Gong turns real sales conversations into original research on pricing, discovery, buyer behavior, and sales performance. These findings then power a wider content library that competitors cannot easily reproduce.

Most sales articles begin with someone’s opinion.

A sales leader explains what worked for their team. A consultant shares a framework. A writer collects several recommendations and turns them into a guide.

Gong built a different kind of content engine.

The company develops software that captures and analyzes interactions between sales teams and potential customers. These interactions include calls, online meetings, emails, and other activities connected with a sales opportunity.

That access gives Gong an unusual source of editorial material. Instead of only asking experienced sellers what they believe works, Gong can study what happened across thousands or millions of real sales interactions.

It can examine how much successful sellers speak, when pricing appears in a conversation, how many people participate in a deal, or which email patterns receive more replies. It can then compare those behaviors with commercial outcomes.

Gong turns the findings into articles, reports, guides, benchmarks, webinars, and practical sales advice through its research program, Gong Labs.

The articles are the visible part of the system. The larger advantage comes from the connection between the product, the data it collects, the questions the research team asks, and the content Gong publishes.

Gong began with a sales problem that existing data could not explain

Gong was founded in 2015 by Amit Bendov and Eilon Reshef.

Bendov had spent more than 20 years working in enterprise software. Before Gong, he served as CEO of the business intelligence company Sisense and held a senior marketing role at the software company Panaya. Reshef was an experienced software entrepreneur who had co-founded the ecommerce technology company Webcollage.

The idea for Gong came from a problem Bendov experienced while leading another software company.

The company had recorded its worst quarter. The leadership team could see that sales performance had fallen, but it could not confidently explain why. Managers had pipeline reports, forecasts, and updates from sales representatives. Those records showed which opportunities were open and which had been lost. They did not show everything customers had said during the conversations that shaped those outcomes.

Bendov concluded that business decisions were being made with an incomplete picture. Sales leaders were hearing summaries of customer conversations rather than the conversations themselves. He contacted his friend Eilon Reshef to explore whether AI could help companies capture and understand what customers were actually saying.

The two founders saw a gap between the information companies recorded and the information they needed.

A sales representative might update an opportunity by selecting a deal stage, entering an estimated value, and adding a short note. The note might say that the buyer was interested, the timing was uncertain, or the price was too high.

Those updates were useful, but they were also subjective. Two representatives could interpret the same customer response differently. Important details could be forgotten, shortened, or softened before reaching a manager.

The real evidence remained inside the original call or email. Gong was built to capture that evidence.

How Gong connects customer conversations with sales outcomes

Gong initially became known as conversation intelligence software.

The product recorded online sales calls, created transcripts, separated the speakers, and used AI to identify subjects and patterns inside the conversation. Gong could show when pricing was discussed, when a competitor was mentioned, which questions were asked, how long each person spoke, and what next steps were agreed upon.

It also connected those conversations with the company’s existing sales records.

Most B2B sales teams use customer relationship management software (CRM). A CRM stores information about potential customers and tracks sales opportunities as they move through the pipeline. It may show when an opportunity was created, its estimated value, the people involved, the expected closing date, and whether the deal was eventually won or lost.

The CRM supplied the outcome. Gong supplied the behavior that led to that outcome.

For example, Gong could identify a group of completed deals and separate them into won and lost opportunities. It could then compare the conversations in both groups.

• Did customers speak more during the won deals?
• Did the seller introduce pricing earlier or later?
• Did successful opportunities involve more people from the buyer’s company?
• Were certain questions more common in calls that progressed?

This connection allowed Gong to study sales performance at a much larger scale than a manager listening to a few recordings.

It also created the foundation for the company’s content strategy.

Once Gong could identify patterns across many interactions, those patterns could be turned into useful research for the wider sales community.

How Gong Labs turns product data into original research

Gong introduced its research ambition publicly in November 2016 through the Sales Conversation Science Manifesto.

The company argued that sales were still guided heavily by personal theories, anecdotes, and intuition. Sales books and training programs offered many different explanations of what successful sellers did. Gong believed its data could test some of those assumptions against real conversations.

The goal was not to create a fixed formula for selling. Human conversations are too varied for that. Gong wanted to identify recurring behaviors that appeared more frequently in successful or unsuccessful opportunities.

Gong does not need to guess what to publish next. Its customer data gives the research team a steady supply of real sales questions to investigate.

Its early research process followed a clear sequence.

First, Gong collected anonymized conversations recorded through its product.

Each call was separated by speaker and converted into a transcript. The product identified topics, key moments, and behaviors within the conversation.

Gong then matched the call with the relevant CRM opportunity. This allowed the research team to see whether the opportunity was won, lost, delayed, or still active.

Finally, the team compared behaviors across groups of calls and looked for patterns linked to the outcomes.

The first large analysis used 25,537 B2B sales conversations from 17 customer organizations. Most of those customers were mid-market software companies. The average call lasted 43 minutes.

That dataset gave Gong enough material for far more than one article.

What Gong learned from its first 25,537 sales calls

Gong’s early analysis examined several parts of a sales conversation.

One of the best-known findings concerned the amount of time sellers and customers spent speaking.

Gong reported that the strongest-performing calls followed an approximate 43% speaking and 57% listening ratio for the seller. By comparison, most representatives spoke during 65% to 75% of their calls.

The finding gave a numerical shape to familiar advice.

Sales trainers had long told representatives to listen more. Gong gave managers a benchmark they could examine during call reviews.

The same study also investigated pricing.

Gong found that three or four pricing questions or mentions from the buyer correlated with higher win rates. It also observed that stronger performers tended to discuss pricing later in the conversation rather than introducing it evenly throughout the call.

Other findings examined customer talking streaks, timeline language, risk-reversal phrases, and the amount of time sellers spent explaining their own company.

Gong could have placed every result into a single long report.

Instead, it separated the findings into individual content topics.

That decision became one of the most important parts of the research engine.

One research study became an entire series of sales articles

The 25,537-call analysis became a source for several focused articles.

Gong published separate pieces about the talk-to-listen ratio, pricing conversations, competitor mentions, customer talking streaks, risk-reversal language, forecast signals, and company introductions. Each article examined one specific behavior rather than repeating the entire study.

Gong turned one study of 25,537 sales calls into content on pricing, competitors, buyer behavior, talk ratios, and more. One dataset became several useful articles without repeating the same story.

This allowed Gong to organize the research around questions that salespeople already had.

• How much should a salesperson speak during a call?
• When should pricing be introduced?
• Does mentioning a competitor help or hurt a deal?
• What language suggests that a buyer may move forward?
• How long should a seller spend explaining their company?

Each question could become a separate entry point into the Gong website.

A sales manager searching for coaching advice might discover the talk-to-listen article. A representative preparing for a pricing call might find the pricing analysis. A revenue leader concerned about forecasting might read the article about buyer language.

The underlying dataset remained the same, but the content addressed different moments in the sales process.

This made the research more productive. It also made the articles more useful.

A reader did not have to search through a large report to find one relevant answer. Gong brought each answer into a focused article with a clear conclusion.

Gong built its content calendar around real sales questions

The research worked because Gong understood the audience it wanted to reach.

Devin Reed, who later led content at Gong, has explained that the team initially avoided writing heavily about the product. It focused on helping sales professionals with the problems they experienced before they were ready to buy software.

The team mapped the sales process and divided it into smaller challenges.

A representative might need help with cold outreach, discovery calls, product demonstrations, pricing conversations, negotiation, deal progression, or executive communication.

Those challenges gave Gong a broad content roadmap.

Gong Labs then made many of those topics more defensible.

An ordinary publisher could write an article about cold calling by interviewing successful representatives or summarizing popular sales books. Gong could analyze recorded cold calls and see how different opening lines performed.

In one analysis, Gong examined more than 90,000 cold calls. It found that calls beginning with “How have you been?” recorded a 10.01% success rate in that dataset, which was 6.6 times higher than the baseline. Opening with “Did I catch you at a bad time?” was associated with a much lower success rate.

Whether every company should copy the exact wording is open to debate. The finding still gave Gong something specific to discuss, test, and explain.

The same approach appeared in its discovery call content.

Gong analyzed 519,000 recorded discovery calls to study how successful representatives uncovered customer problems and structured the conversation. The analysis found that stronger calls often explored three or four business problems in depth rather than covering a long list superficially.

This gave the company a consistent editorial method.

The content calendar did not depend entirely on brainstorming new headlines. The questions came from the work the audience was already doing.

Storytelling made Gong’s research easier to understand

Proprietary data gave Gong authority. It did not automatically make the articles readable.

A research article can become difficult when it begins with methodology, technical terms, and a long description of the dataset. Gong often began with a familiar sales situation instead.

Its articles about company introductions described a representative spending too much time explaining the history and achievements of the seller’s organization. The competitor article began with buyers asking how Gong compared with rival products. The reader first saw a recognizable situation and then learned what the data showed about it.

Reed later described this approach as using the storytelling technique in medias res. The reader is placed inside the situation before receiving the analysis.

The structure helped Gong avoid producing content that felt like a formal research paper.

A typical article followed a practical sequence.

The data answered the question raised at the beginning.

This connection between evidence and storytelling became an important part of Gong’s content advantage. The research gave the article substance. The writing helped the reader remember and use it.

Research findings strengthened Gong’s wider content library

Gong Labs articles did not remain isolated from the rest of the website.

The findings were brought into larger guides about cold calling, prospecting, discovery, email, sales processes, and closing deals.

For example, Gong’s wider guide to the sales process was based on an analysis of more than one million conversations connected with 384,923 deals. It combined findings from different stages of the buyer journey into one broader educational resource.

Its cold-calling guide brought several separate observations together into a list of 17 techniques. The article included research on call length, opening lines, speaking patterns, interaction levels, and meeting-booking language.

Its collection of sales statistics turned individual Gong findings into a highly scannable reference page. Its sales-techniques article organized research-backed advice according to prospecting, meetings, discovery, pitching, objections, and closing.

This created a layered content library.

• A focused Gong Labs article could explain the original study.
• A larger guide could place the findings inside a complete workflow.
• A statistics page could make the result easy to find and cite.
• A downloadable checklist or cheat sheet could help the reader apply it.
• A later article could revisit the same question with new data.
• The value of the research, therefore, extended beyond the first publication.

Each finding could improve several related pages across the site.

The research demonstrated the product without becoming a sales pitch

Gong’s research also helped potential customers understand the product.

Consider the capabilities needed to calculate a talk-to-listen ratio across thousands of calls.

The system must record conversations, identify individual speakers, measure how long each person speaks, create usable transcripts, connect the conversation with an opportunity, and compare the results across won and lost deals.

