The folks over at Schema.org are now providing aggregate usage statistics for Schema.org terms across the public web. This will show you how many domains are using a specific schema/structured data element.

Schema.org announced, “we are pleased to share a new dataset providing aggregate usage statistics for Schema.org terms across the public web.” The dataset are updated monthly and they are aggregated at the domain level and presented in popularity range buckets. “This approach helps filter daily noise while highlighting meaningful adoption trends for researchers and toolmakers,” Schema.org wrote.

What it looks like. Here is a screenshot of two of these Schema.org pages,  author schema and event schema showing the usage statistics towards the top:

More about the data. Schema.org shared more information about the usage statistics over here, here is a summary:

• Schema.org term frequencies are measured within Google’s public web crawling infrastructure. The data is aggregated at the domain level (like example.com), not by individual pages. This means if you use the same term on 100 pages of your site, it still only counts as one domain using it.
• Instead of showing exact, raw numbers (which change daily and can be noisy), websites are grouped into range “buckets” (like “10K – 100K” domains). This keeps the data more stable and protects website privacy.
• HubThe raw files are available on the GitHub site at Google Public Stats dataset on GitHub. They are available in JSON and CSV with the same data and in a JSON summary format with aggregated bucket distributions. They are updated monthly.
• Term Type: The type of term. This is either Type (like “Person” or “Event”) or Property (like “price” or “telephone”).
• URI: The official URI of the term (for example: http://schema.org/Person).
• Domain Count Bucket: The range of unique domains using the term (for example: 100K - 1M domains).

Here is a screenshot of GitHub:

Why we care. Because we love data. Okay, outside of just loving data, knowing if a specific schema element is popular or not may be enough to convince your development team to implement that schema code on your website.

Google Marketing Live may be over, but advertisers still have questions. To help answer some of the biggest ones, Google Ads Liaison Ginny Marvin joined Julie Bacchini and the PPC Chat community for an extended Q&A covering everything from AI Max and AI Overviews to first-party data, AI Brief, measurement and the future of search advertising.

The discussion offered several notable clarifications, particularly around AI Search eligibility, reporting limitations and Google’s growing emphasis on data quality.

AI Max isn’t required for AI Search ads

One of the biggest takeaways from the discussion was Marvin’s clarification that advertisers don’t necessarily need AI Max enabled to appear in AI-powered search experiences.

According to Marvin, campaigns using broad match keywords remain eligible to show ads in AI Overviews and AI Mode. AI Max expands eligibility by applying broad match behavior to phrase and exact match keywords while also enabling keywordless matching.

That distinction matters because many advertisers have assumed AI Max would become a mandatory gateway into Google’s emerging AI search surfaces.

The clarification suggests advertisers still have multiple paths into AI Search inventory, at least for now.

Don’t expect AI Search reporting anytime soon

Advertisers hoping for placement-level reporting received less encouraging news.

Marvin confirmed that ads appearing in AI Overviews and AI Mode are currently reported alongside other top-of-page ads, with no separate performance breakdown available. Google is still evaluating what reporting should look like as these experiences evolve.

That means advertisers will continue to face limited visibility into how much traffic and performance is being driven specifically by AI-powered search experiences.

AI Brief is coming and Google wants advertisers to guide the machine

Several questions focused on AI Brief, Google’s upcoming control layer for AI Max campaigns.

Marvin said advertisers will be able to provide both positive and negative guidance, including instructions such as “never mention prices” or specifying audiences, messaging themes and search intent they want AI systems to prioritize. The feature will also preview sample assets and queries before deployment.

The rollout will begin with AI Max for Search campaigns in English over the coming months before expanding to Performance Max and AI Max for Shopping.

For advertisers concerned about losing control in increasingly automated campaigns, AI Brief appears to be Google’s answer.

First-party data remains Google’s favorite answer

If there was one consistent theme throughout the discussion, it was data.

Marvin repeatedly emphasized the importance of what Google now calls “Data Strength” – the quality and completeness of first-party data flowing into advertising accounts. She highlighted tools such as Enhanced Conversions, Google Tag Gateway, Data Manager and direct database integrations as critical inputs for future bidding and measurement systems.

The message aligns closely with Google’s broader GML narrative: better data fuels better AI.

A new metric aims to measure what traditional attribution misses

Marvin also shed additional light on Qualified Future Conversions (QFC), one of the more intriguing measurement announcements from GML.

The metric is designed to estimate conversions that may occur up to 180 days after an ad interaction, helping advertisers understand the long-term impact of campaigns that don’t immediately generate revenue. The feature is particularly relevant for B2B and lead generation advertisers with lengthy sales cycles.

QFC is currently being tested with a limited group of advertisers, with broader availability expected later this year.

What excites Google most

When asked which GML announcements she was personally most excited about, Marvin pointed to three areas:

• New ad formats for AI Search.
• Measurement innovations such as Qualified Future Conversions.
• YouTube Creator Partnerships.

Her answer offers a useful glimpse into where Google appears to be investing most heavily: AI-driven discovery, more sophisticated measurement and creator-led advertising.

Why we care

The Q&A provided important context that was missing from the GML keynote presentations. Advertisers learned that broad match remains a viable route into AI Search, AI-specific reporting is still a work in progress, and Google’s long-term vision continues to revolve around automation powered by strong first-party data.

Perhaps most importantly, it showed that while Google is adding more AI to its products, it’s also building new controls like AI Brief to help advertisers influence how those systems behave — a balancing act that will likely define the next chapter of Google Ads.

Dig deeper

All Your Google Marketing Live 2026 Questions Answered with Ginny Marvin

]]> https://searchengineland.com/publishers-common-crawl-content-ai-training-479831 Wed, 10 Jun 2026 15:57:15 +0000 https://searchengineland.com/?p=479831 Could AI lose a key source of training data? Major publishers want Common Crawl to stop collecting and sharing their content.]]>

Digital Content Next (DCN) sent the Common Crawl Foundation a cease-and-desist letter demanding that it stop scraping and distributing protected publisher content.

The U.S. trade group, which represents major digital publishers (e.g., the AP, the New York Times, NBC Universal, Bloomberg, NPR, and Fox), also asked Common Crawl to remove DCN members’ content from its datasets, including paywalled and subscriber-only news articles.

Publishers question opt-outs. DCN’s lawyers raised concerns about whether Common Crawl honored publisher opt-out requests and removed older content when asked.

• The letter said Common Crawl had, in some cases, told publishers it was complying, only to later say technical costs and delays prevented full removal. DCN’s lawyers said they were reviewing whether those statements may have been inaccurate or misleading.
• Common Crawl publishes a registry of sites that have opted out of scraping. The list includes many large news publishers.

DCN alleges infringement. The letter argued that copyright law is not an opt-out system. DCN said Common Crawl “flagrantly infringed” publisher copyrights by creating and distributing datasets containing protected content without permission or compensation.

