I rely on Google Search Console because it is excellent at collecting search data. The challenge is that it still does not make interpretation easy.
When I open almost any property, I usually find thousands of queries, landing pages, impressions, clicks, rankings, and click-through rates. That volume is useful, but it can quickly become overwhelming when I am trying to answer one simple question: what should I do next?
For years, my workflow was familiar: export the data into Excel or Google Sheets, build pivot tables, apply filters, and start digging for patterns. That approach works, but it is slow. More often than not, I am searching for insights without knowing exactly what I am looking for.
That is where AI makes the workflow more useful. I use it to speed up the hardest part of Search Console analysis: finding meaningful patterns hidden across thousands of rows of search data.
I think of Google Search Console as my source of truth and AI, whether ChatGPT or Claude, as the analyst sitting beside me. GSC shows me what happened. AI helps me explore why it happened, uncover opportunities I might miss, and organize messy data into decisions I can act on.
A quick note on regex
Most of the examples I use start in the same place inside Google Search Console: Performance → Queries → + Add Filter → Query → Custom (regex).
From there, I enter a regular expression to filter query data before exporting it for analysis.
The useful part is that I no longer need to memorize regex syntax. I can ask ChatGPT to write it for me. For example, I might prompt: Create a regex for Google Search Console that matches queries beginning with question words.
ChatGPT may return something like (?i)^(who|what|why|how|can|does|will|should)b.
If I need something more specific, I simply describe the pattern I want. I might ask for a regex that matches queries containing five or more words, identifies comparison searches, or finds branded queries that include product names.
The better I describe the pattern, the better the regex usually becomes.
Here are seven practical ways I combine Google Search Console with AI so I can spend less time sifting through data and more time making decisions.
1. I stop looking only at queries and start looking at intent
Most Search Console analysis still happens at the keyword level. The problem is that people do not really search by keyword. They search with intent.
Instead of reviewing thousands of individual queries one by one, I use regex to isolate investigation-focused searches before exporting the data.
One useful regex is (?i)^(best|top|vs|review|reviews|compare|comparison).
After exporting the filtered query data, I ask Claude or ChatGPT to classify intent. My prompt is usually something like: Categorize these queries into informational, navigational, investigation, transactional, and local intent. Return a CSV with classifications and confidence scores.
This helps me spot patterns that are difficult to see keyword by keyword. Informational traffic may be growing while commercial investigation queries are declining. Transactional queries may rank well but earn weak click-through rates. Comparison searches may be driving impressions without having dedicated content to support them.
When I segment by intent, the next steps become much clearer.
2. I discover questions my audience is already asking
Question-based keyword research is not new, but AI helps me identify themes across hundreds of question-oriented searches much faster.
I start with a regex like (?i)^(who|what|where|when|why|how|can|does|should|will)b.
Then I export the results and ask Claude or ChatGPT: Group these questions into common themes and identify unanswered topics.
A Google Search Console query filter highlights how regex can narrow SEO performance data, helping marketers turn thousands of search terms into focused insights.
Instead of manually reviewing hundreds of questions, I can quickly see broader patterns around pricing concerns, product comparisons, implementation challenges, and industry-specific use cases.
This becomes more than a content exercise. I can use these themes to improve FAQs, support resources, sales enablement materials, and AI Overview optimization.
The best opportunities are often not hidden in one query. They are hidden in clusters of related questions.
3. I find queries likely to trigger AI Overviews
Google does not give me a filter for queries likely to trigger AI Overviews, but I can build a useful approximation.
I start by isolating common informational and comparison patterns with a regex like (?i)^(what is|how to|best|vs|difference between|guide to).
Then I export the matching queries and ask Claude or ChatGPT: Review these queries and group them by the content format needed to answer them effectively.
The themes often fall into definitions, tutorials, comparisons, or expert recommendations.
This helps me see where my content may need to shift from simply ranking for keywords to becoming the best available answer. Increasingly, those are not always the same thing.
4. I track emerging trends earlier
Traditional keyword research can be reactive. By the time a trend is obvious in keyword tools, competitors may already be building content around it.
Google Search Console can help me identify shifts earlier, as long as I know how to look for them.
Instead of searching for individual keywords, I use ChatGPT to build regex around broader concepts. For example, I might prompt: Create a Google Search Console regex to identify searches related to AI agents, copilots, assistants, automation, and autonomous workflows.
The output may look like (?i)(ai agent|agentic|copilot|assistant|automation).
This same approach works for new technologies, product categories, competitors, industry buzzwords, and changing customer concerns.
Once I filter and export the data, I let AI look for emerging themes. A prompt I like is: Review these queries and identify emerging themes, new terminology, and shifts in search behavior. Highlight which topics appear to be gaining traction, recommend whether they deserve a new content asset or an update to an existing page, and identify any patterns that could influence our content strategy.
Instead of only confirming that a trend exists, AI helps me decide whether the trend is meaningful enough to act on and what the next move should be.
5. I surface conversion intent inside informational traffic
One of the most overlooked opportunities in Search Console is finding bottom-of-funnel signals inside queries that appear informational at first glance.
I might ask ChatGPT: Create a regex for searches that indicate evaluation, comparison, pricing, alternatives, migration, implementation, or vendor selection intent.
An example output is (?i)(cost|pricing|price|vs|alternative|compare|implementation|migration).
I apply that regex to the query report, export the filtered data, and then ask Claude or ChatGPT to analyze it.
My prompt usually looks like this: Review these Google Search Console queries and identify recurring buying signals. Group them into themes such as pricing, comparisons, implementation, and vendor evaluation. Recommend which existing pages should better address this intent, and identify opportunities to improve content through stronger CTAs, internal links, comparison tables, FAQs, or supporting resources.
A visual metaphor for AI turning messy Google Search Console queries into clear SEO decisions, separating qualified intent from irrelevant traffic signals.
I often find that pages created for top-of-funnel education are already attracting visitors who are evaluating solutions. In that case, the best opportunity may not be creating a new page. It may be improving the page that already earns the visit, so users can take the next step without breaking the informational experience.
Sometimes the biggest content opportunity is recognizing the conversion intent already reaching the pages I have.
6. I find audience-specific opportunities
One of my favorite ways to uncover new content opportunities is filtering queries by industry, audience, or customer segment. It quickly shows me whether my content is resonating with the audiences I intended to reach or revealing opportunities I had not considered.
I start by asking ChatGPT to create a regex based on the audience segments that matter most to the business.
For example, I might prompt: Create a Google Search Console regex that identifies queries related to healthcare, manufacturing, retail, education, financial services, government, and nonprofit organizations.
An example output is (?i)(healthcare|hospital|medical|manufacturing|factory|retail|education|school|financial|bank|government|public sector|nonprofit).
After applying the filter and exporting the results, I ask Claude or ChatGPT: Analyze these queries and group them by audience segment. Identify which industries show the strongest search demand, what recurring questions or pain points each audience has, and recommend opportunities for new content, landing pages, case studies, or internal linking that would better serve those audiences.
The differences can be valuable. Healthcare searches may consistently focus on compliance, while manufacturing queries may revolve around implementation. Retail searches may reveal entirely different use cases than financial services searches.
7. I uncover striking-distance opportunities at scale
Every SEO knows the classic advice: look at keywords ranking in positions 5-15 to identify opportunities within striking distance.
The challenge is doing that at scale. A report with hundreds of queries where a site is close to stronger rankings can become overwhelming fast.
I take the regex patterns above a step further. I apply the filters that match my goals, then narrow the report to positions 5-15 before exporting the queries.
Then I ask my AI analyst: Identify recurring themes across these queries and recommend page-level optimizations rather than keyword-level optimizations.
Instead of getting tiny recommendations for individual keywords, I often uncover larger opportunities. A page may be missing subtopics, comparison details, stronger internal links, or use cases that would make it more complete.
The result is usually fewer optimizations, but more meaningful ones.
Turning Search Console data into decisions
As an SEO, I do not have a data shortage. I have a prioritization problem.
Google Search Console remains one of the richest sources of insight into how people discover a business. The difficult part is turning thousands of rows into something actionable.
That is where AI fits into my workflow. It helps me uncover patterns, organize information, and surface opportunities I might otherwise miss. It is not a replacement for SEO strategy, experience, or critical thinking.
The real advantage is not writing better regex or exporting cleaner spreadsheets. It is spending less time searching for insights and more time acting on them.
Because data does not improve SEO. Better decisions do.
I can publish consistently, follow SEO best practices, and still watch competitors outrank me. When that happens, I usually find that the issue is not content quality alone. It is content coverage. Competitors are answering questions my audience is already asking, while my site is not fully part of that conversation yet.
That is where I use content gap analysis. It helps me identify the topics competitors rank for that I do not, then decide which opportunities are actually worth pursuing.
Finding gaps is rarely the hard part. SEO tools make that fairly easy. The real challenge is making sense of thousands of keywords across several reports and deciding what deserves attention first.
My workflow combines competitor data, first-party search data, and AI so I can prioritize content opportunities around business impact instead of search volume alone.
I bring my SEO data together before analyzing it
In this workflow, I use Semrush to identify competitive opportunities, Google Search Console to validate where my site already shows signs of authority, and Google Analytics to add business context. Then I use Claude to bring those datasets together, group related opportunities, identify patterns, and help me decide what belongs on the content roadmap.
I follow this process in one of two ways.
I export reports directly from the platforms and upload them to Claude.
If I have connected those platforms through MCP (Model Context Protocol, a standard that allows AI models to connect securely to data sources), I let Claude pull the data directly without manual exports. The workflow changes, but the analysis does not.
Here is the process I use to turn a pile of SEO data into a prioritized content plan.
Step 1: I choose the right competitors
A content gap analysis is only as useful as the competitors I compare myself against. That sounds obvious, but it is one of the easiest places to go wrong.
If I compare my site to Amazon, Reddit, or Wikipedia, I will end up with thousands of keyword “opportunities” that were never realistic in the first place. My goal is not to find every site ranking for my target keywords. My goal is to find businesses competing for the same audience.
I usually start with Semrush’s Organic Competitors report. Instead of relying only on a list of known competitors, I use this report to find domains that compete across many of the same keywords. From there, I narrow the list to three to five sites that closely match the business and target audience I am analyzing.