The research shows what Gong’s product can uncover without constantly talking about the product. Readers get useful sales insights while seeing the value of Gong in practice.

A reader may visit the article because they want to improve their sales calls. While reading, they also learn that Gong can measure these behaviors across an entire team.

Thus, the research becomes a practical demonstration of the product’s capabilities.

The same pattern appears in newer studies.

In 2025, Gong analyzed 1.8 million new business opportunities. It found that 77% involved more than one buyer contact. Won deals had twice as many buyer contacts as lost deals. For opportunities worth more than $50,000, multi-threading was associated with a 130% improvement in win rates. This includes building relationships with several people inside the buyer’s organization rather than relying on one contact.

The article teaches sales teams why wider stakeholder engagement can strengthen an opportunity. It also creates a natural reason to use technology that can identify which people are involved and where important relationships are missing.

The product does not need to appear in every paragraph.

The research makes the problem visible. The reader can then understand why a tool that tracks the problem could be useful.

Gong revisited successful studies as its dataset grew

The first talk-to-listen study was published almost a decade ago. Sales technology, buyer behavior, remote work, and Gong’s own dataset have changed considerably since then.

In March 2025, Gong returned to the question with an analysis of 326,000 calls that lasted at least ten minutes.

The updated research found that representatives in the won deals spoke during an average of 57% of the conversation. Representatives in lost deals averaged 62%. The broader difference was smaller than the original 43:57 benchmark might suggest.

The new study also added a more detailed conclusion. Strong performers kept their speaking patterns relatively consistent across won and lost deals. Weaker performers showed larger variations.

That changed the article from a simple rule about talking less into a more useful discussion about consistency, question quality, and buyer participation.

Gong preserved the recognition attached to the original study while improving it with a larger dataset.

The updated finding then appeared in Gong’s roundup of its strongest 2025 sales research. It was placed alongside newer work on multi-threading, team selling, cold email, and AI adoption.

This creates a research maintenance loop.

• Older questions can be reanalyzed.
• Popular statistics can be tested again.
• New datasets can reveal whether the original pattern still holds.
• The revised result can update the original page and support several new pieces of content.

The content library becomes stronger as the product collects more information.

Major studies expanded into reports, press coverage, and executive content

Gong Labs has gradually expanded beyond articles about individual sales techniques.

The State of Revenue AI 2026 research combined a survey of 3,048 revenue leaders with an analysis of 7.1 million sales opportunities from 3,613 companies. The study examined how companies were adopting AI and how deeper use related to revenue performance, productivity, and decision-making.

Research at this scale can support a wider content campaign.

The complete report gives executives a detailed view of the findings.

• A press release presents the most newsworthy statistics in a format that journalists and industry publications can reference.
• Webinars allow Gong researchers and external experts to interpret what the findings mean.
• Blog articles can explore individual conclusions without asking readers to consume the complete report.
• Year-end roundups can reconnect the research with other studies.
• Product and solution pages can use selected findings where the evidence is relevant.

This is more useful than simply copying one article into several formats.

Each asset serves a different reading need.

Gong Labs is expanding from sales research into market intelligence

Gong currently describes Gong Labs as an original research engine that turns billions of sales conversations, deals, and workflows into insights.

Its newer Gong Labs Trends initiative broadens the purpose of the research.

Instead of only asking which sales behaviors produce stronger results, Trends examines what buyers and sellers are discussing across millions of customer interactions.

The June 2026 dataset covered 33.5 million deals worked since February 2024. Gong found that the share of deals containing AI discussions had increased by 85%. Discussions related to hiring remained broadly flat during the same period.

This type of research can reveal changes in buyer priorities.

It may show when AI agents begin appearing more frequently in commercial conversations, when budget pressure increases, when security becomes a stronger concern, or when businesses start discussing new pricing models.

The audience also becomes wider.

Research about discovery questions mainly serves sales teams. Research about shifting buyer priorities can interest founders, chief revenue officers, marketers, product leaders, investors, analysts, and business journalists.

Therefore, Gong is using the same underlying data to move into a larger field of business research.

The product still captures customer interactions. The research question has expanded from “How should a seller handle this call?” to “What are businesses discussing across the market?”

Why proprietary research remains valuable in AI search

Generative search creates a new challenge for every publisher.

A reader can ask an AI assistant how much a salesperson should speak during a discovery call. The answer may include Gong’s findings without requiring the reader to visit Gong’s website.

That can reduce traffic to the original article.

However, the AI system still needs a source for the statistics. It cannot independently analyze Gong’s private customer interactions or compare those conversations with commercial outcomes.

It can reproduce the published finding. It cannot create the underlying evidence.

This gives proprietary research a stronger position than content built entirely from public information.

A general explanation can be summarized and recreated easily. A newly observed data point must originate somewhere.

The opportunity for Gong is to remain clearly associated with that origin.

The company can strengthen this association by publishing transparent methodologies, using consistent research branding, updating source pages, creating distinctive charts, and ensuring that every reused statistic points towards the original study.

The objective may gradually shift from winning every click to becoming the source behind the answer.

What founders can learn from Gong’s content engine

Gong’s real advantage is the connection between its product, its customer data, and its publishing. The software gives the company access to patterns inside billions of sales conversations, deals, and workflows. Gong Labs turns those patterns into research that helps revenue teams understand buyer behavior and compare their performance.

Founders should begin by asking what their own business can observe more clearly than the wider market. That advantage may come from product usage, transactions, customer support, implementation work, audits, or repeated client projects.

Access to data is only the starting point. Useful research begins with a decision the customer is already trying to make. Gong studies practical questions such as how much a seller should speak, which stakeholders should join a deal, and which behaviors appear more often in successful opportunities.

It also returns to important questions as its dataset grows. In 2025, Gong re-examined its well-known talk-to-listen research using 326,000 sales calls rather than continuing to rely on its original benchmark. This keeps the content current and shows that the company is willing to refine earlier conclusions.

The research should then strengthen the entire content library. One substantial study can support a detailed source article, several focused explanations, an executive report, a practical guide, and later updates. Each asset should answer a different question rather than repeat the same statistic. Over time, the company becomes associated with a specific field of knowledge because its findings continue to appear in relevant articles, training material, industry discussions, and external citations.

The wider lesson is to treat content as an output of the business rather than a separate marketing activity. Gong’s customer interactions generate data. Its research team turns that data into evidence. Its editorial team makes the evidence understandable.

The resulting content earns attention before many readers are ready to buy and quietly demonstrates what the product can uncover. That system is harder to reproduce than a large publishing calendar because competitors can copy the topic, but they cannot easily recreate the data, methodology, and accumulated authority behind it.

Turn the data your business already owns into insights people can use, remember, and come back to.

First Round built the Review around questions founders could not easily answer elsewhere. Its editors went beyond executive opinions to document how experienced operators handled hiring, product-market fit, management, customer feedback, and scaling. Those interviews became detailed articles, memorable frameworks, and eventually a startup knowledge library that readers could return to years later.

Most venture capital firms publish opinions about markets, technology, and fundraising.

First Round Capital went directly to the people building companies and asked them to explain how the work was actually done.

How did Slack collect and use customer feedback during its launch? How should managers hand over responsibilities as a startup grows? How can a company measure product-market fit before the answer becomes obvious? How should a founder find the right co-founder?

These questions helped First Round create a growing library of startup operating knowledge.

Their publication, First Round Review, launched in June 2013 with an ambitious goal: to build the “Harvard Business Review for startups.” Within its first few months, the site had published more than 30 articles, increased visitors to First Round’s website fivefold, and regularly generated more than 1,000 tweets for individual stories. Its first article, about Etsy increasing the number of female engineers, received tens of thousands of views and led to follow-up coverage from several major publications.

By 2015, articles were between 2,000 and 5,000 words. The site was generating around 480,000 monthly page views, while an average article was shared approximately 2,000 times.

A decade after launch, First Round reported nearly 600 articles, millions of page views, hundreds of thousands of newsletter subscribers, and more than 100 episodes of its In Depth podcast. Its current homepage promotes an archive of more than 600 pieces and a subscriber base of over 200,000.

The impressive part is that First Round found a way to convert private experience into public intellectual property. Every strong interview added another framework, phrase, process, or case study to its archive. Over time, those individual articles became a recognizable body of knowledge about building startups.

That is the real First Round Review content engine.

The Review Began With a Gap in Startup Media

When First Round introduced the Review, startup media was already crowded.

Funding announcements, product launches, valuations, acquisitions, and technology trends received constant coverage. Founders could easily learn what another company had announced. They had far fewer places to learn how that company had solved the problems behind the announcement.

First Round saw that gap clearly, and its launch article argued that most startup coverage explained what companies were doing while very little explained how they were doing it. The firm believed much of the useful knowledge remained inside the heads of experienced operators, engineers, product managers, marketers, and founders.

This distinction shaped the publication.

A conventional interview might ask a founder about their vision, leadership philosophy, or predictions for an industry. Those answers can be interesting, but they are rarely useful enough to change how another company operates.

First Round wanted instructions.

The publication would cover topics such as interviewing product managers, building high-performing teams, hiring executives, setting prices, handling customer feedback, finding product-market fit, and navigating management problems.

That gave every article a clear test. A reader should be able to use at least part of the advice immediately.

The Review still describes its value in similar terms. Its current homepage promises tactical company-building advice, detailed steps, and articles that can reach roughly 5,000 words when the topic requires that depth.

The format has evolved. The editorial promise has remained remarkably stable.

First Round Already Had the Raw Material

First Round did not begin with a large newsroom because it already had access.

By the time the publication was formally introduced, the firm had worked with more than 600 entrepreneurs. It was also running more than 70 events a year for its community. Founders and operators were already sharing advice inside meetings, summits, online groups, and private conversations.

The publication gave that knowledge a permanent form.

First Round had access to 600+ entrepreneurs and 70+ community events a year. The Review gave those private founder conversations a much bigger audience and a much longer life.

An answer shared at a private CEO event could help the people in the room once. An edited article could help thousands of founders for years.

This was an important operational advantage. First Round did not have to compete with technology publications for every source. Its investing and community work naturally placed the firm close to people who had built notable companies or solved difficult operating problems.

The network also helped the editors find valuable experts who were not famous.

First Round has said its strongest stories can come from a well-known CEO or from a vice president most readers have never heard of. The determining factor is whether that person is willing to explain the work in depth and share how it was done.

That approach expanded the possible source pool.

A publication driven by celebrity needs a continuous supply of recognizable names. A publication driven by expertise can interview the person who designed the hiring process, built the first sales team, repaired a failing onboarding system, or managed the company through hypergrowth.

For founders, the second person may be far more useful.