• The group also said Common Crawl made that content available to companies developing AI tools and large language models.
• DCN CEO Jason Kint said the legal notice challenges the idea that online content can be collected, stored, and reused simply because it is accessible.

Common Crawl pushes back. Executive Director Rich Skrenta denied that CCBot bypasses paywalls to scrape websites. He also denied misleading publishers after The Atlantic reported in November that some content from publishers that had requested removal remained available.

• “When a publisher asks us to remove previously crawled material, we respond promptly and initiate a removal process that reflects the technical design of our dataset,” Skrenta said.

Why we care. This fight could shape how much publisher content AI search engines can use without permission. If courts or settlements impose stricter consent requirements, AI responses may rely more on licensed sources and less on the open web.

AI training stakes. Since 2008, Common Crawl has scraped billions of webpages to build a free public archive. Its datasets have been widely used to train AI models. The New York Times’ 2023 copyright lawsuit against OpenAI cited Common Crawl as making up 60% of GPT-3’s training data, Press Gazette reported.

• A 2024 Mozilla Foundation paper said that, in its current form, generative AI likely would not have been possible without Common Crawl.
• Common Crawl has been working on open standards for AI crawling preferences, Skrenta said this week. DCN’s letter asks for a harder line: stop scraping protected publisher content and remove member content already in the datasets.

By now, you understand that LLMs are probabilistic systems and that AI answers are highly variable. That fact has convinced a lot of people that prompt tracking is extra noise. But discounting prompt tracking as nonsense is the wrong conclusion.

Even though prompt tracking is much less deterministic than keyword tracking, we can significantly increase the accuracy of tracking AI mentions and citations. Repeated runs, fixed sampling rules, and confidence intervals turn variance from a reason to quit into a number you can defend. 

By the end of this Memo, you’ll know how to build that system.

View embedded content

This memo assumes that you’re already:

• Operating under the philosophy of persona-based prompt design, argued for in Synthetic Personas for Better Prompt Tracking.
• Bought into doing AI SEO / AEO and need a measurement system that actually tracks your progress vs. noise. Check out How Much Can We Influence AI Responses to learn more.

The prompt-tracking backlash is only half-right

Prompt tracking critics are not wrong. Five people running the same prompt get five different answers. Within-LLM variance from sampling alone hits 10-34% on identical prompts.

Reporting a point estimate from one run is astrology. Together with AirOps, I looked at 815,000 prompt-page pairs and found that after running the same prompt 3x in ChatGPT, only 2.2% of citations remain.

Every prompt is n = 1. Given that the average prompt is 5x longer than classic search keywords, the chance that 2 people around the world use the same exact prompt is close to 0. We currently don’t have any insight into what users prompt, and we might never get that data (although both Bing and Google are keeping us satiated, for now, by offering some AI-visibility data).

But “probabilistic = unmeasurable” is lazy thinking. The weather is probabilistic. Credit scores are probabilistic. We still forecast and track them. 

Keyword tracking was never as clean as we’d like to remember

Classic keyword tracking was more deterministic, but not as much as you think:

• For local searches, results were personalized by location and device. 
• Google rescores results daily, so every rank tracker reports a position range, not a fixed number.

The industry standardized the sampling, fixed location, clean profile, daily crawl, etc., until the noise disappeared. Prompt tracking needs the same move, applied to a harder problem. An added challenge: Keyword tracking was focused on Google, but now we have tons of engines. As the market consolidates, tracking simplifies.

I’d argue there’s no escaping this either as Google transitions from classic search to AI search. More searches than ever show AI Overviews, all while AI Overviews and AI Mode increasingly merge.

At I/O 2026, Search head Liz Reid said users increasingly ask “longer, more natural-language questions,” and Sundar Pichai described Search as “less about individual queries” and “more like an ongoing conversation.”

Where common prompt tracking breaks

Over the last 2 years, prompt-tracking tools have multiplied, while the methodology behind them has stalled. Where’s the innovation?

The common prompt-tracking approach looks something like this:

• Define 25-50 prompts (brand/category/problem split).
• Run each prompt once per platform.
• Track daily.
• Score for citation, mention, sentiment, position.

Here are the problems I see with that approach:

• Variance: Only 2.3% of citations remain after three prompt runs [The Consensus Gap]. One run is a coin flip with the answer hidden. 
• Reasoning: High vs. low reasoning opens an 18 percentage point citation-rate gap and changes how the model searches, with high reasoning firing 4.6x more fan-out queries [Reasoning Lift]. An aggregate score blends two different engines into one misleading number. 
• Personalization: Most prompt-tracking is not persona-specific, so it reports generic answers that no one sees.
• Monthly cadence: SISTRIX tracked 82,619 prompts over 17 weeks and found Google AI Mode replaces 56% of its cited sources every week, while ChatGPT replaces 74%. At that drift, monthly tracking is like checking your bank account once a quarter.
• Cross-platform aggregation: Blending your ChatGPT + Perplexity + Gemini visibility into one “AI visibility score” is like averaging your Google rank with your Bing rank.
• Conversations: A single Turn 1 query tells you whether you get mentioned. It says nothing about whether you survive Turn 2 onward, when the user asks about alternatives, pricing, integrations, or risk. AI is a conversational interface, so the journey is the unit of measurement, and a one-shot prompt misses most of it. 
• Context: Pure mention counting with no context treats every appearance as a win. Get named first for “what are the worst CRMs to avoid?” and a mention tracker still records a victory. 

So, while we can’t remove AI answer variance, we can run prompts multiple times and measure what parts, brand mentions, and citations of the AI answer remain.

Mirroring follow-up prompts is hard because we don’t know exactly what people will ask, but we can use AI to estimate likely follow-ups, enrich them with real conversation transcripts, and track the follow-ups LLMs suggest inside their own answers. We can also record the attributes a brand gets mentioned with, not only whether it shows up.

What good prompt tracking looks like in practice

Worked example: B2B SaaS, CRM category.

• Prompt set: 40 seed prompts, weighted toward problem prompts where purchase intent lives (12 brand, 12 category, 16 problem).
• Platforms: ChatGPT, Perplexity, Gemini, Google AI Overviews. Tracked separately.
• Run config: Five reps per prompt per platform, every week.
• Personas: The 28 category and problem prompts are customized for three key personas (CFO, IT, marketing).
• Metrics: Mention rate (± CI), citation rate (± CI), average position when mentioned (1-5), sentiment, and the attributes attached to each mention.

Level it up by adding the journey layer. A flat list of 40 prompts only measures Turn 1. To measure conversations, build the high-intent prompts into journeys that follow the buyer across the five stages from Reasoning Lift: Problem, Exploration, Comparison, Validation, Selection. 

Each seed prompt for Turn 1 becomes the “seed prompt,” and each stage adds a natural follow-up prompt on subsequent turns.