I do not worry if a few familiar names do not make the cut. Business competitors and organic search competitors are not always the same.
I also filter out sites that can distort the analysis, including large marketplaces like Amazon, community-driven sites like Reddit or Quora, reference sites like Wikipedia, local directories, review sites, and publishers that do not directly compete with the business.
There are exceptions. If I am analyzing a publisher, comparing against other editorial sites makes sense. The key is choosing competitors that create the type of content I am realistically trying to outperform.
A Semrush competitor analysis view turns organic search data into a clear map of rival domains, traffic potential, keyword overlap, and content gap opportunities.
Before I move forward, I sanity-check the competitor list with stakeholders. Sales or product teams may know about newer competitors or strategically important niches that do not yet show up clearly in Semrush.
Once I have settled on the right competitors, I am ready to find the gaps that matter most.
Step 2: I gather and prepare the data
With the competitor list finalized, I collect the data Claude will analyze. Whether I upload exports or connect through MCP, the goal is the same: bring together competitive rankings, my site’s search performance, and engagement data so I can separate meaningful opportunities from noisy keyword lists.
I like to pull data from three core sources.
Semrush: I find the gaps
I start with Semrush’s Keyword Gap tool using the competitors selected in Step 1.
From there, I pay close attention to three buckets: keywords competitors rank for and I do not, keywords where I rank but competitors rank higher, and keywords where I rank but competitors do not.
The first bucket often points to missing topics or content hubs. The second bucket can reveal quicker wins, especially when my site already appears on Page 1 or Page 2. The third bucket shows existing strengths that I should protect and continue building around.
Google Search Console: I validate the opportunity
Next, I check Google Search Console before assuming every missing keyword deserves a new page.
For example, Semrush may show that I do not rank for a keyword, but GSC might reveal that I already receive impressions for closely related queries. That tells me Google has started associating my site with the topic, even if rankings have not caught up yet.
Those “almost there” topics often deserve a higher priority than topics where I would be starting from scratch.
In GSC, I look for queries with high impressions and average positions between 8 and 20, existing pages ranking for related terms, and long-tail queries that reveal additional search intent.
Google Analytics: I add business context
Search volume is only part of the story. Engagement metrics help me answer a more important question: if I improve visibility for this topic, is it likely to support business goals?
A Semrush content gap analysis view reveals where a competitor ranks and the analyzed site does not, turning keyword overlap data into a practical roadmap for SEO content opportunities.
I review metrics such as organic sessions, engagement rate, average engagement time, key events or conversions, and landing page performance.
If a related content hub already drives engaged visitors or conversions, expanding that topic may be a smarter investment than chasing a completely new keyword with higher search volume.
I clean the data before handing it to Claude
If I am manually downloading the data and uploading it to Claude, I clean it first. Claude is excellent at finding patterns, but it can only work with the data I provide. Cleaner data leads to cleaner topic clusters and better recommendations.
I remove duplicate keywords, competitor-branded terms, careers queries, login queries, support queries, locations or product lines outside the business, keywords with clearly different search intent, and high-intent commercial keywords that are too broad to compete for.
For a manual workflow, I export Keyword Gap data from Semrush, query data from Google Search Console, and landing page performance data from Google Analytics, then upload the files to Claude. For a connected MCP workflow, I ask Claude to retrieve the Keyword Gap report, GSC query data, and GA4 landing page metrics directly from connected accounts.
Step 3: I ask Claude to find the story in the data
At this point, I should have a clean dataset that combines competitive keyword gaps, Search Console performance, and Google Analytics data.
This is where the workflow becomes much more useful. Instead of scrolling through thousands of rows looking for patterns, I ask Claude to organize the data into something I can actually build a strategy around.
The mistake I see most often is asking AI to “cluster these keywords.” That usually produces clusters based on keyword similarity alone. That can be useful, but it does not tell me what to do next.
Instead, I ask Claude to think like an SEO strategist. I give it context about the business, including products or services, target audience, primary business goals, content priorities or constraints, and the exported or connected data from Semrush, GSC, and Google Analytics.
Then I ask Claude to organize opportunities by search intent, funnel stage, business relevance, existing authority signals from GSC, user engagement from GA4, recommended content format, and internal linking opportunities.
Rather than returning a spreadsheet of grouped keywords, I want Claude to produce topic clusters with a clear recommendation for each one.
For example, one cluster might be labeled Technical SEO Audits and include supporting keywords, estimated opportunity, existing pages that could be updated, whether a new page is needed, internal linking recommendations, a priority score, and the reasoning behind the recommendation.
A content gap workflow turns scattered SEO signals into topical clusters, showing where AI search visibility, privacy-first analytics, and technical SEO need deeper coverage.
Another cluster might reveal that several competitor keywords can be addressed by expanding an existing guide instead of publishing three separate articles. That is the kind of insight that is hard to spot manually but much easier for AI to surface.
I separate quick wins from long-term investments
Not every opportunity belongs on the same roadmap. As part of my prompt, I ask Claude to classify each cluster into quick wins, new content opportunities, and authority plays.
Quick wins are existing pages that can be refreshed, expanded, or better optimized. New content opportunities are topics that deserve dedicated content because the site has little or no visibility. Authority plays are larger subject areas that may require multiple pieces of content and ongoing investment to compete effectively.
This simple step helps me move from an overwhelming keyword list to a roadmap with both short-term wins and long-term initiatives.
I do not skip the human review
Claude can organize information remarkably well, but it does not know the business the way I do.
Before moving on, I ask whether the topic supports business goals, whether multiple search intents are being combined into one cluster, whether existing content could already satisfy the need, whether the opportunity is realistic given authority and resources, and whether I would actually assign the topic to a writer.
If the answer is no, I refine the cluster or remove it.
The goal is not to accept every AI recommendation. The goal is to spend less time organizing data and more time making strategic decisions.
The biggest prompt lesson is simple: I do not ask Claude to organize keywords. I ask it to recommend what my content strategy should be based on the data I have provided.
Step 4: I score and prioritize the opportunities
Once Claude has grouped the keywords into topic clusters, the next step is deciding what deserves attention first.
This is where many content gap analyses fall apart. Teams naturally gravitate toward the biggest search volumes, but volume is only one piece of the puzzle. A topic that attracts qualified visitors and supports business goals is often a better investment than a high-volume keyword that is difficult to rank for or unlikely to convert.
I score each opportunity across several criteria before I build a roadmap.
A prioritized content gap roadmap turns scattered SEO data into clear next moves, ranking quick wins by impact, effort and AI visibility.
Business relevance
I start with a simple question: if this content performs well, does it help the business?
Topics aligned with products, services, or the customer journey should carry more weight than informational topics with little commercial value.
Existing authority
Next, I look at signals from Google Search Console. If my site already earns impressions or ranks on the second page for related queries, Google has likely established some level of topical authority.
In those cases, improving an existing page or expanding a content hub may produce results much faster than starting from scratch.
Search demand
Search volume matters, but I do not let it dominate the scoring model.
A collection of related long-tail queries with moderate demand can sometimes generate more qualified traffic than one broad keyword.
Ranking difficulty
I review the current search results before committing to a topic. I look at whether authoritative brands dominate the first page, whether the intent is informational, commercial, or transactional, what types of content are ranking, and whether I can realistically create something more useful or complete.
This quick reality check keeps me from chasing opportunities that are not practical.
Estimated effort
Finally, I consider the work involved. Some opportunities require a light refresh of an existing article. Others call for a new content hub supported by multiple pages.
Both can be worthwhile, but they should not carry the same priority when resources are limited.
I let Claude apply the framework
Once I define the scoring criteria, Claude can evaluate every topic cluster consistently.
For example, I may ask Claude to score each opportunity on a five-point scale for business relevance, existing authority, search demand, ranking difficulty, and content effort. Then I ask it to calculate an overall priority score and explain why each recommendation received that score.
A tactical SEO refresh brief turns AI-assisted content gap analysis into page-level priorities, surfacing validation lessons, effort estimates, and the biggest opportunities.
The explanation is just as valuable as the number. If I disagree with a recommendation, I can adjust the weighting, add more business context, and ask Claude to score the opportunities again.
By the end of this step, I have more than a list of content ideas. I have a prioritized content strategy that shows what to tackle next, what can wait, and what is not worth pursuing.
Step 5: I turn priorities into page-level recommendations
Once I have prioritized the opportunities, the next step is figuring out exactly what to change.
Rather than handing a team a ranked list of topics, I ask Claude to generate page-level recommendations for the highest-priority opportunities. This is where connected data becomes especially valuable.
Because Claude has access to Semrush research, Google Search Console performance, Google Analytics metrics, and my prioritization framework, it can evaluate each page in context instead of treating every recommendation the same.
For each priority page, I ask Claude to produce a recommendation that explains why the page was selected, the primary keyword cluster, current rankings and impression data, supporting evidence from GSC and competitor research, recommended updates, estimated effort, expected impact, and priority level.
One of the biggest advantages of this approach is validation.
Before recommending a refresh, Claude can compare URL-level Search Console data against the original analysis. Sometimes what looks like a strong opportunity turns out to be misleading. A keyword may have inflated impression counts, a URL could have been mislabeled in an export, or the page may not be as close to ranking as it first appeared.
Catching those issues before assigning work can save hours of unnecessary effort.
The recommendations also make stakeholder conversations easier. Instead of saying, “I think we should update this page,” I can point to the supporting data, explain why it is a priority, estimate the effort involved, and tie the recommendation back to the larger content strategy.
I treat these recommendations as implementation plans rather than full content briefs. They help SEO and content teams understand what should change, why it matters, and where to focus first. Writers can then use those recommendations to create or update content with confidence.
Step 6: I measure whether the gap is closing
Publishing the content is not the finish line. It is the start of the next round of analysis.
Rows of illuminated data cabinets and paper files stretch into the distance, capturing the pressure on marketers to turn fragmented customer data into a smarter performance engine.
I begin with Google Search Console, tracking whether target queries are gaining impressions, improving in average position, and generating more clicks. When I refresh an existing page, I compare performance before and after the update to see whether the changes actually moved the needle.