First Round also looked beyond its own portfolio. Camille Ricketts, the editor who shaped the early Review, explained that the publication was not intended simply to promote First Round companies. It wanted to feature outstanding people across the technology ecosystem and make their advice widely available.

That decision made the Review feel larger than the fund behind it.

Readers did not need to care which companies First Round had financed. They could visit because they needed to hire their first marketer, improve customer discovery, manage a growing team, or make decisions faster.

The business benefited because the publication was useful beyond its own boundaries.

The Interview Was Treated as Editorial Research

Expert-led content often fails because the interview is treated as a recording session.

Someone schedules a call, asks broad questions, receives broad answers, and turns the transcript into a lightly edited article. The final piece sounds credible because an expert participated, but it rarely contains anything that the expert has not said elsewhere.

First Round developed a more deliberate process.

Before an interview, the editorial team worked with the subject to identify the right topic. The strongest topic sat between three things: what the person understood deeply, what they were genuinely willing to discuss, and what could produce actionable advice for startup readers.

The team then sent questions in advance. An interview lasted around an hour and was transcribed before the article was developed. The preparation changed the conversation.

An executive who receives “What are your leadership principles?” can respond with familiar ideas.

An executive who receives questions about the exact moment a management system failed, what they changed the following week, how they communicated the decision, and what they would measure differently has to retrieve real experience.

The value, therefore, came from the specificity of the inquiry.

The transcript was raw material. It still required an editor to locate the strongest idea, remove repetition, find supporting examples, create a logical sequence, name the framework, and turn the experience into something another leader could apply.

Ricketts originally managed the interviews, ideation, and writing herself. Because attaching the same writer’s name to every story felt distracting, she often removed the editorial byline. She wanted readers to feel that they were hearing directly from the person being profiled.

This was more than a stylistic choice.

It placed the expert at the center of the article. The editor became the person shaping the knowledge rather than the personality competing with it.

Ricketts also treated subjects as partners in the production process. Sources could review what was planned for publication. She found that this made people more forthcoming about mistakes, weaknesses, and personal experiences because the purpose was educational, and they were not worried about being surprised by the final story.

That arrangement created a useful exchange.

The expert received a carefully produced account of their ideas, the audience received uncommon detail, and First Round received original material that could not be gathered from public quotes and existing articles.

Trust became an editorial input.

First Round Turned Experience Into Memorable Frameworks

A transcript preserves a conversation. A strong article improves its usefulness.

The Review repeatedly converted complex experiences into ideas that readers could remember and repeat.

Molly Graham‘s advice about handing over responsibilities during hypergrowth became “Give Away Your Legos.” The metaphor explained a difficult emotional problem. As new employees arrive, early team members can feel that important parts of their work are being taken away. Graham argued that people who want to grow with the company must repeatedly give away parts of their existing job and find a larger problem to own.

Dave Girouard‘s experience at Google and Upstart became “Speed as a Habit.” The article moved beyond the general instruction to work faster. It divided organizational speed into decision-making and execution. It then gave leaders concrete practices, such as deciding when a decision will be made, questioning delivery dates, removing dependencies, and asking why important work cannot be completed sooner.

Kim Scott‘s management experience became “Radical Candor.” The Review article explained the framework through stories, a two-axis model, contrasting management behaviors, and an acronym that made the advice easier to use. The piece helped turn a broad idea about honest feedback into a practical management language.

First Round had access to 600+ entrepreneurs and 70+ community events a year. The Review gave those private founder conversations a much bigger audience and a much longer life.

These concepts, including Radical Candor and Superhuman’s product-market-fit framework, gained early attention through its pages. Phrases such as “Give Away Your Legos” and “Speed as a Habit” became part of startup vocabulary.

These examples show what the editor adds.

That structure might be a sequence, diagnostic question, maturity model, score, operating rule, metaphor, decision tree, or set of warning signs.

Once the idea has a clear form, it becomes easier to remember. Once it is easier to remember, it is easier to share. Once it is shared widely, the article can influence how an entire professional community talks about the problem.

First Round Wrote for Founders, But Built an Audience Much Larger Than That

When First Round Review launched, the audience was much narrower. The editorial team wanted to help founders build companies. Every article started with a practical problem: hiring early employees, finding product-market fit, managing a growing team, launching a product, or making faster decisions.

That focus influenced the way the articles were written.

Instead of asking founders what they believed about leadership, the Review asked how a particular decision had been made. Instead of summarizing success, it unpacked the sequence of events that produced it.

The interview with Slack co-founder Stewart Butterfield is a good example.

Rather than simply saying Slack listened closely to customers, the article explains how the company invited its first external teams onto the platform, what changed when a 120-person organization started using the product, how support requests were categorized, how Twitter conversations were monitored, and how customer feedback reached product teams. Readers can follow the operating system behind Slack’s early growth instead of seeing only the outcome.

The same pattern appears in Rahul Vohra‘s product-market-fit article.

Instead of encouraging founders to “talk to users,” Vohra walks through the survey methodology Superhuman used, explains how responses were segmented, shows how those responses influenced product priorities, and even links to a tool that lets readers apply the framework themselves.

This level of detail gave the articles a much longer lifespan.

News loses value quickly. A funding announcement is interesting for a few days. A product launch matters until the next launch arrives.

Operating decisions are different.

Every generation of founders has to hire managers, improve onboarding, understand customers, build product teams, and decide when a company is ready to scale. The context changes, but the underlying questions stay remarkably similar.

That helps explain why many of the Review’s best-known articles are still among its most-read years after publication. They attract readers because they solve recurring problems rather than documenting one moment in startup history.

Interestingly, this focus also expanded the publication’s audience.

Although the Review was written with founders in mind, product managers, recruiters, engineering leaders, marketers, designers, executives, and even investors found themselves reading the same articles. Once an article genuinely helps people do their jobs better, the audience naturally grows beyond the original target.

That wider readership wasn’t created by covering more topics. It happened because the publication stayed disciplined about the problems it wanted to solve.

Readers returned because they knew what to expect. Whatever company appeared in the next story, the article would explain how experienced operators approached a difficult business problem and provide enough detail for someone else to adapt the same thinking inside their own organization.

The Archive Became More Valuable as It Grew

A publication can become harder to navigate as its archive expands.

First Round responded by turning individual articles into structured collections.

In 2015, the Review reorganized its work into nine digital magazines covering areas such as management, fundraising, product, design, and company culture. The goal was to help a sales leader, designer, founder, or manager discover more material relevant to their work instead of scrolling through a general feed.

The current site uses topic pages, curated collections, classics, newsletters, series, and podcast formats.

Its collections now gather advice around clear reader situations. Examples include becoming a better manager, joining an early-stage startup, building a team, improving customer conversations, and navigating career decisions.

This makes the archive behave more like a knowledge library.

An article may first reach readers through a newsletter or recommendation. Years later, it can resurface inside a management collection, a list of classic articles, a related story, a podcast compilation, or a new thematic series.

The Review demonstrated this compounding model when it marked its tenth anniversary by selecting 100 pieces of advice from its archive. It has also published collections of lessons from podcast episodes and product-market-fit interviews.

Older reporting became material for new editorial products.

That helps because original interviews are expensive. They require access, preparation, calls, transcripts, editing, fact-checking, source coordination, design, and distribution.

A strong archive improves the return on that investment. One interview can remain discoverable for years. A group of related interviews can later support a collection. A mature collection can become a series, learning program, newsletter theme, or podcast discussion.

The asset continues producing value after the original publication date.

The Review Built Trust Before the First Founder Meeting

A founder choosing an investor has limited information.

Capital is visible. The working relationship is harder to judge.

Every venture firm can say that it understands founders, has a strong network, and provides valuable support. The Review allowed First Round to demonstrate those claims publicly.

Its articles showed that the firm could reach experienced operators. They showed that it understood the problems founders faced. They showed that it was willing to invest time and resources into making useful knowledge available.

Ricketts described brand identity as one of the publication’s main goals. First Round wanted to be seen as a constructive participant in the startup ecosystem. She also acknowledged the commercial effect. Wider content reach could encourage more founders to approach the firm, giving First Round a stronger pool of potential investments.

She made the relationship even clearer in another interview. When founders see high-quality content, deep knowledge, and strong connections coming from a firm, they may become more likely to want that firm involved in their company.

This does not mean every reader became a portfolio founder.

First Round gave founders years to understand the firm before ever discussing an investment. More than 200,000 subscribers now have regular access to the people, ideas, and operating knowledge connected to its network.

First Round had not established a direct statistical connection between Review traffic and new investments when Fralic discussed the publication in 2016. The primary goal remained broad reach and practical usefulness.

That lack of direct attribution does not remove the business value.

A founder might read the Review for years before raising capital. By the time First Round appears in a financing conversation, the firm is familiar. The founder already knows its voice, network, areas of expertise, and attitude toward company building.

The content has completed part of the trust-building process before a partner joins the meeting.

Why the Model Is Difficult to Copy Quickly

A competitor can copy the article length.

It can create similar category pages, interview founders, launch a newsletter, and publish titles containing words such as “framework,” “playbook,” and “tactical guide.”

The underlying system is much harder to reproduce.

First Round’s advantage came from several connected assets:

• Access to founders and operators with useful experience
• Enough trust for those people to discuss mistakes
• Editors who could find a narrow and valuable topic
• Interview questions that extracted operating details
• Editorial judgment that turned conversation into structure
• A recognized publication that experts wanted to appear in
• An archive that increased the value of every new article
• A distribution network of founders, executives, and investors

Each asset reinforced the others.

A stronger publication attracted better experts. Better experts produced better articles. Better articles expanded the audience. A larger and more credible audience made participation more attractive to the next expert.

The loop took years to build.

This is why interview-led thought leadership should not be reduced to “book experts and write articles.” Access without editorial skill produces ordinary interviews. Editorial skill without access produces secondhand summaries. Distribution without substance creates a temporary traffic spike.

The engine works when all three are present.

Expert-Led Content Has a Different Role in the AI Era

AI systems can summarize a First Round Review article.

They can extract the main framework, shorten the advice, compare it with other management models, or turn it into a checklist.

That does not remove the need for the original work.

The distinctive information existed because an editor interviewed someone with direct experience. The source remembered the event, explained the decision, disclosed the mistake, and provided the example. The editor then selected and organized the material.

An AI system working only from public information cannot independently remember what happened inside Slack’s early customer-support operation. It cannot recreate a private performance conversation at Google. It cannot know how a founder interpreted a failed product launch unless someone records that experience.

It can process the knowledge after publication. The interview creates the knowledge before publication.

Current search guidance is moving in the same direction. Google advises publishers to demonstrate first-hand expertise and depth. It warns against pages that mainly summarize what others have already said. Its guidance for generative search specifically recommends unique, expert-led, non-commodity content that adds value beyond common knowledge.