For a buyer evaluating CRMs, one journey runs:

• Problem: “How do I know if my sales team needs a CRM?”
• Exploration: “What types of CRM software exist for B2B SaaS?”
• Comparison: “HubSpot vs. Salesforce vs. Pipedrive for a 50-person sales team”
• Validation: “Is HubSpot worth the price for mid-market B2B?”
• Selection: “How do I get started with HubSpot Sales Hub?”

Run the full sequence as one conversation rather than five isolated prompts, and score every turn. The payoff is persistence: in Reasoning Lift, a brand cited at the Problem stage carried all the way to Selection in four journeys under high reasoning and in zero under minimal. Persistence is the metric a one-shot tracker can never see.

Scope it so the run volume stays sane. Track all 40 seed prompts at Turn 1 for breadth, and build the 16 problem prompts into full five-stage journeys for depth.

Insight example: HubSpot is mentioned in 78% ± 6pp of fiproblem prompts on ChatGPT vs. 34% ± 9pp on Perplexity. Perplexity pulls from comparison posts (G2, Capterra); ChatGPT pulls from HubSpot’s own blog plus integration and compliance docs.

Action: invest in integration guides and API docs to win ChatGPT. Invest in G2 review velocity and comparison content to win Perplexity.

The next generation of tracking looks like polling

Prompt tracking won’t become keyword tracking. AI answers are too variable, too personalized, and too dependent on source selection. But that doesn’t make them unmeasurable.

The next iteration of prompt tracking will look less like rank tracking and more like polling: repeated runs, clear sampling rules, confidence intervals, segmented panels, and raw-answer audits.

This post first appeared on the author’s website and is republished here with permission. ]]> Where the prompt critique breaks down https://searchengineland.com/google-liability-false-ai-overview-claims-germany-479820 Wed, 10 Jun 2026 14:55:58 +0000 https://searchengineland.com/?p=479820 The injunction targets AI-generated accusations that tied two publishers to scams despite no support in the linked pages.]]>

Google can be directly liable for false claims in AI Overviews, a German court ruled. The Regional Court of Munich found that AI-generated summaries are Google’s own content and not protected as traditional search results are.

The court issued a temporary injunction barring Google from repeating false claims about two Munich publishers, according to The Decoder. This case centered around AI Overviews wrongly linking the publishers to scams, subscription traps, and questionable business practices.

AI Overviews aren‘t search results. AI Overviews do more than help users find third-party content, the court said. They rewrite, combine, and evaluate information “in its own words and according to its own structure.”

• In the disputed searches, Google’s AI Overview allegedly presented standalone claims about questionable business practices, along with warnings and red flags. The court found those claims didn’t appear in the linked sources.
• Because Google created the feature, controls its presentation, and controls the underlying algorithms, the court treated the statements as Google’s own content.

Search protections didn’t apply. Google argued that German case law limiting liability for traditional search engines and autocomplete should apply. Those rules generally treat search providers as indirect infringers when they surface third-party content.

• The court rejected that argument. Unlike traditional search results, which direct users to external pages, AI Overviews generate new substantive statements from multiple sources.
• The court also rejected Google’s claim that users could verify the information by reviewing the cited links. It found the AI Overview stood on its own and presented its claims as a complete answer.

Why we care. The ruling doesn’t treat AI Overviews like neutral search links. If an AI-generated summary makes false claims about a company, Google may be directly liable for those statements.

Wrong links, wrong claims. The court found that the AI Overview conflated information about other companies with that of the two publishers and created unsupported connections to the source material.

• Because the linked sites didn’t make the disputed claims, the publishers would have had no clear third party to sue if Google were treated only as a search intermediary.
• The court said Google can compare AI-generated statements with underlying sources, at least in cases like this.

Google must stop repeating the claims. The injunction bars Google from repeating most of the challenged claims, including allegations involving scams, subscription traps, questionable company ties, phone calls that allegedly never occurred, and lack of availability.

• Google must pay 80% of the legal costs. Each publisher must pay 10%.
• The court said the risk of repeat violations remains because Google did not provide a cease-and-desist declaration with a penalty clause, and the same algorithms could generate similar claims again.
• The ruling is temporary and may still be challenged. However, it gives publishers and brands a path to challenge false AI Overviews as Google’s own statements rather than merely search results.

SEO reports often contain solid research, including keyword data, technical findings, competitor insights, content gaps, and recommendations. The problem starts when stakeholders finish reading and still don’t know what should happen next.

For example, a report might say internal linking should be improved, but not which pages should be linked, who should make the changes, when the work should happen, or what results to expect. It might identify a crawl issue, but not explain whether fixing it matters more than the content or commercial page gaps already blocking growth.

This is where many SEO reports lose impact. The analysis may be correct, but the decisions and action plan are still unclear.

A strong SEO report should help readers understand what matters now, why it matters to the business, and what should happen next. It should reduce the need for another round of interpretation before work can begin.

Research is useful, but it’s not the final output

Keyword research, SERP analysis, technical crawls, competitor reviews, and content audits are all important SEO activities. They uncover gaps, risks, and opportunities that would otherwise remain hidden.

But those inputs shouldn’t dominate the final report.

Stakeholders don’t need to see every export, screenshot, or crawl detail. They need the conclusions that come from the research. They need to understand which findings matter, which improvements can wait, and which actions deserve priority.

For example, a crawl might identify 300 pages with missing meta descriptions. That finding is only useful if the report explains whether those pages matter. Missing descriptions on low-value archive pages may not deserve immediate attention. Missing descriptions on high-intent service pages with strong impressions may be worth fixing quickly.

The same applies to keyword gaps. A list of 200 missed keywords is less useful than identifying the five opportunities that align with commercial intent, existing authority, and realistic execution capacity.

A useful SEO report shows the work behind a recommendation only when it helps stakeholders understand why it deserves priority.

Dig deeper: Stop reporting traffic and activity. Start reporting progress.

Your customers search everywhere. Make sure your brand shows up.

The SEO toolkit you know, plus the AI visibility data you need.

Start Free Trial

Get started with

Where SEO reports lose stakeholders

Most SEO reports lose momentum when the findings are too generic, the recommendations aren’t actionable, or the priorities aren’t clear. The following are among the most common reasons reports fail to drive action.

The takeaway could apply to any website

Many SEO reports lose impact when the main takeaways are too broad.

Phrases like “improve content quality,” “strengthen internal linking,” and “target high-intent keywords” may be correct, but they don’t tell the reader anything specific about their site compared to others. These recommendations are too broad because they could apply to almost any website.

A stronger takeaway explains what’s different about the website, the market, or the opportunity.

For example:

• Weak: “Create more bottom-funnel content.”
• Strong: “Competitors are winning comparison and pricing-intent queries, while your site mostly ranks for educational searches. The next content priority should be three comparison pages that connect to existing high-traffic guides.”