Next, I look at Google Analytics. Better rankings do not always translate into better business outcomes, so I review organic traffic alongside engagement and conversion metrics. If an updated page attracts more visitors but fails to keep them engaged or contribute to conversions, I know it is time for another round of optimization.
If I am using Claude through MCP, I can also ask it to compare performance over time and summarize what changed. I might ask which refreshed pages improved the most, which content clusters gained the most visibility, which recommendations drove the strongest business results, and which opportunities still need attention.
Instead of comparing reports month after month, Claude can quickly surface significant changes and point me toward the pages that deserve attention.
I do not treat content gap analysis as a one-time exercise. Competitors publish new content, search behavior shifts, and my own site authority evolves. I like to repeat this workflow every quarter, or more often in fast-moving industries, so I can keep finding new opportunities and stay ahead of competitors.
The tools will continue to improve, but the repeatable workflow is what creates the advantage.
I build a repeatable content gap analysis process
A content gap analysis helps me prioritize opportunities worth pursuing instead of chasing every possible keyword.
Semrush helps me uncover competitive gaps. Google Search Console shows where I already have momentum. Google Analytics adds the business context that rankings alone cannot provide. Claude brings those datasets together, helping me identify patterns, prioritize opportunities, and create actionable recommendations in a fraction of the time it would take manually.
Whether I upload reports or connect my tools through MCP, the workflow stays the same. I gather the right data, validate the opportunities, let AI organize the information, and apply my own expertise to decide what comes next. That is the part AI cannot replace.
The biggest advantage is not simply having better prompts or faster analysis. It is having a repeatable process that helps a team make smarter content decisions every quarter.
Prompt template: My prioritized content gap roadmap
Here is the prompt I use after I have gathered the data, whether I have uploaded exports from Semrush, Google Search Console, and Google Analytics or connected those tools to Claude through MCP.
“You are an experienced SEO strategist helping me perform a content gap analysis.
I’ll either provide exported reports from Semrush, Google Search Console, and Google Analytics, or you’ll access those tools through connected MCP integrations.
My goal is to identify the highest-impact content opportunities based on competitor visibility, existing authority, business value, and implementation effort.
Here’s my business context:
– Company: – Industry: – Products/services: – Target audience: – Primary business goals: – Geographic focus: – Any strategic priorities or constraints: – Tone of voice: [Insert brand voice adjectives here (e.g., authoritative, conversational, technical)].
Using the available data, complete the following tasks.
1. Identify content gaps
Organize keywords into these categories: – Competitors rank and we don’t. – We rank below competitors. – We rank and competitors don’t.
Highlight any content gaps, opportunities to consolidate pages, or keyword cannibalization issues.
2. Validate the opportunities
Use Google Search Console data to determine: – Which topics already receive impressions. – Which pages rank between positions 8 and 20. – Which existing URLs have the strongest chance of improving with optimization.
Use Google Analytics data to determine: – Which pages drive meaningful engagement. – Which pages contribute to conversions. – Which content hubs are worth expanding.
3. Create strategic topic clusters
Group related opportunities by: – Search intent – Business relevance – Funnel stage – Recommended content type – Internal linking opportunities
Don’t cluster based only on keyword similarity. Focus on topics that should become part of the same content strategy.
4. Prioritize every opportunity
Score each topic cluster using: – Business relevance – Existing authority – Search demand – Ranking difficulty – Estimated effort
Assign each opportunity a priority (High, Medium, Low) and explain why.
Separate recommendations into: – Quick wins – New content opportunities – Long-term authority investments
5. Recommend next steps
For every high-priority opportunity, recommend whether we should: – Refresh an existing page – Consolidate multiple pages – Create a new page – Build a pillar page with supporting content
Include supporting evidence for every recommendation.
6. Deliver the results
Create: – An executive summary – Prioritized topic clusters – A scored opportunity table – Page-level recommendations for the highest-priority URLs – A phased implementation roadmap (30, 60, and 90+ days)
If you find conflicting data between Semrush, Google Search Console, and Google Analytics, explain the discrepancy and recommend which source should guide the decision. The output should both be HTML and a Google Sheet.
Before presenting your final recommendations, validate your own analysis. If reviewing Search Console or Analytics data changes your original recommendation, explain why and update your prioritization accordingly.”
This prompt is only a starting point. I add business context, editorial guidelines, and scoring criteria that are unique to the organization I am analyzing. The more context I give Claude, the more useful and actionable its recommendations become.
A year ago, I watched the industry place its bets on which AI platform would own discovery. Perplexity looked like the search-native challenger. Copilot looked like the enterprise Trojan horse. In the data I’m seeing now, neither bet has really paid off.
Previsible (disclosure: I’m its CPO and co-founder) just published its third AI Traffic Study, based on 6.77 million LLM-driven sessions. What stands out to me is the level of consolidation. Monthly LLM sessions grew 9.9x, reaching 644,478 in May 2026, and 92.4% of that traffic came from one platform.
The plateau was a pause
In mid-2025, it looked like AI traffic might be topping out in some sectors. I don’t think that was the real story.
Sessions climbed from 65,249 in November 2024 to 396,278 by August 2025. Then they dropped sharply in November 2025 before reaching new highs of 428,203 in February 2026 and 644,478 in May.
That November dip deserves context.
Sessions fell 50% in a single month, driven almost entirely by ChatGPT referrals dropping from 448,412 to 213,345. Other platforms were mostly steady. To me, that points to a model-related change. We’ve already seen small product shifts create major swings in referral traffic, including last fall, when many sites lost half their ChatGPT traffic because the model began favoring Wikipedia and Reddit. By December, sessions had recovered to 442,609.
The lesson I take from this is simple: one vendor’s product decision can cut your AI traffic in half overnight. I would plan for that volatility instead of treating AI referrals as a stable channel.
Consolidation, not competition
When we last published in December 2025, ChatGPT held about 84% share. Perplexity followed at 8.9%, Gemini at 4.5%, Copilot at 2.1%, and Claude at 0.6%. Six months later, the field had moved even more decisively toward the leader.
Across the full dataset, ChatGPT now commands 92.4% of trackable LLM referral traffic. It grew 12.8x over 19 months, with no clear sign of slowing. It is the only LLM sending meaningful referral volume at scale, which means I would not talk about “AI visibility” without putting ChatGPT first.
There is one important caveat. This study measures standalone LLM referral traffic. AI discovery inside Google’s own results, including AI Overviews, almost certainly drives more AI traffic than all standalone platforms combined. But that operates under a different measurement model, so it is not included here.
The challengers flipped
The surprise is not that ChatGPT is on top. What I find more interesting is the movement beneath it.
Claude
Claude grew 64x, moving from 133 sessions in November 2024 to 8,528 in May 2026. It overtook Perplexity in March 2026 for the first time, and it stayed ahead.
Claude was mostly flat through 2025, then accelerated 4x in two months as its agentic tools and enterprise integrations gained adoption. The enterprise advantage many people expected Copilot to win may be materializing for Claude instead.
If your audience includes technical buyers, developers, or professional services, I would treat Claude visibility as material now. The early positioning window is still open, but it may not stay that way for long.
Gemini
Gemini is the quiet number two in this dataset. It delivered 3.2x growth with very little volatility. Because Gemini is tied into Workspace and Android, I suspect referral numbers undercount its real discovery footprint.
Perplexity & Copilot
Perplexity peaked at 17,507 monthly sessions in March 2025 and has fallen 61% since. Copilot fell even harder, dropping 96% from its August 2025 peak, from 8,651 sessions to 339.
I no longer see either platform as a strong traffic-acquisition growth bet. Both are shifting toward experiences that keep users inside their own environments, including browsers, agents, and modes where they do not need to send traffic out at all.
Where LLMs send users, and why it should change your roadmap
The most actionable finding in the study is not market share. It is where LLMs send people after they decide a site is worth visiting.
ChatGPT sends 28.8% of its traffic to internal search results pages. Across industries, roughly 25% of AI-referred traffic lands on internal search.
Old search marketing tools give way to a faster, connected future, with data streams, AI icons, and a glowing search hub symbolizing SEO innovation and community growth.
My read is that the model trusts the domain but cannot always identify the exact right page. So it sends users to the site’s search box and lets them navigate from there. Because this pattern holds across verticals and time periods, I see it as structural to retrieval-augmented generation rather than a temporary quirk.
That changes the role of internal search. The model already did the hard work of choosing your domain. Now your internal search experience decides whether that high-intent visit converts or bounces.
For most sites, internal search is still treated like a neglected navigation feature. I think it needs to be treated as an acquisition surface.
The vertical-level data tells several different stories. SaaS traffic lands on search pages 34.6% of the time. Publisher traffic lands on news pages 54% of the time, but against 120+ million organic sessions, publisher penetration is only 0.11%. Publishers create the content LLMs cite, yet they capture almost none of the resulting traffic.
Ecommerce traffic tends to land on product pages, often with purchase intent already formed. Education traffic lands directly on course pages 52% of the time, bypassing marketing content. Health traffic lands on About pages 42.1% of the time, suggesting users are evaluating the source before trusting the content. Legal traffic spreads across blog, about, contact, and location pages, which reflects the full evaluation arc.
The platforms have distinct behaviors, too. ChatGPT and Gemini act more like search-pattern models: they show domain trust but page-level uncertainty. Perplexity and Claude behave more like content-selection models, picking specific pages and over-indexing on long-form content.
If your strategy depends on editorial content driving qualified traffic, I would give Perplexity and Claude more attention than their raw share suggests.
What I would do now
First, I would optimize for ChatGPT before anything else and expand to other platforms only when the volume justifies the work. ChatGPT is where the measurable standalone LLM referral traffic is concentrated.
Second, I would monitor Claude closely. It overtook Perplexity in March 2026, and early visibility advantages can compound quickly when a platform is still forming its citation and recommendation patterns.
Third, I would treat product pages as AI entry points. Product pages capture 43% of ecommerce LLM traffic, which makes structured, comparable product data a discoverability requirement rather than a nice-to-have.
Fourth, I would make pricing machine-readable wherever possible. “Contact us for pricing” gives AI systems very little to summarize, compare, or recommend.