Google also states that generative AI can help research and structure original material, while large-scale generation without added value may violate its spam policies.

OpenAI describes ChatGPT search as a way to connect users with original, high-quality web sources through citations and links.

None of this guarantees that an interview-based article will receive rankings, traffic, or AI citations.

It does show why original source material remains strategically valuable. Every answer system needs information to retrieve, compare, summarize, and cite. Publications that create new information have a stronger foundation than those repeating the same public consensus.

First Round’s defensibility lies upstream of the article.

The moat is the relationship with the expert, the quality of the questions, the willingness to reveal specifics, and the editorial ability to make those specifics useful.

First Round Is Expanding the Same Editorial Principle

The Review has added new formats without abandoning its original idea.

Its Paths to Product-Market Fit series examines the early years of companies, including false starts, difficult pivots, customer discoveries, and the decisions that eventually unlocked demand. First Round partner Todd Jackson conducts many of the interviews. The series includes detailed company histories from businesses such as Linear, Clay, Vanta, Vercel, and others.

Its Zero to $5M series focuses on founders and early sales hires navigating the first few million dollars in revenue. The In Depth podcast applies the publication’s interview style to long conversations with founders and operators.

In early 2026, First Round introduced another podcast called Executive Function. It focuses on senior executives at fast-growing companies and aims to examine how exceptional leaders hire, make decisions, build systems, and keep pace with companies that may change fivefold or tenfold within a year.

The publication has also begun featuring more personal, narrative work. Recent homepage stories include firsthand accounts from early employees and deeper profiles of influential technology figures.

These formats may look different, but the operating logic is consistent.

Find knowledge that is difficult to access. Ask the person closest to it. Go deeper than the public story. Give the reader enough detail to use what was learned.

What Can Founders and Content Leaders Learn from First Round?

First Round Review worked because it built its content around an advantage that competitors could not easily copy. The firm already had access to experienced founders and operators. Instead of using that access for promotional interviews, it turned real company-building experience into practical editorial assets.

The publication also stayed disciplined about the problems it covered. Rather than chasing startup news or broad leadership advice, it focused on questions that founders repeatedly face, from hiring and product-market fit to sales, management, and scaling. Those problems rarely disappear, which is why many of its articles continue to attract readers years after publication.

The editorial process deserves just as much credit as the interviews themselves. Every conversation was shaped into a clear framework with examples, practical steps, and memorable ideas. The result was content that readers could immediately apply, not simply admire. As the archive expanded, First Round organized it through collections, series, newsletters, and podcasts, allowing older articles to keep creating value instead of disappearing into a blog archive.

Perhaps the biggest lesson is that thought leadership begins long before the writing starts. Original ideas come from original conversations. The interview uncovers the experience, the editor turns that experience into something others can use, and the publication gradually builds a knowledge library that becomes more valuable with every article. That is why First Round is remembered not just for publishing excellent startup content, but for documenting how great companies are actually built.

A founder’s experience becomes far more useful when the reader can see the decisions behind the outcome.

ClickUp matches each content format to a specific buyer intent, from templates for task-seekers to comparisons for buyers evaluating tools. Together, these assets guide users from discovery and evaluation to product use.

How This Teardown Was Researched

This analysis draws only on public sources: ClickUp’s website, template directory, comparison hub, team pages, funding and growth announcements, and a published partner case study. Company- and vendor-reported figures are labeled as such throughout. No private analytics, conversion, or revenue-attribution data was used or assumed.

We reviewed six areas:

1. ClickUp’s template directory and individual template pages
2. Its competitor comparison hub and individual comparison pages
3. Team and department landing pages
4. Editorial blog content and software roundups
5. Public company announcements and founder statements
6. A published case study on ClickUp’s content-production workflow

We did not use, and this analysis does not assume access to: Google Analytics or Search Console data, content conversion rates, product-qualified lead data, CAC, revenue attribution, retention data, internal briefs, or editorial guidelines. Any statement about acquisition, activation, or pipeline is our external read of a public content structure, not a confirmed internal metric.

The Core System: Matching Content Format to Buyer Intent

ClickUp doesn’t use one blog format for every query. It routes search demand into distinct buckets: informational, functional, evaluative, and transactional, and it builds a purpose-specific page type and CTA for each one. A “template” searcher and a “vs” searcher get completely different experiences because they’re at completely different points in the buying journey.

A visitor searching “what is an SOP” and a visitor searching “ClickUp vs Asana” are both in ClickUp’s target market. They are not, however, the same buyer. One is learning a concept. The other has a shortlist and a decision to make. ClickUp’s content architecture treats that difference as the organizing principle for the entire library.

Reader Need	ClickUp Content Type	Likely Next Step
Learn how a process works	Educational article	Explore a related method or template
Start a process quickly	Individual template	Add the template / create a workspace
Compare software categories	Software roundup	Review ClickUp alongside other platforms
Evaluate named vendors	Comparison page	Sign up, compare features, consider switching
Assess departmental fit	Team / department page	Explore relevant workflows
Implement a feature	Help / product content	Use the product more effectively

ClickUp uses content to present one platform through many different customer needs and evaluation stages, rather than forcing every visitor through the same funnel.

What Growth Actually Coincided With This Strategy

ClickUp’s public milestones show strong company growth running alongside heavy investment in organic, product-led content. The available evidence does not isolate how much revenue the content specifically drove. What it does show is a model where specific content assets bring in a first use case, and a broad product expands usage from there.

In October 2021, alongside a $400 million raise at a $4 billion valuation, ClickUp’s revenue had tripled year-over-year. Its customer base had grown from 200,000 to 800,000 teams, and more than 90% of customers were already using at least three ClickUp products.

In March 2024, it served more than 10 million users across 2 million teams. In a February 2026 article, ClickUp stated revenue had grown tenfold since it reported nearly $20 million in ARR in 2020.

The available figures show ClickUp’s overall growth, not the specific impact of its content strategy. Growth also came from product quality, freemium distribution, sales, paid marketing, partnerships, and retention. What the pattern does support is a two-stage read of the funnel:

1. A specific piece of content, such as a template, a comparison, or a team page, gives a visitor one relevant reason to try ClickUp.
2. Once inside, ClickUp’s product breadth (90%+ multi-product adoption, per its own 2021 disclosure) does the work of expanding that single use case into a full workspace.

A template built for a marketing team can be the entry point for an account that later adopts docs, dashboards, and automations. A comparison page aimed at Asana switchers can be the entry point for a company that eventually replaces three or four other tools. Content’s job isn’t to close the deal on its own; it’s to win the first, narrow reason to log in.

Templates: The Fastest Route From Search to Product Use

A template page does something a normal landing page can’t: it lets someone use the product before they’ve signed up. That’s why it converts better than a typical blog post or feature page.

Here’s the logic. Someone searching “SOP template” isn’t looking to learn what an SOP is; they already know. They want a ready-made structure they can start filling in today. ClickUp built its 1,000+ template library specifically for this kind of searcher, across different categories:

A new user setting up a workflow from scratch normally has to decide on structure, build the workspace, configure fields, choose views, assign owners, and set deadlines before doing any real work. A template page compresses that into a shortcut. So, they are not only introduced to your product, they are already using it.

Thus, template pages don’t just attract traffic; they collapse the gap between “learning about the product” and “using the product” into a single click. That’s the entire reason this format outperforms a standard landing page.

Comparison Pages: Meeting Buyers Who Already Have a Shortlist

ClickUp’s comparison hub covers 24+ named competitors, targeting readers who’ve already picked a category and narrowed it to specific vendors. This is lower-volume, higher-intent traffic than a broad informational query, and it’s traffic that ClickUp would otherwise have to buy at high CPCs on Google Ads if it didn’t own the organic result.

ClickUp even targets comparisons that don’t include its name, such as “Trello vs Jira.” This puts ClickUp in front of buyers while they are still deciding which tools to consider.

Bidding on terms like “Asana alternatives” or “Monday vs Jira” in paid search is expensive precisely because the buyer intent is so strong; advertisers know these searchers are close to a decision. ClickUp’s comparison hub, which covers Monday.com, Notion, Asana, Jira, Trello, Airtable, Smartsheet, Wrike, and roughly 15 others, earns that same high-intent placement organically.

What’s notable is that some of these comparisons don’t mention ClickUp in the query at all. “Trello vs Jira” is two competitors’ traffic, captured by a third party who shows up with a balanced breakdown and then earns the right to pitch. Competitors rarely defend paid budget against searches that name two other companies, which makes this some of the cheapest high-intent traffic available.

Thus, a vendor writing about its own comparison has an obvious conflict of interest, and buyers know it. ClickUp’s main Asana comparison leans into product breadth as the differentiator, which is reasonable, but it’s also exactly what you’d expect a vendor to say.

Its help-center comparison is more convincing, because it explicitly concedes that Asana may suit teams that want more structure with less configuration. That kind of acknowledged trade-off is what separates a comparison page that gets cited from one that gets dismissed as marketing.

Team Pages: Making One Broad Platform Feel Built for You

Next comes the Team Pages. ClickUp solves the “does everything, means nothing” problem of broad platforms by splitting its messaging along department lines. Marketing, sales, and engineering each get a page that speaks their specific workflow language instead of a generic feature list, which matters both for first conversion and for internal expansion once one team is already inside.

A general project management pitch rarely lands with a software engineer the same way it lands with a marketing coordinator. However, the vocabulary, the pain points, and the proof points are different.

ClickUp content topics by audience

Audience	Topics Emphasized
Marketing	Briefs, campaigns, calendars, approvals, reporting
Development	Product requirements, sprints, backlogs, bugs, releases
Sales	Lead handoffs, deal tracking, customer onboarding
Design	Creative production, feedback, asset delivery
Product Management	Roadmaps, priorities, execution

The underappreciated role of these pages is internal expansion, not just first-touch acquisition. A single project manager who onboards ClickUp for one team has a ready-made page to forward to marketing, design, or sales colleagues, one that explains, in their language, why the same tool fits their workflow too. That’s a meaningfully cheaper expansion motion than a second cold acquisition effort.

The Blog Loop: Turning Educational Traffic Into Product Trials

ClickUp’s blog is structured to catch readers who aren’t searching for ClickUp at all, they’re searching for how to do something. The strongest articles don’t stop at explaining the process; they embed the exact template or tool needed to execute it, so the “next step” is inside the product rather than a separate search.