The second version gives stakeholders a clearer reason to act. It explains the gap, the business relevance, and the first move.

The recommendation stops before the work begins

A recommendation can sound useful but still be difficult to execute.

“Improve internal linking to commercial pages” is a common example. It identifies the right area, but leaves too many open questions:

• Which pages should link?
• Which commercial pages matter most?
• Who updates the copy?
• When should it be done?
• How will impact be measured?

A more useful recommendation would look like this:

• “Week 1: Map the top 10 informational pages by organic clicks to the three commercial pages with the highest lead value.
• Week 2: Add contextual links using descriptive anchor text. SEO owns the mapping and QA. Content owns the copy updates. Review impact after four to six weeks through crawl depth, impressions, and ranking movement.”

That level of detail helps work begin without another strategy meeting.

Priorities are listed, not sequenced

Many reports list 15 or 20 recommendations at the same level of importance. This creates confusion because stakeholders can’t tell what should happen now, what comes next, and what can wait.

Priority should account for impact, effort, dependency, and timing.

For example, while fixing broken links might be easy, it may not be the highest-impact task. Consolidating cannibalized pages may take longer, but it could create more commercial value if those pages are competing for the same high-intent queries.

A useful report explains the order of operations. It shows which actions create momentum first, which actions depend on other teams, and which improvements should be treated as later-stage work.

Tool outputs are mistaken for strategy

Automated audits can surface useful issues, but they lack business context.

A tool can flag missing schema, duplicate titles, slow pages, and broken links. However, it can’t reliably determine which issue matters most for a specific business at a specific moment.

For example, a tool may flag site speed as a critical issue. But if competitors in the niche are equally slow and the bigger gap is weak commercial content coverage, site speed may not deserve top priority.

Tools identify potential issues. SEO judgment determines which issues deserve action.

Dig deeper: Why most SEO failures are organizational, not technical

Tailor reports to the stakeholder

A common reason SEO reports fail to drive action is that they treat every stakeholder the same.

The same finding may need to be explained differently depending on who has to act on it. A CEO, a marketing lead, a developer, and a content manager don’t need the same level of detail. They need the same truth translated into the context of their decisions.

For a CEO or founder, the report should focus on business opportunity, risk, resource needs, and expected impact.

• Example: “The site is losing visibility in commercial comparison searches where competitors are capturing high-intent demand. The top priority is building three comparison pages and linking them from existing high-traffic informational content. The pages can be published within four to six weeks, with early visibility signals reviewed after indexing.”

For a marketing lead, the report should connect SEO work to demand generation, campaigns, and content direction.

• Example: “The blog attracts early-stage visitors, but these pages don’t support product discovery. The next step is adding commercial pathways from the top informational pages into product, demo, or comparison content.”

For a developer or product team, the report should remove ambiguity from the technical requirement.

• Example: “Update canonical tags on these 12 filtered category URLs to point to the main category page. Acceptance check: The canonical target returns 200, is indexable, and appears consistently in rendered HTML.”

For a content team, the report should make page-level action clear.

• Example: “Update this guide to include a comparison section, add two internal links to commercial pages, and answer the pricing-related query that competitors cover but the page currently misses.”

A strong SEO report presents the findings in a way that each stakeholder can act on.

What a decision-ready SEO report should show

A useful SEO report should clearly answer a small number of questions. These questions may vary by stakeholder, but the underlying purpose stays the same: helping people decide what to do next.

Start with the opportunity

Where can SEO create business value? This could be a topic cluster where competitors are weak, a set of underperforming commercial pages, or an existing content asset that attracts demand but doesn’t support conversion.

Identify the main constraint

Is growth blocked by crawlability? Content depth? Competitor dominance? Something structural in how the site maps to buyer intent?

The answer shapes everything that follows.

For example, a B2B site may have strong informational rankings but weak visibility for comparison queries. In that case, the problem isn’t simply a content gap – it’s a missing bridge between educational demand and commercial intent.

Define the first move

The report should explain which action deserves priority and why it should happen before other work. This is where impact, effort, dependency, and timing matter.

For example, instead of saying “create a buyer guide,” the report should specify which guide to create, which SERP pattern justifies it, which existing pages should link to it, who owns the draft, when it should go live, and what signals should be reviewed after launch.

Establish how progress will be measured

Early signals may include indexing, crawl depth, impressions, or ranking movement. Later signals may include qualified traffic, demo requests, assisted conversions, or pipeline influence.

Dig deeper: SEO execution: Understanding goals, strategy, and planning

Turn every finding into a clear next step

Every important finding in an SEO report should answer three simple questions:

• What did we find?
• Why does it matter?
• What should happen next?

This helps prevent findings from sitting in the report as observations with no clear path forward.

Consider a report that identifies five high-traffic informational pages with no links to relevant commercial pages. That’s a useful finding, but without interpretation, stakeholders only know what exists. They don’t know why it matters or what action it should trigger. The site is already winning organic visibility, but the journey from education to action is weak.

The next step should be specific.

• Example: “Add contextual links from those five informational pages to the two most relevant commercial destinations. Content owns the updates. SEO reviews the mapping and anchor text. Measure changes in crawl depth, impressions, clicks, and assisted conversions after four to six weeks.”

This turns a finding into work that can be assigned, implemented, and reviewed.

If a finding can’t be connected to a next step, it may belong in a supporting document rather than the main report.

What to cut from SEO reports

SEO reports often become clearer when they remove material that doesn’t help the reader make a decision.

Tool screenshots are a common example. A screenshot can support a point when it shows a clear pattern, but adding screenshots for every issue usually makes the report harder to read.

The same applies to large keyword exports, crawl tables, and raw audit scores. These details are useful as supporting material, but they shouldn’t dominate the main report.

Generic best-practice advice should also be removed or rewritten. “Add schema,” “improve page speed,” and “optimize title tags” only help when the report explains why those actions matter for this site right now.

Long methodology sections can usually be shortened. Methodology matters when trust, reproducibility, or stakeholder education is needed. But most senior stakeholders don’t need to see the entire research process. They need the conclusion that came from it.

Before adding something to the main report, ask whether it helps the reader understand the priority or take the next step. If it doesn’t, it’s likely better placed in an appendix or supporting document.

Dig deeper: SEO in the black box era: Why reports will look more like Mad Men than Search Console

See the complete picture of your search visibility.

Track, optimize, and win in Google and AI search from one platform.

Start Free Trial

Get started with

The best SEO reports make the next step obvious

SEO reporting should reduce uncertainty. After reading a report, stakeholders should understand what matters most and what work should move forward.

That doesn’t mean every recommendation needs to come with a perfect forecast. SEO rarely works with that level of certainty. But each recommendation should explain the expected direction of impact and the signals that will be used to evaluate progress.