Fifth, I would prioritize internal search. It is not just a navigation feature anymore. For AI-referred users, it may be the first real conversion point.
Finally, I would track AI traffic by page type instead of relying only on site-wide averages. Your overall AI traffic number can hide where the real concentration is. A pricing page, for example, might run 3x your site-wide penetration.
The next question I want answered is conversion rate by LLM platform. Which platforms send users who buy, and which send users who bounce?
We built this dataset to answer that. If the last 19 months are any guide, I expect the answers to change faster than most teams are prepared for.
About the data
This analysis includes 166 GA4 properties from November 2024 through May 2026, spanning SaaS, ecommerce, finance, legal, health, insurance, education, publishing, and ticketing. All 166 properties are present throughout the full 19-month window, so I’m looking at behavioral change rather than sample expansion.
Agentic SEO is the newest branch of search optimization, and I see it as one of the most important shifts marketers need to understand now. Instead of focusing only on traditional search rankings, agentic SEO is about earning visibility, trust, and conversions inside agentic search platforms such as ChatGPT Agent and Claude CoWork. Many marketers expect it to become a major acquisition channel by 2027.
To identify the top agentic SEO companies, I evaluated 38 firms in Q2 2026 and scored each one across five weighted factors.
AI Visibility Score (30%): How often an agency’s clients appear in AI citations.
SEO, GEO & ASO Expertise (25%): The depth of an agency’s approach to SEO (Search Engine Optimization), GEO (Generative Engine Optimization), and ASO (Agentic Search Optimization).
Notable Clients (20%): The recognizable brands an agency has worked with.
Average Review Score (15%): Direct client satisfaction feedback.
Leadership Experience Score (10%): How long senior staff have worked in the search optimization industry.
After reviewing the data, I found six companies that stood out from the rest. Below, I break down each agency’s strengths, specialty, scores, and client review themes.
I ranked First Page Sage first because it is the only firm on this list with a dedicated agentic SEO practice. Its work is built on a commercial GEO methodology the firm has been running since 2023. I also like that its offering connects SEO, GEO, and Agentic SEO into one integrated program, so clients do not need separate teams managing traditional organic search and agentic AI visibility.
First Page Sage’s approach focuses heavily on content, thought leadership, on-site authority, and independent third-party coverage. These are the kinds of signals AI models often draw from when evaluating whether a brand should be cited, recommended, or included in a comparison. Its Agentic SEO and GEO work also helps clarify how major models position a client at the comparison stage, including content that explains who a product is for and reduces the ambiguity that can cause AI agents to skip a vendor.
In one campaign, a skincare brand’s AI sentiment score rose 4 points across tracked models in roughly 14 weeks, and agents recommended the brand more often across the board. I see this approach as especially strong for considered-purchase categories such as B2B software, financial services, and healthcare, where AI endorsement can influence which vendors make a buyer’s shortlist.
Summary of online reviews: In client feedback, First Page Sage’s Agentic SEO work is described as “incredibly innovative and well-executed,” with reviewers saying the agency has “become a game-changing part of [their] marketing strategy.” Several clients also say the team “helped [them] gain a first mover advantage within [their] industry” and describe staff as “extremely detail-oriented and communicative.”
Genevate
I ranked Genevate second because it is one of the few firms built specifically for the generative-search era. Genevate focuses on the external signals large language models pull from, including media placements, authoritative third-party mentions, and a consistent brand narrative across the web. By combining GEO with digital PR, the firm helps shape how ChatGPT, Perplexity, and Claude describe and recommend a brand to prospective buyers.
The main limitation I found is Genevate’s shorter track record. Founded in 2025, it has one of the briefest operating histories on this list, and its scope is intentionally narrow. The firm focuses on GEO and reputation management rather than broader agentic search or traditional SEO. For brands that already have media momentum and want tighter control over their AI narrative, that focus can be an advantage. For companies that need a wider service mix, I would weigh that limitation carefully.
Summary of online reviews: Early clients describe Genevate as a “responsive partner” with a strong grasp of “how AI platforms describe and recommend brands,” and they speak positively about the firm’s “PR-led approach.” Teams looking for a full-service marketing partner “may find the offering is best paired with separate performance marketing.”
Driven Metrics
I see Driven Metrics as a strong fit for small and mid-sized businesses that want disciplined SEO and GEO without the cost structure of a larger agency. Its methodology starts with buyer intent, mapping each keyword to a funnel stage instead of chasing search volume alone. From there, the firm produces content across landing pages, pillar articles, guides, FAQs, and other formats while also handling the technical work needed to keep a site indexable and ready for generative AI systems.
Because Driven Metrics is a younger and smaller firm, I would place it in a different category than the larger enterprise agencies on this list. Its GEO work appears credible and structured, but its operating history is shorter and its client base leans more mid-market than enterprise. For growth-stage companies that want a practical SEO and GEO foundation, it is a natural starting point.
AI Visibility Score: 4.4
SEO, GEO & ASO Expertise: 4.4
Notable Clients: AutoStar Transport Express, Dignity Health, Affirmed Home Care
Average Review Score: 4.7
Leadership Experience Score: 4.3
Specialty: Results-focused SMB SEO with GEO capabilities
Summary of online reviews: Clients appreciate Driven Metrics’ “clear, well-constructed process” and describe the team as “communicative and easy to work with.” Reviewers also note that the agency has “a limited track record and few documented wins to point to,” and that buyers from larger firms “may find the experience underwhelming.”
Seer Interactive
I ranked Seer Interactive highly because it brings more than two decades of SEO experience into AI search. Data is the center of its model. The firm runs large-scale analyses to identify where clients should focus, then builds content and technical improvements around what the numbers show. Seer has also packaged its GEO offering as a defined product and has published AI search experiments since 2023, with work cited by Search Engine Land, Semrush, and Ahrefs.
On the technical side, Seer is especially useful for enterprise-scale problems that smaller agencies may struggle to handle. That includes large site architectures, technical markup at scale, and measurement systems that track how a brand appears in AI answers across thousands of pages.
The tradeoff is that Seer’s breadth can dilute specialization. Most of its case studies focus on traffic and conversions rather than pipeline or revenue, and the full-service model spreads senior attention across many client needs. I would consider Seer a strong fit for larger organizations that want AI search handled alongside a broader SEO program, but not necessarily for buyers who want a dedicated agentic specialist above all else.
AI Visibility Score: 4.2
SEO, GEO & ASO Expertise: 4.8
Notable Clients: LinkedIn, Intuit, Capital One, Autodesk
Summary of online reviews: Clients credit Seer with “deep analytical horsepower and senior, experienced teams,” and they value its “measurement-first approach.” The process is also described as “thorough, but slow,” and “lean teams wanting quick turnarounds will feel it.”
Omniscient Digital
I view Omniscient Digital as the content-led choice on this list. The agency builds GEO around editorial quality, original expertise, and the brand and author authority signals AI models consider when selecting sources to cite. Its work also includes digital PR, machine-readable markup, and tracking across major LLMs as part of each engagement.
The downside is that this kind of editorial depth takes time. Results may build more gradually than they would with a narrow technical fix. Engagements also start at around $10,000 per month, with no self-serve option, which puts Omniscient out of reach for many earlier-stage teams. The firm also shares relatively little about its GEO methodology publicly, so buyers have limited visibility into exactly how citations are tracked and improved.
For well-funded B2B software companies that want to own the language of their category, I think Omniscient can be a strong option. For teams with tighter budgets or urgent timelines, the ramp-up period and pricing are important considerations.
Summary of online reviews: Clients praise Omniscient for “editorial-quality writing and thoughtful, well-scoped strategy.” Some also note that the work is “slow to ramp.”
Go Fish Digital
I included Go Fish Digital because it brings a long technical SEO history into the GEO conversation. The agency built its GEO practice around Barracuda, a proprietary AI platform that scores pages against the signals AI systems use to evaluate and cite sources. Go Fish maps how AI interprets a site’s coverage, then improves machine-readable markup, authority signals, and fact density to make a brand more likely to appear in AI answers.
The biggest limitation I found is the lack of published GEO-specific results. That is common in such a new field, but it still matters when comparing providers. I would consider Go Fish Digital a good fit for established brands that want a technical, platform-backed AI search approach from an agency with a longer operating history.
Summary of online reviews: Reviewers consistently praise Go Fish Digital’s “full-package capability” across technical SEO, content, and PR, and call the team “responsive and well-organized.” The clearest gap is GEO proof: “the process is clearly defined, but public results in AI search are still thin,” which reflects how new the practice remains.
The Top Agentic AI Optimization Companies by Specialty
I also compared these leading agentic AI optimization companies by buyer segment, since the right agency depends heavily on company size, budget, and search maturity.
On June 11, 2026, I saw more than 1,000 marketing leaders come together in New York for Zero Click New York, Profound’s largest AI Marketing summit to date.
What stood out to me was the range of leaders and brands shaping the conversation. Speakers from Coca-Cola, LinkedIn, Delta Air Lines, U.S. Bank, and CVS Health shared how they are rethinking marketing strategy, team design, and measurement as AI changes the way audiences discover and trust information.
I also found the research sessions especially important. The summit explored Claude’s citation mechanics, ChatGPT’s emerging ads business, and the data behind the kinds of content AI systems are most likely to trust. Together, these conversations made Zero Click New York 2026 feel like a clear marker for where AI Marketing is heading next.
I see vibe coding as one of the most accessible ways to create small pieces of software with AI tools like ChatGPT, Cursor, Replit, and Gemini. Instead of writing code line by line, I describe what I want in plain language, receive working code in return, paste it into an environment like Google Colab, run it, and test the result.
In this guide, I’ll explain how I approach vibe coding, where I think it works well, where it breaks down, and which SEO examples can inspire practical projects of your own.
Vibe coding variations
I use “vibe coding” as a broad term, but it helps to separate it from nearby approaches:
Type
Description
Tools
AI-assisted coding
AI helps write, refactor, explain, or debug code. I usually associate this with developers or engineers who already understand the systems they are building.
GitHub Copilot, Cursor, Claude, Google AI Studio
Vibe coding
AI handles most of the work after I provide the idea or prompt.