Most B2B blogs treat educational content and product features as separate systems. ClickUp’s process guides are built to close that gap directly:

The product reference works because it’s positioned as the easiest way to do the thing the article already taught, not as an interruption. Its task-management software roundup, ranking ClickUp alongside Todoist, Asana, Monday.com, Trello, TickTick, and Notion, does the same job for category-level, non-branded search. It puts ClickUp inside the comparison set before the reader has even decided to compare vendors directly.

Scaling Production Without Losing the Plot

ClickUp includes about 150+ new articles and 130+ optimizations of existing pages over 12 months, correlated with an 85% increase in non-branded organic traffic. These are vendor-reported figures, not independently audited ones, but the operating model they describe is instructive regardless of exact attribution.

ClickUp published 150+ new articles and optimized 130+ existing pages in 12 months. Almost half of its 280+ content actions were updates, not new content.

At roughly 280 combined publish and optimize actions over a year, that’s about one significant editorial action per working day. The real detail here is that updating existing content was treated as equal in priority to publishing new content.

Articles moved into an optimization queue when rankings slipped, with results tracked on a four-to-six-week cycle. That’s a maintenance discipline most content teams skip in favor of always chasing new posts, and it’s usually the difference between a library that compounds and one that slowly decays.

What a Library This Size Costs You

Scaling to hundreds of comparison pages, templates, and guides creates real operational challenges. Near-duplicate pages can begin competing for the same search query, making it harder to maintain clear search intent.

The workload also grows over time. Competitor pricing and feature information can become outdated, while some pages continue attracting traffic without contributing to signups, product adoption, or pipeline.

None of this is a reason to avoid scale. It’s a reason to budget for maintenance from day one, not as an afterthought.

Four risks show up predictably at this scale:

• Search-intent overlap. A single topic can spawn a definition article, a template roundup, an individual template, a software list, and a department page, all technically about the same subject. Without a clear content map defining each page’s distinct job, these pages cannibalize each other’s rankings.
• Repeated editorial formats. Templated structures (“explain the topic → list options → place ClickUp prominently → CTA”) produce content faster. But it also produces content that starts to look interchangeable, which is exactly the kind of content generative search engines summarize instead of citing.
• Stale product claims. Competitor pricing, feature limits, and integrations change multiple times a year. A comparison page with 2024 pricing data is unhelpful and also actively damages trust with a buyer who’s already checked the competitor’s current site.
• Traffic without business value. Visits are not the same as a qualified pipeline. Every page type should have its own success metric. Template pages should measure template use, comparison pages should measure trial or demo starts, and educational content should measure qualified click-throughs.

Which of These Assets Survives the Shift to AI Search

Generative search engines can now answer basic definitional queries directly, with zero clicks to the source. What survives that shift is content that offers something an AI answer can’t: a usable artifact, a transparent test, or genuinely proprietary data. ClickUp’s templates and interactive tools sit in a much stronger position here than its definitional and list-style content.

AI search vulnerability by content asset

Content Asset	AI Search Vulnerability	Strategic Response
Basic definition guides	High — summarized instantly, low click-through	Rebuild as interactive hubs, not static explainers
Competitor comparison pages	Medium — AI can extract feature lists but needs verified sourcing	Add original screenshots, transparent testing methodology
Templates & workspace assets	Low — a functional structure can’t be reproduced in a text summary	Double down; this is the most defensible format ClickUp has
Proprietary research & data	Very low — AI search must cite the original source of new statistics	Publish first-party data other publishers can’t replicate

Google’s own guidance for AI Overviews and AI Mode reinforces this: it recommends content that is unique, useful, and based on information or experience that adds value beyond what’s already summarizable. It should not be a different technical playbook, just a higher bar for originality. A template clears that bar automatically, because the value only exists inside the product. A definition article does not, no matter how well it’s written.

The Replicable Framework

You don’t need ClickUp’s budget to run this playbook. You need the same discipline: map real customer intent to the right format, connect every piece of content to a specific next action inside your product, and measure each format against the metric that actually matches its job.

Match the asset to the task

Reader Intent	Right Format
Understand a process	Educational guide
Start a task	Template, checklist, or tool
Research a category	Software evaluation
Compare vendors	Competitor comparison
Assess team fit	Department or use-case page
Build internal approval	ROI model or buying guide

Measure by format, not by traffic:

• Educational content → qualified click-through to product pages
• Comparison pages → trial starts, demo requests, influenced pipeline
• Template pages → workspace activation and template-use rate

And build maintenance into the plan from the start: fact-checking cadence, internal-link reviews, and a clear removal or consolidation policy. Don’t treat your content library as something you only ever add to.

Final Takeaway

ClickUp’s content strength comes from assigning a clear commercial purpose to every format in its library.

Templates reduce the time between search and product use. Comparison pages help buyers evaluate software at the moment they are making a decision. Team pages explain the same platform in language that different departments understand. Behind all of these, you find a disciplined optimization process that keeps the library useful as it grows.

As AI search changes how people discover information, simply publishing more articles is unlikely to create a lasting advantage. Practical assets, first-hand evidence, transparent comparisons, and continuously maintained content are harder to replace with an AI-generated summary and more valuable to both readers and search engines.

That’s the principle behind how we think about modern content strategy. AI is excellent at accelerating research, drafting, and maintenance. But deciding which customer problems deserve content, which format best serves the reader, and how each asset supports product adoption or revenue still requires human judgment. A high-performing content engine is built on that combination, not on volume alone.

Want to build a content engine like this? Every teardown on this site reverse-engineers a public growth system to uncover the strategic decisions behind it. If you’re building a B2B content engine that prioritizes qualified pipeline over pageviews, we’d love to talk.

Book a Strategy Call

The real advantage is not publishing more content, but building the right asset for each buyer need and keeping it useful over time.

Strengthen technical SEO, publish original evidence and first-hand expertise, and ignore AI-search shortcuts that add work without improving visibility.

Marketers trying to improve visibility in Google’s AI search should focus on strong content, technical SEO, first-hand expertise, supporting evidence, images and video, accurate structured data, and measurement. They can ignore many of the new GEO tactics built around special AI files, artificial content structures, and other supposed shortcuts.

That distinction has become more important as AI search reaches a much larger audience.

On May 15, 2026, Google published a detailed guide to improving visibility in generative search features such as AI Overviews and AI Mode. The document addresses content quality, technical SEO, images, ecommerce data, measurement, and several popular tactics that websites can safely ignore.

The guidance arrived as these features moved well beyond the experimental stage. AI Overviews now has more than 2.5 billion monthly active users, while AI Mode has crossed one billion monthly users. Queries in AI Mode have more than doubled every quarter since its launch.

At the same time, organic search is becoming less predictable.

A December 2025 analysis of 300,000 keywords found that AI Overviews were associated with a 58% lower average click-through rate for the top-ranking result. The dataset compared 150,000 keywords with AI Overviews against 150,000 informational keywords without them.

The challenge for marketers is therefore broader than winning the first organic position. Content also needs to be useful enough to support an AI-generated answer and persuasive enough to earn a click when the searcher already has a summary in front of them.

So, what deserves attention, and which AI search tactics are creating unnecessary work?

The guidance in one view

A quick summary of the practices worth investing in and the AI search tactics that can be deprioritized.

What to prioritize for Google AI Search

Area	Worth investing in	Safe to deprioritize
Content	Original research, expert insight, first-hand experience, and complete topic coverage	Generic summaries and separate pages for every keyword variation
Technical SEO	Crawlability, indexing, internal links, correct canonicals, and a strong page experience	A separate technical infrastructure created only for AI search
Formatting	Clear sections, useful headings, readable paragraphs, and relevant visuals	Breaking every article into tiny “AI-friendly” chunks
AI-assisted writing	Research support, structure, editing, and content maintenance with human review	Publishing large volumes of lightly reviewed pages
Structured data	Valid markup for established search features and accurate commercial information	Invented “AI schema” or excessive markup added only for AI Overviews
Brand authority	Genuine expertise, customer evidence, and authentic third-party discussion	Purchased mentions designed to influence AI citations
Measurement	AI impressions, qualified traffic, conversions, and branded demand	Treating citation counts or rankings as the only success metric

The rest of the article explains why these priorities are taking shape and how marketing teams can apply them.

How Does Google AI Search Work?

Google AI Search works by understanding the meaning and intent behind a query, retrieving relevant information from multiple sources, and using AI to combine those findings into a direct answer. A page still needs to be discovered, crawled, and indexed before it can appear as a supporting source in Google’s AI search features.

How Does AI Search Understand a Query?

AI Search understands a query by analyzing its meaning, context, and relationships rather than relying only on exact keyword matches.

• Natural language processing: AI search analyzes the complete question to understand what the user means, including context and relationships between different parts of the query.
• Semantic matching: It can connect a query with relevant information based on meaning, even when the page does not contain the searcher’s exact words.
• Vector embeddings: Many AI retrieval systems represent words, passages, and queries numerically, helping identify information that is semantically related rather than simply keyword-matched.

How Does AI Search Retrieve Information?

AI Search retrieves information by finding relevant external sources that can help answer the user’s question. Retrieval-Augmented Generation (RAG) uses this information to ground the generated answer rather than relying only on information learned during model training.

Google AI Search can also use query fan-out. A complex question can be broken into subtopics, with multiple related searches performed simultaneously to gather information that addresses different parts of the question.

For example, a search for “Which CRM is suitable for a 50-person SaaS company with a small sales team?” could expand into questions about:

• pricing and implementation time;
• integrations and data migration;
• reporting and sales automation; and
• scalability and customer support.

Different pages can contribute useful information to different parts of the final answer. This means a page does not necessarily have to answer the entire original query. A strong section, comparison, statistic, or explanation can provide relevant information for one of its supporting questions.

How Does AI Search Generate Answers?

AI Search generates answers by combining the relevant information it retrieves into a coherent, conversational response.

• Information synthesis: The model combines relevant information retrieved from multiple sources around the user’s question.
• Response generation: It turns those findings into a direct, conversational answer rather than simply presenting a list of search results.
• Source links: Supporting links allow users to explore the webpages behind parts of the generated response and verify or learn more about the information presented.

For marketers, this changes how topic coverage and keywords should be approached. A page can become relevant because one well-developed section answers a supporting question, even when it does not use the exact phrase entered by the searcher.

The opportunity is to build content around real questions, related decisions, entities, evidence, and complete topic coverage rather than creating multiple near-identical pages for slight keyword variations.

How Is AI Changing Google Search?

AI is changing Google Search by making it more conversational, interactive, multimodal, and capable of answering complex questions directly. Users can ask follow-up questions, search with text, images, or voice, evaluate sources, and continue researching without repeatedly returning to a traditional results page.

The change also affects how users interact with Search:

• Multimodal inputs expand discovery: Users can search using text, voice, images, or combinations of these inputs.
• Context carries across follow-ups: Users can refine or narrow a question without explaining the entire request again.
• Complex intent can appear in one query: Requirements such as budget, location, features, preferences, and constraints can be included together.