A useful report connects research to decisions. It shows the business context behind the finding, the rationale for prioritizing the recommendation, and the practical path to execution.

When reports stop at analysis, they create more work for the readers. When reports translate analysis into action, they help teams move faster.

The strongest SEO reports leave stakeholders with a clear priority, a practical next step, and a way to judge whether the work is moving in the right direction.

Most people aren’t using AI the way GEO discussions often assume. Two surveys of AI users conducted by Stella Rising found that many prompts still look remarkably similar to traditional search queries. (Disclosure: I’m the VP of SEO at Stella Rising.)

One survey focused on a beauty-oriented consumer panel in August 2025, while the other surveyed a broader general-audience population in January 2026. Across both studies, prompts were short, often keyword-driven, and much closer to a Google search than the elaborate prompt templates popular in AI marketing circles.

At the same time, a growing share of users are adding personal context, such as their budget, location, profession, age, health concerns, or preferences. Those details give AI systems far more information than a traditional search query ever could, creating a new layer of personalization that influences recommendations and brand visibility.

The combined findings suggest that GEO strategies need to account for both realities: Many AI searches still resemble classic keyword queries, while the highest-value recommendations increasingly emerge from prompts rich with personal context. That’s where the opportunity — and the measurement challenge — lies.

A lot of people are still typing like it’s 2008

The biggest takeaway across both surveys is that the median AI user is still throwing a keyword over the wall and hoping for the best.

In the general-audience study from January:

• Two-thirds of respondents reported writing prompts of 15 words or fewer. 
• Only 12% wrote something that would qualify as a “real” prompt by the standards of an AI influencer thread. 
• About 60% phrased their queries as questions, while only 9% gave a direct command. 

That mirrors what Pew Research has been seeing more broadly — 34% of all U.S. adults now use ChatGPT, roughly double the 2023 share, and 58% of adults under 30 use it.

When we ran a scenario task asking respondents to write the prompt they’d send if they needed a new pair of shoes, the median answer was eight words. Real examples from the panel included:

• “Shoes nearby”
• “Tennis shoes”
• “Nike”
• “Ladies tennis shoes size 7 near me”
• “Best price for hiking shoes”

This lines up with Semrush’s clickstream data on ChatGPT’s search mode, which shows the average prompt length is 4.2 to 8.7 words, essentially the same as a Google query.

Longer, structured prompts tend to appear only when users are doing something other than search, such as drafting, coding, or creative work.

For AEO and GEO work, that’s the part to internalize. If you’re optimizing for prompts like “Compare the top five orthopedic-approved walking shoes under $150 for plantar fasciitis with 4.5+ star ratings,” you’re optimizing for the wrong distribution. 

Real prompts run 71% longer than the synthetic ones marketers tend to invent, but the median is still only 12 words, Otterly.AI’s analysis found.

Your customers search everywhere. Make sure your brand shows up.

The SEO toolkit you know, plus the AI visibility data you need.

Start Free Trial

Get started with

The shift between the two surveys

In the August 2025 survey, we classified roughly 50% of the free-text prompts as “SEO-keyword-shaped,” meaning short, ambiguous, and brand-and-attribute-driven. By the time the January 2026 survey came back, that share had dropped closer to 30%. The remaining 70% had grown longer and more contextualized.

A few findings are worth carrying forward:

• 24.5% of all prompts include the word “best.” If you’re not appearing in “best [category]” responses, you’re missing one of the highest-intent slots.
• 28% of prompts mention price or budget constraints. Users aren’t just shopping. They’re shopping with a number in their head.
• 16% of prompts are explicitly location-based. The “near me” query pattern has successfully migrated from Google to LLMs.
• 32% of prompts include personal attributes (e.g., size, profession, health condition, life stage, etc.). This is the most important number on the page, and we’ll come back to it.

On location specifically, the 16% figure lines up with what Local Falcon’s 2025 research showed for AI search overall: AI Overviews now appear on 92% of informational local queries, but only on 15% of simple local-pack queries. The intent is moving into LLMs faster than the supply of optimized local content for AI engines.

One caveat: These were two different surveys with two different audiences. The January 2026 general-audience sample was structurally more transactional than the August 2025 beauty-focused panel, which partially explains why fewer prompts looked like keyword-style searches and more looked like full requests. I wouldn’t over-index on the “prompts are evolving” narrative, but I’d absolutely take the directional read.

The user embedding layer is where this gets interesting

The 32% figure (prompts containing real personal context) is the most under-discussed finding in the dataset.

Nearly one-third of users are willingly handing LLMs information that no Google query would normally carry, such as their size, job, training plan, living situation, or kids’ ages. We see prompts in the data like:

• “What shoes would you recommend for daily standing at work?”
• “Find me a cost-effective pair of running shoes that I can order on Amazon. My size is men’s 10.”
• “Please tell me the top five shoes for wide feet in a size eight for women that are comfortable, stylish, under $120, and that younger people won’t make fun of for a Gen X person like me.”

That last one alone packs in gender, foot width, size, budget, style intent, generational identity, and a real social anxiety. No traditional search query was ever going to surface all of that.

This is the user embedding layer at work. When someone interacts with ChatGPT or Gemini repeatedly, the model builds a profile of who they are, which increasingly persists through memory. The user is, in effect, training the assistant on themselves. Once that trust is established, they stop writing surface queries and start writing requests that assume the assistant knows them.

That shift has two implications for how brands should think about visibility:

• The prompts that drive purchase decisions are often not the ones that show up in a SERP or keyword tool. A real Gen X woman asking about wide-fit, $120, “won’t-get-made-fun-of” sneakers will never appear as a tracked SERP keyword. But that’s the prompt that decides whether your shoe ends up in the recommendation set.
• The value of a brand citation increases substantially when it appears inside a context-rich prompt. If the model is already factoring in user attributes, the brands it surfaces are prefiltered for relevance. That’s a much higher-quality impression than a generic blue link.

Where synthetic prompts fit — and where they don’t

A common tactic in GEO prompt research is to construct synthetic personas (“I’m a 38-year-old product manager training for a half marathon in Boston who prefers brands focused on sustainability…”) and then use those personas to stress-test which brands an LLM surfaces under different scenarios. There’s real merit to the approach. If the user embedding layer is doing the heavy lifting in the answer, the only way to simulate the answer is to simulate the user.

But synthetic prompts don’t capture everything. Real prompts are messy, layered, and influenced by recent conversation history, persistent memory, and signals the model has picked up over weeks of use. You can craft a 50-word persona and still miss the nuance of a user who has been talking to ChatGPT about their day, preferences, and family for 6 months.

Instead, use synthetic prompts to map the personas your brand needs to be visible to, but don’t treat the resulting visibility scores as ground truth. Combine them with real prompt data wherever possible. That can mean customer interviews, social search patterns, support tickets, or regex pulls of question-shaped queries from Google Search Console.