ChatGPT, Replit, Gemini, Google AI Studio
No-code platforms
Platforms handle what I ask for through visual interfaces, drag-and-drop workflows, or background automation. Many now use AI, but they existed before AI became mainstream.
Notion, Zapier, Wix
For this guide, I’m focusing only on vibe coding.
The barrier to entry is low. In most cases, I only need a ChatGPT account, free or paid, and access to a Google account. Depending on the project, I may also need API access or subscriptions to SEO tools such as Semrush or Screaming Frog.
I also like to set expectations early: by the end of this kind of workflow, I’m usually aiming to run a small program in the cloud. If I want to build a SaaS product or software I plan to sell, I treat AI-assisted coding as the more realistic path because it usually requires more technical knowledge, more testing, and more budget.
Vibe coding use cases
I find vibe coding most useful when I’m working with clear buckets of data and need a helpful outcome, not a perfect one. That might mean finding related internal links, adding pre-selected tags to articles, comparing groups of URLs, or building something playful where the output does not need to be exact.
For example, I built an app that creates a daily drawing for my daughter. I type a short phrase about something she told me, such as “I had carrot cake at daycare.” The app uses examples of drawing styles I like and a few pictures of her, then generates a drawing as the final output.
When I ask for precise changes, the tool often gets worse. I once asked it to remove a mustache, and instead it recolored the image. That is exactly the kind of limitation I expect with this approach.
If my daughter were a client reviewing every detail, I would need someone with Photoshop or similar skills to make exact edits. For this use case, though, the result is good enough, and that is where vibe coding shines.
I would be cautious about building commercial applications solely through vibe coding. Some companies may even need vibe coding cleaners to clean up AI-generated work. But for demos, MVPs, internal tools, and quick experiments, I see vibe coding as a useful shortcut.
How I create SEO tools with vibe coding
When I create an SEO tool with vibe coding, I usually follow three steps:
I write a prompt describing the code I need.
I paste the code into a tool such as Google Colab.
I run the code and check whether the results match what I expected.
Here’s a real prompt example from a tool I built to map related links at scale. After crawling a website with Screaming Frog and extracting vector embeddings through the crawler’s OpenAI integration, I vibe coded a tool to compare topical distance between the vectors for each URL.
This is exactly what I wrote in ChatGPT:
I need a Google Colab code that will use OpenAI to:
Check the vector embeddings existing in column C. Use cosine similarity to match with two suggestions from each locale (locale identified in Column A).
The goal is to find which pages from each locale are the most similar to each other, so we can add hreflang between these pages.
I’ll upload a CSV with these columns and expect a CSV in return with the answers.
After ChatGPT generated the code, I pasted it into Google Colab, which is a free Jupyter Notebook environment for running Python in a browser. I then used “Run all” to test whether the program produced the output I wanted.
That is the clean version of the process. In practice, AI can make the workflow look perfect while still producing code that does not behave the way I need.
I expect issues along the way, and most of them are simple to troubleshoot if I keep the prompt and testing process clear.
First, I always state the platform I’m using. If I want code for Google Colab, I say that directly in the prompt.
Sometimes I still get code that depends on packages that are not installed. When that happens, I paste the error back into ChatGPT and ask it to fix the code or suggest an alternative. I do not need to fully understand the missing package to move forward. I can also ask Gemini inside Google Colab to identify the problem and update the code directly.
I also check outputs carefully because AI can sound confident while inventing data. One time, I forgot to specify that the source data would come from a CSV file, so the tool created fake URLs, traffic, and graphs. “It looks good” is not the same as “it is correct.”
If I connect to an API, especially a paid API from a provider such as Semrush, OpenAI, Google Cloud, or another platform, I need to request my own API key and keep usage costs in mind.
A Semrush subscription screen highlights 2 million Standard API units and a masked API key, underscoring the setup step needed for SEO automation and vibe-coded tools.
If I want an even lower execution barrier than Google Colab, I can use Replit.
With Replit, I can prompt what I want, and the platform can generate the code, design the interface, and let me test everything in one place. That reduces copy-and-paste work and gives me a shareable URL quickly. I still need to review poor outputs and keep iterating until the app behaves properly.
The tradeoff is cost. Google Colab is free unless I use paid API keys, while Replit charges a monthly subscription and usage-based API fees. The more the app runs, the more expensive it can become.
SEO vibe-coded tools that inspire me
Google Colab is the easiest place for me to start, but SEOs are taking vibe coding much further. I’ve seen people create Chrome extensions, Google Sheets automations, and even browser games.
I’m sharing these examples because they show what is possible when useful SEO ideas meet practical AI tooling. If I see a tool and wish it had a different feature, that is often a sign that I could try building a version for myself.
A Replit vibe-coding session turns a simple prompt for toddler-friendly daily jokes into a published web app, illustrating how AI tools can quickly prototype playful ideas.
GBP Reviews Sentiment Analyzer by Celeste Gonzalez
After vibe coding SEO tools in Google Colab, Celeste Gonzalez, Director of SEO Testing at RicketyRoo Inc, pushed the idea further by creating a Chrome extension. “I realized that I don’t need to build something big, just something useful,” she explained.
Her extension, the GBP Reviews Sentiment Analyzer, summarizes sentiment analysis from reviews over the last 30 days and shows review velocity. It also exports the information to CSV and works on Google Maps and Google Business Profile pages.
Instead of relying only on ChatGPT, Celeste used Claude to create stronger prompts and Cursor to turn those prompts into code.
AI tools used: Claude (Sunner 4.5 model) and Cursor
APIs used: Google Business Profile API (free)
A vibe-coded GBP Sentiment Analyzer turns review data into a quick snapshot, showing negative sentiment trends, key topics, and an export option for SEO workflows.
Platform hosting: Chrome Extension
Knowledge Panel Tracker by Gus Pelogia
I became obsessed with the Knowledge Graph in 2022, when I learned how to create and manage my own knowledge panel. Later, I discovered that Google’s Knowledge Graph Search API lets me check the confidence score for any entity.
That led me to build a vibe-coded tracker that checks entity scores daily, or at any frequency I choose, and returns the results in a Google Sheet. I can track multiple entities at once and add new ones whenever I need to.
The Knowledge Panel Tracker runs entirely in Google Sheets, and the Knowledge Graph Search API is free to use. This guide explains how to create and run it in your own Google account, or you can see the spreadsheet here and update the API key under Extensions > App Scripts.
AI models used: ChatGPT 5.1
A spreadsheet-based Knowledge Panel Tracker turns entity searches into structured SEO data, comparing names, entity types, descriptions, and confidence scores at a glance.
APIs used: Google Knowledge Graph API (free)
Platform hosting: Google Sheets
Inbox Hero Game by Vince Nero
I also like the idea of vibe coding a link building asset. That is what Vince Nero from BuzzStream did with the Inbox Hero Game. The game asks players to use the keyboard to accept or reject a pitch within seconds, and it ends if they accept too many bad pitches.
Inbox Hero Game is more complex than running a small script in Google Colab, and it took Vince about 20 hours to build from scratch. “I learned you have to build things in pieces. Design the guy first, then the backgrounds, then one aspect of the game mechanics, etc.,” he said.
The game was built with HTML, CSS, and JavaScript. “I uploaded the files to GitHub to make it work. ChatGPT walked me through everything,” Vince explained.
A retro arcade-style Inbox Hero screen turns PR pitch triage into a fast keyboard game, challenging players to accept or reject emails before time runs out.
He also found that longer prompt threads became less useful over time, “to the point where [he’d] have to restart in a new chat.”
That became one of the hardest parts of the project. Vince would add a feature, such as a score, and ChatGPT would “guarantee” it had found the error, update the file, and still return the same problem.
In the end, Inbox Hero Game shows that it is possible to create a simple game without coding knowledge. It also shows where a developer becomes valuable when the goal shifts from “working prototype” to polished product.
AI models used: ChatGPT
APIs used: None
Platform hosting: Webpage
How I think about vibe coding with intent
I do not expect vibe coding to replace developers, and I do not think it should. What I do see is a practical way for SEOs to prototype ideas, automate repetitive tasks, and explore creative experiments without a heavy technical lift.
The key is realism. I use vibe coding where precision is not mission-critical, I validate outputs carefully, and I stay alert for the moment when a project grows beyond “good enough” and needs stronger technical support.
When I approach vibe coding thoughtfully, it becomes less about shipping perfect software and more about expanding what I can test. For internal tools, proofs of concept, and SEO side projects, the best results come from pairing curiosity with restraint.
I see existing content as a goldmine, but only when I have a practical way to improve it. The hard part is usually finding the time, and that is where Claude has made a large, messy job feel much more manageable for me.
I do not start by building a giant content audit system. I start with one article, run one focused audit, refine the output, and then turn the prompt into a reusable Claude skill. Over time, those one-off audits become a working library I can improve every time I use it.
I use Claude to uncover topical gaps, flag outdated information, check brand voice, and evaluate whether a page is easy for AI systems to retrieve and cite. The real value comes from iteration: each time I improve a skill, the next audit becomes faster and more useful.
Here are six content audit workflows I would build in Claude. The first four work at the page level, so I can start with a single article before moving into larger library-wide analysis.
Page-level audits
When I am not ready to build a full workflow, I start with page-level audits. These audits only require one article, which means I do not need a content inventory, a data export, or a complicated setup. After each session, I ask Claude to turn the process into a reusable skill for future page-level reviews.
1. Brand voice consistency
I use a brand voice consistency audit when a content library has drifted over time. Voice can shift because of new writers, changing services, product updates, or evolving positioning. This audit helps me spot where a page no longer sounds aligned with the brand.
If I do not have detailed brand guidelines with strong examples, I let Claude extract the voice guide from high-quality content. That usually works better than relying on vague phrases like “conversational but authoritative” or “educational, not too formal.”
I pick three to five articles that represent the brand at its best. If possible, I download them as markdown files and ask Claude to describe how the voice works in concrete terms.
How the articles usually open, such as whether they begin with a direct claim, a counterintuitive statement, or a specific scenario.
How sentences and paragraphs are built, including average length, range, rhythm, and how paragraphs tend to close.