How Do AI Answers Affect Search Clicks?

AI answers can reduce the need for some search clicks by giving users explanations, comparisons, recommendations, and supporting information directly within Search. Users may no longer need to open several pages simply to assemble a basic answer.

As a result, the value of a click is also changing:

• Basic information may no longer earn the click: Definitions and simple summaries can often be satisfied within the AI response.
• Deeper information creates a stronger reason to visit: Original research, detailed comparisons, tools, first-hand evidence, and proprietary data can extend what the AI answer provides.
• Visibility can happen without a website visit: A brand or publisher can appear as a supporting source even when the user continues researching within Search.

What Can Google AI Mode Do?

Google AI Mode can help users research complex questions, compare options, refine their requirements, and work through multi-step tasks within Search. With Gemini 3.5 Flash as its default model, AI Mode is designed to handle detailed questions while producing responses with less delay.

This expands the types of tasks users can complete through Search:

• Research can develop over several stages: A broad question can progress into comparisons, evaluation, and more specific follow-ups.
• Search can assist with decisions: Users can provide multiple requirements and ask Search to evaluate suitable options.
• The interaction becomes iterative: Each response can provide the context for the user’s next question.

How Do Citations Work in Google AI Search?

Citations in Google AI Search connect generated information with supporting web sources, giving users a direct route to the pages behind individual claims. Inline citations and desktop hover previews can also provide information about a source before the user decides whether to visit it.

That creates several additional considerations for publishers:

• Page titles influence source selection: A specific title can help users understand what additional information they will find after clicking.
• Brand recognition can influence trust: Familiar publishers may be easier for users to identify among several cited sources.
• Citation placement creates new visibility: A source can appear beside the exact fact, explanation, or recommendation for which it was selected.

How Does Google AI Search Surface First-Hand Perspectives?

Google AI Search can surface first-hand perspectives from forums, reviews, creator communities, social platforms, and other sources alongside synthesized information. Features such as Expert Advice can give practical experiences greater visibility within the search journey.

These sources can contribute information that traditional articles may not always provide:

• Product experience adds practical context: Users can see how products or services perform in real situations.
• Community discussions reveal different viewpoints: Forums and social platforms can surface questions, problems, and experiences from multiple users.
• Creators can contribute specialized knowledge: People with direct expertise can provide details that generic summaries may overlook.

How Do Follow-Up Questions Work in AI Search?

Follow-up questions in AI Search allow users to continue researching a topic without starting a new search from scratch. Explore More, related questions, and personalized recommendations can move users from an initial question into progressively more specific areas.

The journey can become more individualized as it continues:

• Related prompts introduce adjacent questions: Users may discover considerations they did not include in their original search.
• Personal context can shape recommendations: AI experiences can use available context and preferences to make results more relevant.
• Preferred sources can receive greater prominence: Publications a user already follows or subscribes to may be highlighted more visibly.

12 Practices to Improve Visibility in Google AI Search

A practical breakdown of the content, technical, and measurement improvements that increase a page’s chances of being discovered and referenced.

1. Get the Technical SEO Foundations Right

Get the technical SEO foundations right by making important pages crawlable, indexable, internally connected, and eligible to appear in Search. AI Overviews and AI Mode still depend on the infrastructure Google uses to discover and process web pages.

Before building a separate AI search program, review whether important pages can already be found and processed correctly.

A practical technical review should cover whether:

• Important URLs are indexable
• Canonical tags point to the preferred page
• Internal links connect related articles and commercial pages
• JavaScript prevents essential content from loading
• Images and videos can be accessed
• The page is eligible to appear with a search snippet
• Duplicate or outdated URLs are consuming crawl resources

These checks rarely make headlines, but they often determine whether a page is eligible to appear at all.

Use relevant structured data where it supports established Search features. Article, Product, Organization, LocalBusiness, and other applicable markup can help Search understand page information and entities, but the markup should accurately represent visible content. There is no separate “AI schema” required for AI Overviews or AI Mode.

Technical clarity also helps teams diagnose performance. When a well-researched page fails to appear, the team can separate access and indexing problems from issues related to content quality or competition.

2. Publish Original Research

Publish original research that competitors cannot easily reproduce. Internal research, customer evidence, product testing, expert analysis, implementation experience, and proprietary data give a page of information with a clear origin. Here’s something to know:

• Commodity content is interchangeable. It may be accurate and well written, yet it adds little beyond what a searcher can already find across dozens of websites. Generic definitions, surface-level checklists, and articles assembled from the current top-ranking results usually fall into this category.
• Non-commodity content introduces information with a clear origin. It may come from internal research, customer interviews, product testing, expert analysis, implementation experience, or a strong point of view developed through years of work.

Consider two articles about improving a SaaS pricing page.

• The first collects seven common conversion tips: simplify the layout, add social proof, clarify the plans, and test the call to action. The advice is familiar and easy to reproduce.
• The second article explains how a company redesigned its pricing structure, which customer objections shaped the new page, why two experiments failed, and what happened to trial conversion after the change. That article contributes evidence that did not exist before the company published it.

Originality does not always require a large research budget. A content team can create distinctive material through:

• Interviews with customers or internal specialists
• Examples from completed projects
• Screenshots of real workflows
• An analysis of public company data
• A comparison based on hands-on testing
• Lessons from a failed launch or implementation
• Benchmarks drawn from anonymized internal data

The aim is to give the page a reason to exist beyond targeting a keyword.

3. Demonstrate First-Hand Experience and Expertise

Demonstrate first-hand experience and expertise by showing how real knowledge shaped the article’s analysis, examples, and recommendations. Author biographies, credentials, expert reviewers, and dedicated author pages can support trust, but the article itself should show evidence of experience.

A warehouse operations guide written by someone who has worked inside distribution environments should contain details about labor constraints, replenishment delays, slotting decisions, and implementation trade-offs. A software comparison should reflect actual product use. A marketing teardown should explain the choices behind a company’s growth system rather than restating its website.

Experience often appears through small details:

• The assumptions that shaped a decision
• The constraint that delayed implementation
• The feature that performed differently in practice
• The customer objection that changed the strategy
• The metric that looked positive but hid a deeper problem
• The compromise the team accepted to keep the project moving

These details make an article more useful because they prepare the reader for real decisions. They also separate expert-led content from summaries that can be produced without first-hand knowledge.

This broader shift is also visible across search results. Community discussions, product reviews, and expert conversations have become more prominent because they often contain practical insights that are difficult to reproduce through generic content.

Rather than competing with those sources, publishers should learn from what makes them valuable: first-hand knowledge, transparent opinions, and evidence from real-world experience.

4. Cover the Topic and Search Intent Completely

Cover the topic and search intent completely by answering the related questions that a reader needs to make a decision. Query fan-out can turn one question into several related searches, creating more opportunities for a comprehensive page to provide useful information.

Suppose the primary topic is choosing CRM software for a growing company. The reader may begin with a comparison, but the decision quickly expands into implementation effort, pricing, integrations, data migration, reporting, and adoption.

A strong guide could therefore move through the decision in a natural sequence:

Define specialized terminology when it is necessary to understand the subject. Clear definitions establish what an entity, process, or concept means before the article builds more complex explanations around it.

Each section supports the same reader and the same decision. The article remains focused even though it covers several related searches.

By comparison, a cluster of thin pages often repeats the same introduction, examples, and product claims. Readers move between pages without gaining a coherent view, while the site creates unnecessary overlap.

Comprehensive does not mean adding every fact available on the subject. It means answering the questions that naturally arise before the reader can make a decision.

5. Structure Content for Readers and Search

Structure content for readers and Search by making answers easy to find while keeping related ideas logically connected. Use clear headings, answer-first openings, full paragraphs for explanations, bullets for grouped information, tables for comparisons, and visuals where they improve understanding.

Some points to consider:

• Very long paragraphs can make a page difficult to scan. A page made entirely from one-sentence paragraphs creates a different problem: every thought appears to carry the same weight, and the reader loses the relationship between ideas.
• Headings should help readers understand where the discussion is going. They should describe the subject of the section rather than force every idea into a question.
• Transitions also deserve attention. A section about original content can lead naturally into first-hand experience because experience is one source of originality. A section about complete topic coverage can then lead into structure because deeper content needs a clear route through it.
• Every section should naturally lead to the next, so the article reads like one continuous argument rather than 12 disconnected tips.

6. Optimize Images and Videos for AI Search

Optimize images and videos for AI Search by using visuals that explain, demonstrate, compare, or provide evidence rather than simply decorate the page. Relevant visual content can also appear alongside web links in AI-powered search experiences.

Useful visuals include:

• A comparison table showing meaningful differences
• A process diagram explaining how a system works
• A product screenshot annotated with important details
• A chart built from original or properly sourced data
• A before-and-after example
• A video demonstrating a complex workflow
• A decision tree that guides the reader towards an option

A technical article about website migration could include a redirect map. A warehouse article could show how inventory travels through a facility. A software comparison could display how the same task is completed in each product.

Captions and surrounding text should explain the relevance of the visual. Descriptive filenames and accurate alt text also provide useful context, while keyword-stuffed alt text creates a poorer experience.

The simplest editorial test is to remove the image temporarily. When the reader loses an explanation, example, or piece of evidence, the visual is doing useful work.

7. Make Important Pages Crawlable and Indexable

Make important pages crawlable, indexable, and eligible to appear with a Search snippet. These conditions are necessary before a page can support an AI Overview or AI Mode response, although eligibility does not guarantee inclusion.

Content teams should work with technical teams to review pages that contribute directly to revenue or authority. These may include product pages, comparison pages, service pages, original research, and high-value educational resources.

Pay particular attention to:

• Pages blocked by robots.txt
• Accidental noindex directives
• Canonicals pointing towards outdated URLs
• Important content available only after login
• CDN or firewall rules that block crawlers
• Orphan pages with no internal links
• Snippet controls that limit how content can appear
• JavaScript that fails to render key information

For larger websites, server log analysis can show whether search crawlers are actually reaching important URLs, how frequently they visit them, and whether crawl resources are being spent on duplicate, outdated, or low-value pages.

Search Console now also includes a setting for inclusion in generative search experiences. Inclusion is the default, while exclusion removes a property’s links and content from supported generative features.

Most commercial websites will want to remain included. Publishers with licensing or content-control concerns may choose to assess the setting more carefully.

8. Add Crawlable Text to Interactive Content

Add crawlable supporting text to calculators, dashboards, videos, animations, and interactive product experiences so their essential information can be understood without relying entirely on the interaction.