What to actually track

This naturally leads to the next question: Should you track SEO keywords in your AI visibility platform if one-third of real prompts look like SEO keywords?

The answer is yes, with one filter.

Across the last quarter, our team has seen web retrieval rates on tracked prompts climb sharply. On several client accounts, more than 90% of monitored prompts now trigger live web search inside ChatGPT or Google’s AI Mode. 

When that happens, the LLM is effectively running a real-time SERP and synthesizing the result. That means the short, keyword-shaped prompts we identified — roughly 30% of the total — are still very much in play. They behave like AI-flavored Google queries and should be tracked accordingly.

The filter is this: Don’t waste tracking slots on prompts that are pure head terms or single-brand queries. Those are likely to be answered from model weights or short canned responses rather than retrieval, and they won’t give you a useful read on visibility.

Here’s a practical setup:

• A synthetic-persona prompt set that exercises the user embedding layer, mapped to the personas your brand actually needs to win. Use this to surface which competitors a model defaults to under different user conditions.
• A real-prompt set sourced from question-shaped GSC queries, customer panel inputs, and regex-extracted “who/what/where/can/should” patterns. These are the short, retrieval-triggering prompts most users still write.
• A small qualitative library of messy, context-rich real prompts pulled from the type of work we did for the study. Use it to sanity-check whether your content actually answers the question the user is asking, not the question your keyword tool says they’re asking.

At that point, you’re not just tracking AI visibility. You’re tracking it across the full spectrum of how real users get to your content, from a three-word “good walking shoes” query to a 40-word “I’m a 60-year-old with plantar fasciitis…” request.

What the broader data tells us about AI search

A handful of additional findings from the January 2026 survey help explain why these prompt patterns matter.

Users increasingly trust AI recommendations

Up to 68% of users trust ChatGPT’s recommendations more than Google’s, with most citing detail, lack of ads, and personalization as the reasons. 

• NIM’s research has found ChatGPT often produces more efficient, accurate consumer decisions than Google.
• A January 2026 study reported that trust in Google weakens as AI usage rises. 
• Yext’s 2026 research puts ChatGPT’s share of “top-quality source” responses at 35% among heavy AI users.

AI search is becoming a daily habit

Half of active AI users use these tools daily or several times per day to complete tasks they used to do on Google.

• Search Engine Land reported 37% of consumers now start searches with an AI tool instead of Google.
• OpenAI’s February 2026 numbers put ChatGPT’s weekly active users at 900 million — more than double a year earlier.

Citations still drive traffic

85% of users click through to cited sources at least some of the time; 21.9% always do. The mention is not the end of the funnel. 

• Conductor’s 2026 benchmarks showed AI referral traffic up 357% year-over-year. 
• Semrush reported outbound referrals from ChatGPT up 206% in 2025. 
• Emarketed saw AI-referred visitors converting at 4.4x the rate of standard organic. 
• Volume is still small (Conductor pegs it at around 1.08% of total traffic), but it punches well above its weight class.

Voice may finally be having its moment

34% of users are now using voice chat daily or more often. This is the first dataset I’ve seen that actually delivers on the “voice search will matter” promise we’ve been hearing for a decade. 

It’s worth pairing all of this with Ahrefs’ latest AI Overviews CTR research: The presence of an AI Overview correlates with a 58% lower clickthrough rate for the top-ranking page. The traffic that does come through is qualified. The traffic that doesn’t is gone

AI search is settling into a richer, more personalized form. The intent stack is the same one Google has always served. What’s new is the embedding layer and the tracking demands it entails. That creates a clear set of priorities for SEO and GEO teams.

See the complete picture of your search visibility.

Track, optimize, and win in Google and AI search from one platform.

Start Free Trial

Get started with

What changes — and what doesn’t

Here are three things you can do with this information if you’re an SEO lead, content lead, or strategy lead:

• Audit your prompt-tracking setup: If it’s all synthetic prompts or all keyword-shaped prompts, you’re missing half the picture. Build the layered framework outlined above.
• Map your content to the user embedding layer: For your top categories, list the personas (e.g., age, life stage, profession, condition, budget) most likely to carry real prompts into AI search. Then check whether your PDPs, blog content, and FAQs actually answer those people’s questions.
• Don’t abandon the SEO-keyword work: Roughly one-third of real prompts still look like classic search queries. With web retrieval running at 90%+ on many of the prompts we monitor, the gap between an SEO keyword and an AI prompt is narrower than the GEO discourse implies.

The behavior change is real. The sophistication of AI thought leaders’ prompting is partly here and partly oversold. Most people are still doing Google-style searches. They’re just searching inside an interface that knows more about them.

If that’s where the audience is, that’s where we have to optimize.

Methodology

Both studies referenced in this article were conducted by the Stella Rising team. You can read it in “New Data: How Consumers Use LLMs for Search in 2026 (And What It Means for GEO).” 

The August 2025 study surveyed 178 members of Stella’s Glimmer Insights community, 113 of whom were active LLM users. 

The January 2026 study surveyed 524 active LLM users via Centiment, defined as having used ChatGPT, Copilot, or Gemini in the previous 30 days, with a margin of error of approximately ±4.3% at the 95% confidence level. 

Given its smaller size and category-specific composition, the August 2025 panel should be viewed as directional rather than statistically representative of the broader U.S. AI user population.

As algorithms take over ad platforms, truly expert marketers are being scrutinized — and rightly so.

Managing ad accounts looks very different today than it did even a few years ago. Platforms have shifted heavily toward automation, with products like Meta’s Advantage+ and Google’s Performance Max leveling parts of the playing field and changing what skilled account management actually looks like.

I’ve conducted many B2B account audits recently, both for companies with in-house teams and for businesses transitioning from other agencies. One thing keeps standing out: operators who truly understand these systems are still surprisingly rare.

Even with all the technology available today — and more on the way — many smart B2B marketers are still making expensive mistakes. Here are some of the most common ones I’ve encountered lately on LinkedIn and Google.

Mistakes advertisers are still making on LinkedIn

LinkedIn has become a crucial platform for reaching B2B audiences, yet advertisers keep making mistakes like these.

1. Using no audience targeting

Yes, I’m still encountering this rookie mistake. The impact is always what you’d expect: Ads are being shown to entry-level folks, college students, interns, and completely irrelevant companies instead of your target customer.

2. Letting audience targeting run without adjusting over time

LinkedIn’s professional targeting is unique, but it’s not airtight. For instance, it’s common to see sales professionals and titles sneak into campaigns with “growth” included in the targeting. 

Account managers who aren’t paying attention will continue to spend money to show ads to irrelevant audiences.

3. Using all automated default settings

I see this error from time to time, so it’s definitely not a one-off “oops.” 

LinkedIn is no different than other platforms. If you leave options like audience expansion turned on, your ads will show on cheaper, lower-quality inventory. Any desired actions will take the path of least resistance (usually the least qualified).