Three to five personality dimensions framed as “We say X, but not Y,” with do and don’t examples.
Words and phrases the brand tends to use, and words or phrases it should avoid.
Specific constructions, phrases, and conventions the brand never uses.
Instead of accepting a vague voice description, I want Claude to return concrete observations. For example, it might say that articles open with a direct claim rather than a scene-setting paragraph, sentences average 15 to 20 words and rarely exceed 30, and transitions are functional, such as “here’s why that matters,” rather than formulaic, such as “furthermore.”
I also want example pairs, such as: “We’d say ‘the data shows three things,’ not ‘there are multiple factors to consider.’” The goal is not to create a voice guide for writers. The goal is to create one an LLM can understand and apply consistently.
Once I like the output, I ask Claude to save it as a skill and evaluate an article against it. If Claude flags issues I disagree with, I update the skill until the feedback becomes useful and repeatable.
I can then use that skill to find voice inconsistencies in older content, check new drafts for alignment, and even generate more on-brand first drafts. I still edit the output, but the starting point is much stronger.
When I need to improve content performance, I use a coverage comparison to find topical gaps. This helps me understand what competing pages cover that my article misses.
I use the Claude in Chrome extension to have Claude review the top three to five ranking pages for my target keyword. Then I ask Claude to compare those pages against my content and highlight the most important gaps.
What competitors are doing well.
What my article already does well.
Where I can improve the piece without bloating it.
If I want the output in a table, I ask Claude to format it that way. If I want a downloadable DOCX for review or handoff, I ask for that instead.
When Claude recommends additions I would never publish, I make a note of those exclusions before packaging the workflow into a skill. That way, the skill gets closer to my editorial standards each time I refine it.
3. Freshness audit
Old content adds up quickly, and it is hard to prioritize refreshes while I am also producing new material. A freshness audit skill helps me identify what needs attention without rereading every older article from scratch.
I give Claude an older article and ask it to flag anything time-sensitive: statistics tied to a specific year, named tools or platforms, references to “current” or “recent” trends, and claims that depend on a market, regulatory, or product context that may have changed. I am not asking Claude to rewrite the article yet. I am asking it to build an issue list I can act on.
If my company has launched new products, removed old services, changed positioning, or updated terminology, I include that context in the input. That helps Claude flag what should be added, removed, or revised.
I use an AEO and AI retrievability audit to understand whether a page is likely to be surfaced in AI-generated answers. Tools such as ChatGPT, Perplexity, and Google AI Overviews tend to favor content that answers questions directly. If an article buries the answer under too much preamble, or structures key information in a way that is hard to extract, it becomes less useful for those systems.
I give Claude the article and the target query, then ask it to evaluate several retrieval signals.
Whether the article answers the main question directly and early.
Whether key statements are specific enough for an LLM to quote or cite.
Where an FAQ-style section would improve clarity.
Whether the page includes authority signals, such as primary research, first-person experience, outbound citations, or specific examples.
Once I save this as a skill, it becomes an extra editor focused specifically on AI visibility and answer retrieval.
Library-level audits
Once I am ready to move beyond individual pages, I use library-level audits. These require performance data, a content inventory, a connector, or a manual export.
5. Performance triage
When I think about a traditional content audit, performance triage is usually what comes to mind. It helps me analyze a content library and identify the pages that deserve attention first.
Before I begin, I make sure Claude has access to the right data through a connector such as BigQuery or the Semrush API. If that is not available, I export the data I normally use for large-scale audits, such as traffic, clicks, engagement metrics, conversions, rankings, and related performance signals.
I ask Claude to prioritize pages that have suffered meaningful performance drops in the past six to 12 months, pages with high impressions but consistently low click-through rates, and pages that have been live long enough to rank but never gained traction.
I also define what a meaningful performance drop looks like for the site I am analyzing, because traffic patterns vary by industry, audience, and page type. Then I ask Claude for a prioritized list of what is worth investigating and why. From there, I use the page-level audits above to diagnose the problem.
If I have run this analysis before, I give Claude the previous output. That helps the skill learn the kind of prioritization and reasoning I expect.
I treat entities as a major part of AEO and semantic search. A topical gap analysis helps me see whether my content library has enough coverage to build authority around the entities tied to my brand.
The core question I ask is simple: what is my content library not covering that it should?
To start, I create a list of target entities. For example, at my agency, I want to be known for SEO and AEO. If I have a clear list of services or products, I can use that instead of a formal entity list.
Using Cowork or Code, I ask Claude to analyze my sitemap and compare it to those target entities. If I have a Screaming Frog export with URLs, page titles, and meta descriptions, I use that as input for a more accurate analysis.
Then I ask Claude to identify topic clusters that are missing or underrepresented based on the target entities, services, or products. If I want prioritization, I can use the Semrush MCP so Claude can check search volume for potential keywords.
Not every gap is worth filling. I filter the results against audience needs, business relevance, and editorial standards. Then I feed those decisions back into Claude so the skill produces better recommendations next time. The final list can go directly into my content creation workflow or be handed off to a content team.
I do not try to audit everything at once
I have seen content audits stall because the scope feels too large, not because the team lacks data. My preferred approach is to pick one audit and one article, run the workflow, save the skill, and use it again on the next piece.
For me, iteration is part of the value. I enjoy taking one Claude skill, improving it, and then chaining it with other skills to uncover more content opportunities. Starting small is what makes the system easier to keep using.
I’m looking at Yahoo! Scout as Yahoo’s most direct return to search and web discovery in years. The new AI-based answer engine is available at scout.yahoo.com, and Yahoo is also weaving it through its major properties, including Yahoo News, Yahoo Finance, Yahoo Mail and Yahoo Search. I think of it as a Yahoo-branded AI companion built to help people move through those familiar Yahoo experiences with more context and guidance.
What Yahoo Scout is. To me, Yahoo Scout is Yahoo’s version of an AI search engine and assistant, similar in broad idea to Google’s AI Mode or OpenAI’s ChatGPT, but with Yahoo’s own personality layered in. Yahoo told me it wanted Scout to feel fun, approachable and easy for people of all ages to understand.
When I first visited Yahoo Scout, the experience felt intentionally warm. The home page includes a search box, a playful slogan and an animated icon above it. Beneath the search box, Yahoo offers suggested searches that can be filtered by topics such as news, finance, sports, shopping and travel. On the left side, I could also see previous queries, making it easier to return to earlier searches and continue where I left off.
The home page also rotates through playful visual treatments. In one version I saw a cowboy hat, while other versions included a crystal ball, a gold medal, a walking cartoon brain and more.
Yahoo Scout’s advantage. The Yahoo Search team gave me early access to try Yahoo Scout. While the interface will feel familiar to anyone who has used other AI answer engines, the Yahoo-specific pieces are what stood out most to me.
Yahoo’s biggest advantage is its existing reach. The company already has a large audience across Yahoo Mail, Yahoo News, Yahoo Finance and Yahoo Search. Yahoo told me it has more than 500 million user profiles, stores signals such as queries, usage and intent, has more than one billion entities in its knowledge graph and processes 18 trillion consumer events and signals across its properties. That gives Yahoo a lot of context it can use to personalize AI search and better categorize queries.
Yahoo also told me it is the second-largest email company and the third-largest search engine.
Because Scout is connected to Yahoo’s own properties, it can bring Yahoo Finance widgets, financial data, tables, citations, weather results, news results and other rich content directly into answers.
“Search is fundamentally changing, and our team has been inspired to use our decades of experience and extremely rare assets to create something uniquely useful for Yahoo’s hundreds of millions of monthly users,” said Jim Lanzone, CEO of Yahoo. “This beta launch is just the starting point. From search to our industry-leading verticals, Yahoo Scout will help our users accomplish their goals online faster and better than ever before.”
Sending traffic to publishers. Jim Lanzone told me Scout is closely tied to Yahoo’s original mission of being a trusted guide to the internet. Because of that, Yahoo says it designed Scout with the open web in mind, including ways to send traffic downstream to content creators and publishers.
In Yahoo Scout responses, I saw large blue highlights over portions of the answer text. When I hovered over those highlights, I could click through to the source. Each response also includes a visible “featured source” area, along with tables, imagery, related news articles and other source-driven elements meant to make publisher links more prominent.
Lanzone told me early AI answer engines have not done enough to send traffic back to the sources behind their answers. Yahoo wants Scout to be an example of how that relationship can work better. Since there is not enough licensing revenue for every publisher to make deals with AI companies, Yahoo is leaning into the historical search model: give users answers, but also send meaningful traffic to the sites that produced the underlying content.
CTR expectations. I asked Yahoo what click-through rate it expects from Yahoo Scout to publishers. The honest answer was that it does not know yet. Yahoo expects to learn from real user data after launch and then iterate to improve downstream clicks.
Yahoo expects queries in Scout to be longer than queries in Yahoo Search. It also expects ad loads to be lighter, and the team hopes click-through rates will be higher than the industry average.
Yahoo also told me it plans to build a way for publishers to see impression and click data in the future. I see that as something like a Yahoo Webmaster Tools-style reporting experience, though crawling and indexing data would still be tied to Microsoft Bing because Bing powers the underlying search index.
Yahoo Scout across Yahoo properties. I expect Scout to show up throughout Yahoo’s ecosystem. Yahoo Search will use Scout-powered AI summaries. Yahoo News will provide article highlights and may include daily digest audio summaries. Yahoo Finance will add an Analyze button powered by Scout. Yahoo Mail will summarize emails and extract action items, such as adding events to a calendar.
Examples of Yahoo Scout in action. Yahoo Scout is not perfect, but for something Yahoo says was built in about six months, I came away impressed.
When I asked Yahoo Scout for help understanding how SEO works, it returned a useful response with citations throughout the summary. SEO is complex, and not everyone would agree with every part of the answer, but the citation structure made the experience more transparent.
I then asked it for sources I could use to find more content on the topic. There were clearly missed opportunities to link out more often, and I shared that feedback with Yahoo. The team agreed there was room to improve.
When I followed up by asking how I could navigate to the sources it had mentioned, Scout did provide links at that point. I also saw citation previews appear when hovering over linked highlights.