Take a pricing calculator. A search system may struggle to understand the page when the pricing assumptions, available options, and interpretation of the output are never explained in text. Visitors may face the same difficulty before they begin entering information.

A supporting explanation should cover:

• What the tool calculates
• Which inputs does it use
• How the result should be interpreted
• Who the tool is designed for
• Which assumptions affect the estimate
• Where the calculation has limitations

The same approach applies to videos and product demonstrations. A written summary helps readers understand what the visual covers and gives search systems enough context to retrieve the page for relevant questions.

Interactive features work best as part of a complete page. They should deepen the explanation rather than carry the entire burden of communicating it.

9. Improve Page Experience After the Click

Improve page experience after the click by giving visitors fast access to useful information that goes deeper than the AI-generated answer they have already seen. As AI responses satisfy more preliminary information needs, the visitors who do click need a clear reason to stay.

A useful landing experience includes:

• Fast and reliable loading
• A clear relationship between the search question and the opening
• Navigation that helps readers reach relevant sections
• Limited interruption from pop-ups and advertisements
• A visible next step
• Mobile formatting that preserves tables and visuals
• Current information, rather than an article left untouched for years

Content duplication also affects the experience. Five overlapping articles can confuse readers about which page is current and force search systems to choose between several weak candidates.

Consolidating related pages often creates a stronger result. The best material can be combined, outdated claims can be removed, and internal links can point towards one authoritative resource.

10. Use AI-Generated Content With Human Review

Use AI-generated content with human review so automation supports research and production without replacing editorial judgment. Search policies focus on the value and quality of the finished page, while large-scale production without meaningful additions can fall under scaled content abuse.

Used carefully, AI can help with:

• Organizing research notes
• Transcribing interviews
• Identifying gaps in an outline
• Creating an initial structure
• Comparing claims across sources
• Updating repeated product information
• Checking consistency across a content library

Editorial responsibility remains with the publisher.

Statistics need verification. Product claims need current evidence. Quotes need accurate attribution. Examples need context. Recommendations should reflect someone who understands the subject and can recognize when an output sounds plausible but is wrong.

Disclosure can also be useful when automation played a substantial role, particularly when explaining the process helps readers understand how the work was produced.

The production method should support the quality of the article. It should never become the reason the page exists.

11. Keep Product and Local Business Information Accurate

Keep product and local business information accurate because AI-powered Search can surface prices, availability, locations, business details, and other commercial information directly during discovery.

For ecommerce and location-based businesses, content alone cannot carry the entire search strategy. Merchant Center feeds and Business Profiles need regular maintenance so the information appearing across search experiences remains accurate.

Review:

• Prices and promotional offers
• Inventory availability
• Store locations
• Opening hours
• Contact details
• Product images
• Delivery terms
• Return information
• Service areas

A useful article may introduce a customer to the brand. Incorrect pricing, availability, or opening hours can end the journey before a purchase or visit takes place.

Hence, commercial data should lie inside the search workflow rather than being treated as a separate administrative task.

12. Measure AI Search Visibility and Business Impact

Measure AI Search visibility against engagement and business outcomes rather than relying on impressions or citations alone. Visibility shows where a brand appears, but clicks, qualified visits, leads, conversions, and revenue show whether that visibility creates value.

A useful measurement framework should combine three levels:

How to measure AI search performance

Measurement level	Questions to answer	Example metrics
Visibility	Where is the brand appearing?	AI impressions, visible pages, countries, and devices
Engagement	What happens after discovery?	Organic clicks, engaged sessions, returning visitors, and branded searches
Commercial impact	Does the visibility contribute to growth?	Leads, demos, trials, subscriptions, assisted conversions, and revenue

A page may gain many AI impressions and very few direct clicks. Another may receive less visibility but attract visitors who are closer to a purchase.

Both cases need analysis at the website and conversion level. Citation or impression counts cannot explain the quality of demand on their own.

7 tactics marketers can safely ignore

The rapid growth of AI search has created demand for fast answers. That environment naturally attracts tactics that promise a simple route to generating results.

The new documentation is useful because it removes several of those distractions.

1. You Do Not Need to Rebuild SEO Around GEO or AEO

You do not need to rebuild your SEO strategy around GEO or AEO to appear in Google’s AI search features. AI Overviews and AI Mode continue to rely on Google’s broader Search infrastructure and systems.

GEO and AEO can be helpful labels when teams discuss AI visibility, citation monitoring, or the way generated answers represent a brand.

They do not require a separate foundation.

AI Overviews and AI Mode continue to use the main Search index, ranking systems, and quality systems. Technical SEO, original content, crawlability, and authority remain central to visibility.

A company can add AI-specific reporting and research to its existing search program. Rebranding the whole discipline does not create an advantage by itself.

2. You Do Not Need llms.txt for Google Search

You do not need an llms.txt file to improve visibility in Google Search because Google does not use it for AI Overviews, AI Mode, or traditional Search.

Other services may choose to support it, so companies can still maintain one for those systems. It should remain well below crawlability, indexing, and content quality on the priority list.

3. You Do Not Need Separate AI Versions of Every Page

You do not need separate Markdown files, simplified AI pages, or machine-readable copies of every page to appear in Google’s generative search features. A standard crawlable and indexable page is sufficient.

Duplicate versions increase maintenance work and create another place where claims, links, or product information can become outdated. They can also complicate canonicalization when several versions cover the same material.

One complete, current, and well-structured page is easier to maintain than several versions created for speculative technical reasons.

4. You Do Not Need to Break Every Article Into Tiny Chunks

You do not need to break every article into tiny chunks for AI Search. Section length should follow the complexity of the subject and what the reader needs to understand it.

Some sections may need a direct 50-word explanation. Others may require examples, evidence, and a table. Forcing every idea into two sentences usually weakens the transitions and makes the article feel fragmented.

Good structure creates clear sections without destroying the relationship between ideas.

5. You Do Not Need a Special AI-Search Writing Style

You do not need a special AI-search writing style to make content understandable to Google. Robotic phrasing, excessive keyword repetition, and question headings above every paragraph are unnecessary.

Modern search systems can understand synonyms, related concepts, and pages that do not repeat the exact wording of every possible query.

Write in the language your audience uses. Define specialist terms when needed. Use direct answers where they help the reader, and expand the explanation when the subject requires context.

A page should sound like a knowledgeable person explaining the topic clearly.

6. There Is No Ideal Word Count for AI Search

There is no ideal word count for AI Search because the appropriate length depends on the question, evidence, and complexity of the reader’s decision.

A narrow question may be fully answered in 600 words. A technical guide or research-heavy teardown may need several thousand. The subject, evidence, and decision complexity should determine the length.

Longer articles become weaker when additional words repeat the same idea. Very short articles fail when they leave out the information a reader needs to act.

Length is an editorial decision, not a ranking formula.

7. Do Not Buy Artificial Mentions or Invent AI Schema

Do not buy artificial mentions or add invented AI schema in an attempt to influence Google’s AI-generated results. Manufactured references provide little durable value, and there is no special schema.org markup that unlocks AI Overviews or AI Mode.

Structured data should continue to support established search features and remain consistent with the visible page.

Authentic reputation develops through useful products, credible research, satisfied customers, and worthwhile coverage. It cannot be replaced by a markup field or a purchased forum comment.

What Remains Uncertain About Google AI Search?

Several parts of Google AI Search remain uncertain, including why individual sources are selected, which fan-out queries run behind a response, how AI visibility affects traffic quality, and how quickly AI Overview coverage will change.

Why Does Google AI Search Cite Some Pages and Not Others?

Google has not published a formula explaining why AI Search cites one eligible page and ignores another. Its documentation explains retrieval and query fan-out, but it does not reveal the weighting used to select individual sources.

A page may rank highly in traditional search and still fail to appear in an AI response. Another may support one part of the answer even when it does not hold a top organic position for the original query.

Marketers can improve eligibility through strong technical foundations, original information, and complete topic coverage. They cannot guarantee a citation for a specific prompt.

Can Higher-Quality Traffic Compensate for Fewer Clicks?

Higher-quality traffic could compensate for fewer clicks, but marketers need to test that through their own conversion and lead-quality data. AI-generated answers may reduce overall visits while sending some users to websites later in the decision journey.

Platform messaging has emphasized prominent links, website previews, and new link designs designed to encourage exploration.

Independent research continues to show significant pressure on organic click-through rates.

A website could receive fewer visitors while attracting a higher proportion of people with strong intent. That possibility has to be tested through each company’s analytics, lead quality, and conversion data.

Can You See Which Query Fan-Out Searches Google Uses?

No, marketers cannot see the exact query fan-out searches Google runs behind an individual AI response. Content teams can predict likely supporting questions, but the precise queries remain hidden.

Content teams can predict common follow-up questions through customer interviews, search data, sales conversations, and product research. They cannot see the precise set of searches generated for every AI response.

The best defense against that uncertainty is a content strategy built around real customer decisions rather than a list of prompt variations.

How Often Do AI Overviews Appear in Google Search?

AI Overview frequency changes significantly over time, so there is no stable percentage that marketers can apply across all Google searches.

In an analysis of more than 10 million keywords between January and November 2025. AI Overviews appeared for 6.49% of the tracked queries in January, climbed to 24.61% in July, and fell to 15.69% in November.

Likewise, an analysis of more than 600,000 commercially focused keywords found an average 71% increase in AI Overview appearances across ten industries over six months.

Forecasts based on one month, industry, or keyword set will age quickly. The direction is clear, AI answers are expanding across the search journey, while the exact frequency remains volatile.

How Do Google Ranking Updates Affect AI Search Visibility?

Google ranking updates can affect AI Search visibility by changing which pages and domains are available or competitive for retrieval. This makes it difficult to separate traffic changes caused by AI-generated answers from those caused by broader ranking changes.

A March 2026 core update was followed by another core update in May, while a global spam update rolled out in late June. These updates were not limited to AI Overviews, but they could still change which pages and domains are available for AI systems to retrieve.

This makes attribution difficult. A loss of traffic may come from an AI-generated answer reducing clicks, a ranking update changing organic visibility, or both happening at roughly the same time. Marketers should compare Search Console, analytics, ranking changes, and conversion data before connecting a performance shift to one cause.

Discover also received a dedicated core update in February 2026. The change affected how articles are surfaced in personalized feeds, reinforcing the need for publishers to produce original, relevant reporting rather than treating Discover as another headline-distribution channel.

What Should Marketers Do Next?

Marketers should start by improving the pages already connected to revenue, product adoption, or brand authority rather than launching a separate AI-search content program.

Confirm that they are crawlable and indexed. Review whether each page answers a complete customer problem. Replace generic claims with evidence, expert commentary, and examples from real work. Consolidate overlapping articles and strengthen the internal links between educational and commercial content.