4. Setting and forgetting creative without rotating in new themes

Any experienced LinkedIn advertiser knows LinkedIn’s ad structure makes creative testing difficult, so you need to be proactive about refreshing creative.

Replacing tiny elements won’t move the needle, either, so failing to produce and use a bank of theme variations is a miss.

Dig deeper: 5 B2B LinkedIn Ads tests to run in 2026

Your customers search everywhere. Make sure your brand shows up.

The SEO toolkit you know, plus the AI visibility data you need.

Start Free Trial

Get started with

Mistakes advertisers still making on Google

Now we’re getting to the most common issues I’ve seen lately. Here are some of the mistakes our audits have uncovered in Google Ads accounts.

5. Running conversion-focused bid strategies without conversion data

I’ve audited accounts running ads with conversion-focused bid strategies, yet they have zero conversion data integration (or even any conversion actions specified). This mistake leads the algorithm to prioritize the absolute easiest actions and the cheapest audiences.

6. Leaving Display and Search Partners turned on

This oversight diverts budget from high-intent search inventory into reliably cheaper, lower-performing placements.

7. Forgetting to use sitelinks

By forgetting to use sitelinks, you inevitably cede a ton of real estate and limit ways to appeal to users. 

8. Letting spend run on non-converting keywords 

If you’ve zoned out and are continuing to spend money on keywords that aren’t converting for months on end without making any adjustments to the ads themselves, you’re flushing money down the toilet.

9. Running all campaigns in broad match only 

Similarly, you’re wasting money if you’re running all of your campaigns in broad match only, with no phrase or exact match to be found.

10. Letting PMax for B2B run with no guardrails beyond remarketing lists

PMax for B2B has gotten a little more viable over the past year, thanks to a slew of moderately useful new functions. 

However, if you’re not even doing the basics of enhanced conversions and OCT, I would turn those campaigns off ASAP or hand them over to marketers who can use PMax to improve lead quality. 

Otherwise, this mistake means you’re setting yourself up to spend on junk leads. 

11. Not testing AI Max 

Google is rapidly changing how ads are matched, served, and optimized. If you don’t test these systems early, you risk falling behind competitors who are giving Google more automation signals and learning data.

Success with AI-driven campaign types depends on proper measurement, quality inputs, and ongoing oversight. Those responsibilities haven’t gone away as automation has increased.

Dig deeper: Top 10 Google Ads mistakes to avoid in 2026

Automation still needs expert oversight

An important layer across channels that often gets overlooked is what happens after someone provides their information. Having a strong follow-up sequence for leads is critical to campaign success, as is passing back the right data to properly tie performance to ad spend. 

Omitting this crucial step does more than cloud the true potential ROAS of the campaigns. It also means you’re getting less down-funnel data that could be fed back into the platforms to teach the algorithms to work harder on your behalf.

Even experienced marketers are making these mistakes. They tend to be good at what used to be critical, such as SKAGs and exhaustive negative keyword management, but they’re not adapting to use the controls still available to them in a more automated landscape.

These mistakes aren’t just one-off “oopsies” that cost brands money before they’re caught and corrected. To me, they’re flashing red lights signifying that the account isn’t in good hands, especially if you’ve got more than one of these scenarios in play, or if one has persisted over time.

Yes, account management involves much more automation than it did in the past. But these platforms still need quality inputs, proper measurement, and ongoing oversight to perform at their best.

Keep that in mind before you decide that paring back to a B-level resource and AI can help save you money in the fixed-cost column.

Google searches ended without a click 68.01% of the time in the U.S. during the first four months of 2026, according to new SparkToro research based on Similarweb clickstream data. That’s up from 60.45% in 2024, a 7.56-point increase in two years.

Fewer searches result in clicks. The share of searches generating at least one click fell 9.51 percentage points between 2024 and 2026 (a 22.9% decline), according to SparkToro. This includes clicks to organic results, paid ads, and Google-owned properties such as Maps and YouTube, but excludes follow-up searches within Google.

• Over the same period, the share of searches that led to another Google search rose 7.2 percentage points.
• This trend reflects Google’s growing ability to answer questions directly in search results while encouraging users to refine or continue their searches within Google, according to SparkToro.

AI Overviews and zero click. SparkToro believes AI Overviews are likely contributing to the increase in zero-click searches, though the study doesn’t isolate the extent to which the overall rise between 2024 and 2026 can be attributed specifically to AI Overviews.

• AI Overviews now appear on more than 20% of Google searches, according to the research. When they do, click-through rates drop by nearly 60%.

AI Mode and zero click. It appears to have played only a limited role during the January to April study period. SparkToro found that just 0.34% of searches transitioned into AI Mode during that time.

• However, Google said at I/O 2026 that AI Mode had surpassed 1 billion monthly users and that query volume was more than doubling each quarter, suggesting its impact on search behavior could grow significantly.

Zero click history. SparkToro has tracked zero-click search behavior for years, though its underlying data sources have changed over time. Because the studies rely on different providers, panels, and methodologies, long-term comparisons are not directly equivalent. Still, the available data consistently points to a rise in zero-click behavior over time, according to SparkToro.

• 2019: 49% of Google searches ended without a click, based on the now-defunct Jumpshot clickstream panel. Later that year in updated findings, the figure rose to 50.33%.
• 2020: 64.82% of Google searches ended without a click between January and December, based on SimilarWeb data.
• 2024: 58.5% of Google searches in the U.S. (and 59.7% in the EU) ended with no clicks, based on Datos data.

Why we care. The findings suggest Google is increasingly satisfying user needs without sending users to external websites. However, you should interpret direct comparisons across years cautiously because SparkToro’s historical analyses rely on different clickstream data providers and panels.

SEO still matters, but… SEO alone may be insufficient for many publishers seeking to regain historical levels of Google-referred traffic. SparkToro co-founder Rand Fishkin recommended investing in brand awareness and influence on the platforms where your audience already spends time, regardless of whether those efforts drive direct website visits.

• Some categories continue to benefit significantly from SEO, including branded searches, local business queries, and high-intent transactional searches, Fishkin said.

About the data. The study used Similarweb desktop and mobile web panel data covering U.S. Google searches from January through April 2026. SparkToro assumed that two-thirds of searches occurred on mobile devices and one-third on desktops. The analysis excludes searches conducted in Google’s mobile search app, where SparkToro said zero-click behavior may be even higher.

The study. In 2026, Less than One Third of Google Searches Still Send a Click ]]> https://searchengineland.com/how-ai-forms-opinions-about-your-brand-479671 Tue, 09 Jun 2026 14:00:00 +0000 https://searchengineland.com/?p=479671 Build a digital footprint that helps AI understand your expertise, recognize your credibility, and recommend your brand.]]>

AI forms opinions about your brand from what it can see online. That’s your digital footprint.