I tried several other types of searches as well. For entertainment queries, Scout pulled in news articles with larger graphics and clickable card-style formats. For finance queries, Yahoo brought in Yahoo Finance, though I was not able to generate stock charts during my own testing, even though I saw that capability in a demo. It may still have been in progress at the time.
For weather, I tested Scout on a Sunday morning as a major snowstorm was touching down in New York. I was able to get a Yahoo Weather chart, along with practical tips on how to stay warm.
For sports, I asked about Super Bowl predictions. As a lifelong Jets fan, I also asked whether the Jets had any chance of winning the Super Bowl in the next 10 years. The answer was not especially encouraging, but I was glad to see a chart embedded directly in the response.
For shopping, Scout gave me advice on how to dress for the weather. That is where Yahoo’s commerce strategy becomes more visible.
Ads and commissions. Yahoo Scout will show ads at the bottom of some responses. Commerce-related queries will also be monetized through affiliate commissions, which is already a common revenue model across the web.
Yahoo told me the ads are still powered by Microsoft Advertising, but Yahoo controls how those ads appear inside the Scout experience.
Those ads will be charged on a CPC basis, not on an impression basis like some other AI engines have announced. I also saw product results labeled with “Yahoo may earn commission from these links.”
How Yahoo Scout came together. Yahoo has been hinting for about three years that it wanted to return to the search game. In 2009, Yahoo made a deal with Microsoft to have Microsoft power Yahoo Search, which effectively ended Yahoo’s work on its own search technology. Since then, Yahoo has outsourced search technology until this new Scout effort.
About six months ago, Yahoo acquired Eric Feng’s company to lead consumer search at Yahoo. Feng co-founded the online video platform Mojiti, which Hulu acquired in 2007. He then became Hulu’s founding CTO and head of product. Before that, he worked in Microsoft Research on search-related problems.
“Yahoo’s deep knowledge base, 30 years in the making, allows us to deliver guidance that our users can trust and easily understand, and will become even more personalized over the coming months,” said Eric Feng, Senior Vice President and General Manager of Yahoo Research Group, the creators of Yahoo Scout. “Yahoo Scout now powers a new generation of intelligence experiences across Yahoo, seamlessly integrated into the products people use every day.”
Lanzone, who also has a long history in search from his years as CEO of Ask.com, told me Feng has been instrumental in building Yahoo Scout over the past six months. Yahoo says this first public release is only the beginning, and more iterations and improvements are expected.
Anthropic and Claude. Yahoo Scout is not built on Yahoo’s own LLM. Yahoo partnered with Anthropic and uses Claude as Scout’s primary foundational AI model. Anthropic, founded in 2021 by former OpenAI employees including Daniela Amodei and Dario Amodei, has become one of the leading AI companies. Amazon announced an investment of up to $4 billion in September 2023, Google committed $2 billion the following month, and as of November 2025 Anthropic had an estimated value of $350 billion.
Even though Scout uses Anthropic’s foundational AI models, Yahoo has customized the experience and combined it with proprietary Yahoo data. Running the same searches directly on Anthropic’s tools would not produce the same Yahoo Scout experience.
“When you’re serving hundreds of millions of users, you need AI that can do more than retrieve information – it has to reason, synthesize, and explain. Yahoo is building toward a more personalized, trustworthy kind of search, and Claude’s ability to deliver that quality of guidance at scale is at the heart of Yahoo Scout,” said Ami Vora, Head of Product at Anthropic.
Microsoft Bing. Microsoft Bing data is also part of Yahoo Scout. Bing provides the underlying search index, but Yahoo says the responses, ranking and overall experience are Yahoo’s. Yahoo wrote that Scout builds on its long-standing Microsoft relationship by using Microsoft Bing’s grounding API, combining that API with Yahoo’s trusted data and content ecosystem so answers are informed by authoritative sources across the open web.
Yahoo is also joining Microsoft’s Publisher Content Marketplace pilot. Microsoft says that marketplace can help support publisher revenue, and Yahoo described the move as “reflecting a shared commitment to expanding publisher reach, connecting original work with new audiences, and supporting sustainable revenue opportunities for publishers.”
Hallucinations. I asked Yahoo about hallucinations, and the company told me it has added many guardrails to reduce them as much as possible. Yahoo says its entity graph, news content and other Yahoo-specific data help ground the answers. The team believes Scout’s hallucination rate should be “very low” compared with other AI engines.
Yahoo Scout blends AI search with commerce, surfacing winter parka recommendations, affiliate shopping cards and trusted weather sources in one answer-style interface.
Agents. Many AI engines are moving toward agentic experiences that can complete tasks for users. Google, OpenAI and Microsoft are all investing heavily in this area.
Yahoo Scout already includes some agent-like elements, especially inside Yahoo Mail, where it can help add calendar events, support smart compose features and surface action items. Yahoo says more is coming on that front.
Why I care. Search is changing quickly, and I find it exciting to see Yahoo step back into the space in a meaningful way. As someone who has followed search for more than 20 years, I appreciate seeing Yahoo try to make search feel fresh again.
Seeing people such as Jim Lanzone, Eric Feng and Brian Provost work on AI search at Yahoo makes this feel like more than just another answer engine launch. I’m interested to see what Yahoo does next.
I have watched the debate around llms.txt become one of the most polarized conversations in web optimization.
Some people treat llms.txt as essential infrastructure for AI discovery. Others, especially longtime SEO practitioners, see it as speculative theater. Platform tools are starting to flag missing llms.txt files as site issues, yet server logs still show that AI crawlers rarely request them.
Google even appeared to adopt it. Sort of. In December, Google added llms.txt files across many developer and documentation sites.
At first, the signal looked obvious to me: if the company behind the sitemap standard was implementing llms.txt, maybe the file really mattered.
Then Google removed it from its Search developer docs within 24 hours.
Google’s John Mueller said the change came from a sitewide CMS update that many content teams didn’t realize was happening. When asked why the files still exist on other Google properties, Mueller said they aren’t “findable by default because they’re not at the top-level” and “it’s safe to assume they’re there for other purposes,” not discovery.
The llms.txt research
I wanted data, not another debate.
So I tracked llms.txt adoption across 10 sites in finance, B2B SaaS, ecommerce, insurance, and pet care. I looked at the 90 days before implementation and the 90 days after.
I measured AI crawl frequency, traffic from ChatGPT, Claude, Perplexity, and Gemini, and the other changes each site made during the same window.
Here is what I found:
Two of the 10 sites saw AI traffic increases of 12.5% and 25%, but llms.txt was not the cause.
Eight sites saw no measurable change.
One site declined by 19.7%.
The 2 ‘success’ stories weren’t about the file
The Neobank: 25% growth
One digital banking platform implemented llms.txt early in Q3 2025. Ninety days later, its AI traffic was up 25%.
That sounds compelling until I looked at what else happened during the same period.
The company ran a PR campaign around its banking license and earned coverage in major national publications.
It restructured product pages with extractable comparison tables for interest rates, fees, and minimums.
It published 12 new FAQ pages optimized for extraction.
It rebuilt its resource center with new banking information and concepts.
It fixed technical SEO issues, including header structure problems.
When a company earns Bloomberg coverage in the same month it launches optimized content and fixes crawl errors, I cannot isolate llms.txt as the growth driver.
The B2B SaaS platform: 12.5% growth
A workflow automation company saw AI traffic jump 12.5% two weeks after implementing llms.txt.
The timing looked perfect. It would be easy to call the case closed. But the surrounding context told a different story.
Three weeks earlier, the company had published 27 downloadable AI templates covering project management frameworks, financial models, and workflow planners. These were functional tools, not ordinary content marketing assets, and they drove the engagement behind the spike.
Google organic traffic to those templates rose 18% during the same period and kept climbing throughout the 90 days I measured.
Search engines and AI models surfaced the templates because they solved real problems and created an entirely new site section. They did not surface them simply because the URLs appeared in an llms.txt file.
The 8 sites where nothing happened after uploading llms.txt
Eight sites saw no measurable change after adding llms.txt. One of them declined by 19.7%.
The decline came from an insurance site that implemented llms.txt in early September. Based on the data, the drop likely had nothing to do with the file.
The same pattern appeared across all traffic channels. Llms.txt did not prevent the decline, and it did not create any visible advantage.
The other seven sites, which included ecommerce brands in pet supplies, home goods, and fashion, plus B2B SaaS, finance, and pet care sites, used llms.txt to document their best existing content. That content included product pages, case studies, API docs, and buying guides.
Ninety days later, nothing changed. Traffic stayed flat. Crawl frequency was identical. The content was already indexed and discoverable, and the file did not change that.
The pattern was clear: sites that launched new, functional content saw gains. Sites that only documented existing content saw no gains.
Why the disconnect?
No major LLM provider has officially committed to parsing llms.txt. Not OpenAI. Not Anthropic. Not Google. Not Meta.
“None of the AI services have said they’re using llms.txt, and you can tell when you look at your server logs that they don’t even check for it.”
That is the reality I saw in the data. The file exists. The advocacy exists. But platform adoption does not show meaningful use yet.
The token efficiency argument and its limits
The strongest case for llms.txt is efficiency. Markdown can save time and tokens when AI agents parse documentation. It gives agents clean structure instead of forcing them through complex HTML, navigation, ads, and JavaScript.
That matters, but mostly for developer tools and API documentation. If your audience uses AI coding assistants like Cursor or GitHub Copilot to interact with your product, token efficiency can improve integration.
For ecommerce brands selling pet supplies, insurance companies explaining coverage, or B2B SaaS companies targeting nontechnical buyers, token efficiency does not automatically translate into traffic.
llms.txt is a sitemap, not a strategy
The closest comparison I can make is a sitemap.
Sitemaps are useful infrastructure. They help search engines discover and index content more efficiently. But I would not credit traffic growth to simply adding a sitemap. The sitemap documents what exists; the content drives discovery.
Llms.txt works in a similar way. It may help AI models parse a site more efficiently if they choose to use it, but it does not make the content more useful, authoritative, or likely to answer user queries.
In my analysis, the sites that grew did so because they:
Created functional assets such as downloadable templates, comparison tables, and structured data.
Earned external visibility through press and backlinks.