Next, review the wider search experience. Product feeds, business details, visuals, page speed, and mobile usability all influence what a visitor sees before and after the click.

Finally, update the measurement. Rankings remain useful, though they no longer describe the entire search journey. AI impressions, branded demand, qualified visits, lead quality, and assisted conversions give a more complete view.

The search industry will continue producing new terms and shortcuts. The official guidance offers a more durable direction: make the website technically accessible, give every important page a clear purpose, and publish information that deserves to be retrieved.

AI has changed the shape of the results page. The standard for useful information remains demanding: original evidence, clear reasoning, genuine experience, and a page worth visiting after the summary ends.

AI search rewards what generic content cannot provide: original evidence, real experience, and information worth retrieving.

First Round already knew founders and business leaders with valuable experience. Ricketts brought that knowledge to a wider audience, turning their lessons, mistakes, and methods into practical articles for founders.

The partners of First Round Capital knew founders, executives, and operators who had built important technology companies. The firm held more than 70 community events each year, while its portfolio leaders regularly exchanged advice on hiring, product development, fundraising, and management. The knowledge was already there. Much of it, however, remained inside meetings, private introductions, and conversations that ended when people left the room.

Camille Ricketts helped First Round bring that private advantage into public view.

She became the founding editor of First Round Review, the firm’s long-form editorial publication for founders and startup operators. Through the Review, Ricketts drew out the thinking of experienced operators and turned it into detailed company-building guidance. Over time, founders outside First Round’s portfolio could learn from the same kind of expertise circulating within its network, often before they had ever spoken with one of its investors.

First Round could say it had a strong founder community. The Review gave people a chance to see what that community actually knew.

Before First Round, she learned how to make expertise useful

Ricketts had wanted to become a journalist from an early age. She edited newspapers in school and college before beginning her professional career at The Wall Street Journal. Later, she returned to San Francisco and joined VentureBeat, where she covered climate and green technology.

Her reporting at VentureBeat frequently brought her into contact with Tesla. After roughly two years, Tesla’s head of marketing and communications approached her about joining the company. The move took her from reporting on companies to helping one explain itself.

At Tesla, Ricketts worked in communications during the company’s early years. She later moved into content strategy at the nonprofit Kiva. Together, those roles showed her that company-produced content did not have to sound like advertising. Reporting, research, and careful editing could still sit at the center of the work. A company could explain complicated ideas without reducing them to slogans or promotional claims.

By the time she arrived at First Round, Ricketts understood both sides of the editorial relationship. She knew how a journalist searched for a story, but she also understood why executives often became cautious once the questions began.

That background became useful almost immediately. First Round needed someone who could recognize valuable operating knowledge, ask enough questions to uncover it, and organize the answers without replacing the expert’s voice with the writer’s own.

The opportunity was larger than producing a steady stream of blog posts. First Round was surrounded by people who had already worked through difficult problems. The challenge was finding the parts of their experience that another founder could use.

First Round had the opportunity before it had the editorial system

First Round formally introduced First Round Review on June 28, 2013, after working on the project for four months. By then, more than 30 articles had already been published. The firm described its ambition as building something similar to a Harvard Business Review for startups.

The idea came from a gap First Round had noticed in startup media. Plenty of publications covered funding rounds, product launches, acquisitions, and industry news. Far fewer explained how companies hired teams, built products, changed direction, or handled difficult operating decisions.

A reader could usually find out that a company had succeeded. Learning how the people inside it reached that result was much harder.

First Round believed those explanations were sitting inside the minds of experienced practitioners. The firm already heard them at portfolio events and in conversations between founders. The Review offered a way to capture that knowledge before it disappeared.

Ricketts joined in the summer of 2013, initially as a freelancer. In the beginning, she handled the interviews, topic development, and writing herself. She later became First Round’s head of content and marketing.

As she took control of the editorial work, the publication began to develop a clearer identity.

Rather than asking First Round’s investors to comment on every market trend, the Review featured people who had carried out the work. An article about hiring could come from someone who had built a team. A piece on pricing could examine the decisions of an operator who had tested different approaches and lived with the consequences.

The subject did not have to work for a First Round portfolio company. What mattered was the quality of the person’s experience and whether it could be made useful to someone else. Ricketts described the larger goal as championing exceptional people and making their advice available, even when they had no direct connection to the firm.

That choice gave the Review a purpose beyond covering successful companies. It could examine the decisions, failed attempts, and working methods behind the visible result.

The editorial process brought out what conventional interviews missed

Much of the Review’s strength came from the work completed before an article was written.

Ricketts and her small editorial team worked with each subject to identify a precise topic. It had to connect what the person understood deeply, what they were prepared to discuss honestly, and what a startup founder could apply.

That early focus kept the interview from drifting into broad career advice. Instead of asking an executive to describe their leadership philosophy, the team could examine a particular hiring decision, product change, management problem, or period of rapid growth.

Questions were shared before the interview. The team then conducted a conversation of roughly an hour, transcribed it, and shaped the material into a structured article. With the central subject already defined, the editor could spend more time asking about the sequence of events, the trade-offs, the failed attempts, and the exact process that had been used.

Ricketts also treated the interviewee as a partner in the process. Subjects were allowed to review what would be published. She found that this made people more comfortable discussing weaknesses, personal mistakes, and difficult moments that might help another founder.

The practice differed from traditional journalism, where allowing a source to review an article can raise concerns about editorial independence. First Round Review was working toward a different outcome. It was not an investigative newsroom trying to hold its subjects to account. Its purpose was to document operating knowledge accurately and in enough detail to make it useful.

Because the subject understood that purpose, the conversation could sometimes move beyond the careful answers executives normally give in public. Collaboration helped the team reach details that a shorter, more controlled interview might have missed.

Another decision shaped the publication’s voice. Ricketts removed the writer’s byline from most pieces. At first, the reason was practical. She was writing nearly everything herself and did not want every article to make the Review look like her personal blog.

Eventually, the absence of a byline became part of the editorial style. Ricketts wanted readers to feel that they were hearing directly from the featured operator. The editor still chose the material, clarified the ideas, and created the structure. Yet the person who had lived through the experience remained at the center of the article.

The pieces were also unusually detailed. Ricketts said they commonly ran between 2,000 and 5,000 words. Still, length was never valuable on its own. The extra space gave the team room to include steps, examples, decisions, and mistakes that would otherwise have been cut.

Her standard was practical: readers should be able to use the advice that day or soon afterward.

As a result, the Review rarely felt like a collection of polished executive opinions. Its best articles showed how a person had thought through a problem and what another operator could learn from the process.

The Review made First Round valuable before the pitch

Ricketts understood that the Review could not earn trust if every article eventually turned into an advertisement for First Round.

For that reason, the publication regularly featured operators from outside the firm’s portfolio. It also avoided attaching an aggressive fundraising message or sales-oriented call to action to every article. Ricketts believed readers would be less willing to share something that felt like a promotional page disguised as useful advice.

First Round had to trust that consistently valuable work would shape how people saw the firm.

At the same time, the Review had a clear business purpose.

First Round wanted promising founders to become familiar with the firm and eventually consider it when raising capital. Ricketts acknowledged that a wider readership could introduce First Round to more entrepreneurs. The publication also showed those founders the quality of the firm’s knowledge and relationships.

In practice, the content became proof of what First Round could offer.

The firm did not need to repeatedly claim that its network could help founders. Readers could see the people it had access to and learn from them directly. With every strong interview, the community surrounding First Round became a little more visible.

Ricketts also worked to expand the publication beyond First Round’s website. Email was part of the distribution strategy from the beginning. Later, publications including Fast Company and Inc. approached the team about syndicating its articles. That early interest led Ricketts to pursue similar arrangements with Quartz and Business Insider. Shortened versions could appear on those sites and direct interested readers toward the complete article.

Meanwhile, the team remained active on social media and paid particular attention to influential readers who shared its work.

To help track that activity, the team built something called the Influencer Bot. When someone in its audience with more than 10,000 followers shared or discussed a Review article, the team received an alert. It could then engage with that person, amplify the mention, or share it internally with the First Round team.

As the archive grew, discoverability became another challenge. Ricketts and First Round platform leader Brett Berson reorganized the publication into nine digital magazines covering subjects such as management, fundraising, product, and company culture.

The team still published only about two new articles each week. The redesign did not create nine separate publishing operations. Instead, it made years of existing knowledge easier for readers with different interests to navigate.

Together, these choices helped the Review become more than a marketing channel. It began to function as a growing library of company-building knowledge.

What did her work achieve?

First Round Review showed encouraging results early. By June 2013, some articles were receiving more than 1,000 tweets, while the new content had contributed to a fivefold increase in visitors to First Round’s website. Its first article, about Etsy’s effort to increase the number of female engineers by almost 500% in one year, received tens of thousands of views and attracted coverage from several major publications.

The audience continued to grow. By 2015, the Review was averaging more than 250,000 unique visitors per month over a three-month period, with readers spending more than four minutes on the page. The publication was also contributing to First Round’s investment pipeline and had helped the firm close deals.

Around the same period, monthly page views reached approximately 480,000, while individual articles averaged about 2,000 social shares. The Review later grew to roughly half a million monthly readers and more than 100,000 email subscribers. Its email audience eventually passed 127,000.

What the figures show is that Ricketts helped build an editorial property that reached the people First Round most wanted to serve. More importantly, she gave the firm something competitors could not reproduce by simply publishing longer articles or increasing output.

The real advantage came from the combination of access, interviewing skills, editorial judgment, and trust. Ricketts created a standard that made experienced people willing to share what they knew and made founders willing to keep reading.

What can leaders learn from Camille Ricketts’ work at First Round?

The clearest lesson from her time at First Round is not that every company should launch a long-form publication.

First Round Review worked because its editorial model grew out of something the firm genuinely possessed: access to experienced operators with useful knowledge. Ricketts created a disciplined method for recognizing that knowledge, drawing it out, and turning it into material that founders could apply.

She also understood that the person with the most valuable story was not always the most famous executive. The operator who designed the hiring process, repaired the product strategy, or built the customer system might have more to teach than the person who regularly appeared on conference stages.

That insight expanded the publication’s range. It also made its advice more practical because the people being interviewed had often worked directly on the problem being discussed.

Most companies talk about their networks, expertise, and customer knowledge. Far fewer make those advantages visible in a form that people outside the organization can use.

Ricketts showed First Round how to do that.

Her work made the firm’s relationships visible, its knowledge reusable, and its support for founders believable. First Round’s network had always been valuable. Through the Review, people outside it could finally understand why.

The best content does more than show expertise. It turns experience into something another person can use.