Fixed technical barriers such as crawl and indexing issues.
Published content optimized for extraction, including FAQs and structured comparisons.
Llms.txt documented those efforts. It did not drive them.
What actually works
The two successful sites showed me what actually matters.
Create functional, extractable assets. The SaaS platform built 27 downloadable templates that users could deploy immediately. AI models surfaced them because they solved real problems, not because they appeared in a markdown file.
Structure content for extraction. The neobank rebuilt product pages with comparison tables for interest rates, fees, and account minimums. That is data AI models can pull directly into answers without heavy interpretation.
Fix technical barriers first. The neobank fixed crawl errors that had blocked content for months. If AI models cannot access your content, no amount of documentation will help.
Earn external validation. Coverage from Bloomberg and other major publications drove referral traffic, branded searches, and likely influenced how AI models assessed authority.
Optimize for user intent. Both sites answered specific queries, such as “best project management templates” and “how do [brand] interest rates compare?” Models surface content that maps to what users ask, not content that is merely well documented.
None of this requires llms.txt. All of it can drive results.
Should you implement an llms.txt file?
If you run a developer tool and AI coding assistants are a primary distribution channel, I would implement llms.txt. In that context, token efficiency matters because your audience is already using agents to work with documentation.
For everyone else, I would treat llms.txt like a sitemap: useful infrastructure, not a growth lever.
It is good practice to have. It likely will not hurt. But the hour spent implementing llms.txt is often better spent restructuring product pages with extractable data, publishing functional assets, fixing technical SEO issues, creating FAQ content, or earning press coverage.
Those tactics have shown real ROI in AI discovery. Llms.txt has not, at least not yet.
The lesson I take from this is not that llms.txt is bad. It is that we are reaching for control in a system where the rules are still being written. Llms.txt offers comfort because it is concrete, actionable, and familiar. It looks like the web standards we already understand.
But looking like infrastructure is not the same as functioning like infrastructure.
My focus would stay on what is already working:
Create useful content.
Structure it for extraction.
Make it technically accessible.
Earn external validation.
Platforms and formats will change. The fundamentals will not.
AI search is reshaping the marketing landscape faster than anything I’ve seen before.
During my time at Zero Click NY, I witnessed how significantly AI search has evolved over the last six months and identified emerging features that might define its future.
Among all the discussions, these seven trends were the most compelling.
From the emergence of marketing engineers, to the way Claude and ChatGPT differ in results, and Claude’s rapid ascent in the business world over the past year, here are the key insights I gathered.
1. Every AI relies on different content
According to Profound data, only 8% of citations are shared between ChatGPT and Claude. This means 92% of the sources that ChatGPT refers to would not be recognized by Claude for the same inquiry. Essentially, a brand may have high visibility in one AI and not exist in another.
Moreover, each AI favors different types of content.
ChatGPT frequently indexes community content: Reddit, Quora, and forums make up around 16% of its citations.
In contrast, Claude cites listicles 36% and opinion content 13.2% of the time, compared to ChatGPT’s ~20% and 7.2%, respectively.
The disparity also applies to traditional search. A significant 64% of websites Claude cites appear in Google’s top 50 for equivalent queries, whereas it’s only 37% with ChatGPT.
Takeaway: It’s vital to inform stakeholders that AI visibility differs between LLMs, and strategic prioritization is necessary to reach your audience.
Track your visibility by engine because effective strategies in one platform may not translate to another. UGC helps drive ChatGPT while listicles boost presence on Claude.
2. Claude is quietly winning B2B — so sequence your optimization by audience
Claude may appear insubstantial in AI traffic-share charts, but it’s a different story when it comes to enterprise usage.
Web traffic doesn’t tell the whole tale. Anthropic derives about 85% of its revenue from enterprise and API usage, not visible in consumer data.
A critical chart from Ramp’s AI Index reveals the true penetration of Anthropic in the business sector. A year ago, only a small number of businesses used Anthropic. Now, it’s at 34.4%, surpassing OpenAI at 32.3%.
This insight led me to reconsider: if more business users are engaging with Claude and consumers are on ChatGPT, shouldn’t our optimization priorities reflect audience preferences?
Should B2B entities focus on Claude first, while B2C aim for ChatGPT visibility?
Currently, few distinguish between ChatGPT, Gemini, or Claude usage. This distinction is bound to grow.
3. ChatGPT ads are here, and this is what we’re seeing
The game has changed: competitors are securing visibility through ChatGPT ads. These ads are now live and available for self-serve directly within the chat interface.
Recent weeks also saw the debut of GPT 5.5, citation chips morphing into clickable links (leading to a 60% spike in referral traffic overnight), and Google integrating AI Mode into its main search functionality.
This wasn’t incidental. The hyperlinks are crucial for an ads business. Analyzing over 100,000 ad placements highlighted three essential revelations.
ChatGPT Ads match on topic
Ads align with topic similarity, not intent. Only 14% of real user prompts express commercial intent, yet 20% show ads, even if the prompt involves a math problem.
Embedding analysis indicates that ad titles and descriptions significantly influence which conversations you appear in, transforming them into tactical targeting tools.
Paying for ads
We have entered a “pay-to-play” era. Approximately one-fifth of ad placements appear when a direct competitor is mentioned, but only 8% of organic references belong to the associated brand.
Competitors are twice as likely to advertise around your brand’s organic mentions than you are.
For instance, Startup CRM Adia is targeting prompts mentioning Salesforce, with Salesforce responding by showing paid ads 40% of the time, defending their position even when organically mentioned.
Ad inventory is scarce and expensive
Currently, ChatGPT presents about one ad per conversation, with the median exchange spanning three turns. Only 30% of eligible users ever see ads, and CPMs/CPCs are about four times Meta’s rates.
Expect future changes: additional ad slots per reply, ads woven deeper into conversations, and engineered suggestions to prolong interactions, ultimately increasing inventory.
The insight: Understanding both organic AEO and paid defense strategies is essential. Monitoring your brand’s organic citations without tracking who advertises against them offers a partial view.
4. Claude is the most directly optimizable AI right now
Claude sources web content directly from Brave searches, not merely influenced by it, as discussed in the presentation I attended.
In recent testing by Profound, 79.2% of Claude’s citations were directly aligned with Brave’s top 10 search results for equivalent queries.
Reshuffling is minimal; no other AI model trusts its search provider so extensively.
This transparency makes Claude the most straightforward AI to optimize for: a visible index, checkable rankings, and, as we’ll explore next, predictable retrieval.
If I’ve convinced you of the importance of Claude for B2B, here’s your approach: identify where you stand on Brave for key prompts and use that as your roadmap for Claude visibility.
This level of transparency won’t last forever. Take advantage now while it’s possible.
5. Claude only performs web searches a third of the time
There’s a significant caveat: ChatGPT initiates web searches for nearly 95% of prompts, but Claude does so only a third of the time, likely due to cost ($5 per thousand searches via Brave’s API).
You can optimize Claude effectively only when it conducts a search.
The encouraging part is its predictable search habits. Prompts framed around recent events (“best X in 2026”) initiate searches about 81% of the time.
Ranking-related prompts lead to 67% search initiation, location-specific prompts 55%, and comparisons 51%.
Prompts concerning definitions and procedures rarely trigger searches, making them poor targets for Claude optimization.
The lesson: Before investing to enhance Claude visibility for a prompt category, determine if Claude actually conducts searches for it.
Focus on recent events, rankings, locations, and comparisons for effective Claude optimization using Brave rankings as a guide.
Other areas rely on internal memory beyond our reach.
6. Query fan-out: A raffle on one platform, near-deterministic on another
Two speakers offered perspectives on query fan-out, presenting a contrast worth exploring.
Query fan-out entails background synthetic queries to collect content prior to providing an AI-generated response.
Mike King of iPullRank viewed it as a raffle: The task is to gain more tickets through a wider content range across owned, earned, and shared channels, and the right content formats make all the difference.
Even if you rank for a fanned-out query, the wrong format renders you ineligible.
According to his research, content-to-query cosine similarity and information gain strongly correlate with success in AI search.
Conversely, Josh Blyskal from Profound notes that Claude’s fan-outs are highly predictable; the same prompt results in consistent fan-out strings 65% of the time. Interestingly, 94% of Claude’s fan-outs are current-year stamped, unlike ChatGPT’s 17%
Where ChatGPT’s fan-outs constantly evolve, Claude’s remain relatively stable. Thus, both perspectives may hold true for different engines.
With stable fan-outs like in Claude, content creation can directly focus on them. The year-stamping trend suggests using the current year in titles.
For volatile fan-outs as in ChatGPT, King’s approach applies: maximize exposure through format variety.
One mechanism demands two strategies, tailored by engine, potentially requiring prioritization between them.
7. The marketing engineer is here, and agents are the new workforce
The role of a “marketing engineer” might sound like a buzzword, but the hiring trends prove otherwise.
Google’s recently hired its first marketing engineer, Figma has an opening at a $295,000 salary, and both RBC and Autodesk have placed hires.
It’s a rapidly growing search term, and Google’s AI marketing lead dubbed it “the hire for 2026.”
What makes someone ideal for this role? Is the priority given to an engineer learning marketing or vice-versa?
The emerging profile emphasizes marketing experiences such as someone with channel expertise who builds and runs AI systems, reports to the marketing head, and supports the team by removing obstacles. They are marketers advancing the state-of-the-art.
The underlying concept is that marketing functions decompose into pipelines: data extraction, transformations, and loading into useful formats. Agents can now automate these pipelines.
Monitoring competitor pricing and auto-generating sales content.
Scheduling and assessing AEO presence and landing page efficiency.
Analyzing sales call objections and drafting relevant content solutions.
What previously were backlogged tasks now become brief agent-building exercises. Creativity replaces headcount as the limiting factor.
If marketing engineering isn’t a role in your team yet, it’s likely only a matter of time before it is.
The job now: Figuring out how this all works
There remains no definitive roadmap for AI search. When a guidebook emerges, the key step will be prioritizing one LLM contingent upon who you wish to reach.
In many instances, that “who” will now be agents, simultaneously assisting us in our endeavors and highlighting the rising need for professionals adept at engineering such systems.