The problem is that AI often sees only fragments of your business. It sees your website, content, reviews, and mentions, but much of the expertise, customer insight, and operational knowledge that makes your business valuable never makes it into the digital footprint.

The solution is to surface that knowledge, organize it into a single source of truth, and turn it into machine-readable signals. Here’s how to collect it, organize it into a single source of truth, and distribute it across the channels AI uses to understand, evaluate, and recommend brands.

What you feed the machines is understandability, credibility, and deliverability (UCD)

Everything you put into your footprint is fodder for three things AI has to decide about you. Together, they provide the fodder for the whole funnel.

Understandability

Does AI know who you are, what you do, and who you serve? You already know where your understandability comes from: 

• Your about page.
• Your product pages.
• Your structured data. 

What often gets missed is the operational detail that explains what you actually do once a client is inside.

Credibility

Does AI believe you’re good at it? This is N-E-E-A-T-T credibility — notability, experience, expertise, authoritativeness, trustworthiness, and transparency, an extension of Google’s E-E-A-T.

You know what credibility signals you currently feed: your case studies, your credentials, and your testimonials. What many businesses don’t realize is how much N-E-E-A-T-T credibility is already embedded in their day-to-day operations.

Deliverability

Does the AI engine have the content to hand you to the subset of its users who are your audience? 

You know where your deliverability comes from: the topical content, the marketing, and the authority pieces you commission. Deliverability is often hiding in plain sight, in the content generated by your business operations and offline activities.

Your customers search everywhere. Make sure your brand shows up.

The SEO toolkit you know, plus the AI visibility data you need.

Start Free Trial

Get started with

5 streams of business data feeding every commercial surface

All three elements of the UCD trio are fed by the five inputs below, and how much each contributes varies by business.

The point isn’t to file each input under one letter. Organized and codified, the five together give AI the fodder it needs from top to bottom of the funnel.

1. Products and services: What you sell, and you already do it

Your products and services data: what you sell, at what price, under what conditions, and with consistent names and identifiers. This is mostly about understandability, with credibility riding alongside it.

Most businesses already do this, so the work is in the depth, not the effort. Don’t just list what you sell. Describe who each offering is for, what problem it solves, what it costs, what it doesn’t do, and how it differs from the next option.

A thin product page tells AI a product exists. An exhaustive one tells it when to recommend that product and to whom.

Keep it accurate, complete, and consistent with everything else in your footprint. A price or product name that differs across pages reads as doubt.

2. Authority content: Your expertise, and almost everybody does it

This is the marketing you already create to show you know your field: your articles, videos, guides, data studies, and the thought leadership you publish to tick the box marked “content created.”

People put effort into it to build authority, rank, do SEO, and position themselves as experts. That’s fine. It leans toward deliverability because it’s what tells AI which territory to surface you in.

But everybody does it, which is exactly why it’s the least differentiating of the five on its own. It earns its weight only when it’s tied to the rest: the same expertise proven by your operations and corroborated by third parties, not just asserted in a blog post.

It’s necessary, but it’s not where your advantage hides.

3. Brand narrative and voice: Who you are, who you serve, and why you’re the best

All marketers create brand narratives, so the work here is about consistency and clarity rather than invention. Everybody communicates who they are, what they do, and who they serve, and keeping that clear and consistent matters enormously. 

But three things are often left out, and AI needs all of them.

• Intent: It isn’t enough to name your ideal customer profile (ICP). You have to pair your ICP with what they’re after: the cohort-to-intent combinations from the funnel query pathway. AI has to know not just whose problem you solve, but which problem, and at which moment, before it can hand you to them.
• Credibility: The thing that feeds your N-E-E-A-T-T. Many people leave it out because they feel awkward saying it. You have to set it out because AI won’t work out your true value on its own. Be clear and bold about why you’re credible, then make sure you can back it up with evidence.
• Making the relationship with your clients explicit: Validation from the people you serve that you deliver on what your narrative and cohort-to-intent mapping promise. Say who you are, what you do, and who you serve. Then explain why a customer should choose you and prove it.

Voice is the part corporations get wrong most often. Narrative is what you say. Voice is how you say it. One team may write the narrative once, but voice escapes through every rep, every support reply, every social post, and every deck. 

When it drifts, and in most large companies it drifts constantly, AI reads the same brand as five different brands and loses confidence in all five.

So standardize your voice and keep it consistent everywhere. Consistency is a credibility signal in itself. Inconsistency is a tax you pay without seeing the bill.

In short, make sure your brand narrative clearly sets out your ICP, who you are, and why you’re the best fit for them, in a voice that stays consistent wherever AI finds it.

4. OPID business operations: The stream almost nobody harvests

This is everything your business generates by running: onboarding, performance, integration, devotion, and all the day-to-day activity around them. 

It’s the most powerful of the five because the material comes from your clients and from the work your team does to serve them, which is exactly the material that rarely makes it online. It sits behind closed doors, buried in a CRM, parked on a platform nobody values, and almost nobody harvests it.

It feeds all three elements of understandability, credibility, and deliverability more effectively than anything else you own. 

• Understandability comes from the granular detail of what you actually do and the exact circumstances in which you help. Most of that is only ever discussed inside the business. A review where a client describes precisely what they got from you puts something on the record you’d never say about yourself, and the machine reads it as fact.
• Credibility is your N-E-E-A-T-T, and this is the most convincing kind because it comes from clients themselves, not from your marketing.
• Deliverability comes from the match. The content here aligns exactly with your cohort-to-intent combinations because it was created around the clients you attracted and served well. Whether it comes from you or from them, it fits the audience and intent you need to communicate to the engines.

Once you start looking, you’ll find the richest material you own:

• Customer voice is the highest signal because it’s real questions in real language: reviews across every platform, written and video testimonials, FAQs, unpublished support questions that should become FAQs, support and sales call transcripts, onboarding and churn-exit interviews, and free-text survey responses.
• Evidence and outcomes provide the proof you need: case studies with real before-and-after numbers, patent filings, academic deposits that are public but underused, and independent third-party studies that corroborate your claims.
• Methodology covers the rest. SOPs, playbooks, training materials, glossaries you currently keep private, and long-form spoken content such as webinars, keynotes, and podcast appearances, transcribed.

Look for material that answers a question an assistive engine or agent actually gets asked, in the questioner’s own words, with a verifiable fact attached. 

A support ticket, churn interview, or sales call transcript will often outperform polished marketing copy in that test because it’s already phrased the way real people ask questions.

That’s the whole point of harvesting OPID business operations: taking information from a place AI can’t see and moving it to a place where it can, while making it visible to your human audience, too. It’s convincing to both because it’s true and because it matches the cohort-to-intent combination exactly.

5. Bringing the offline online: The stream almost nobody runs

This section is all about the marketing and audience engagement you do offline: the talks you give, the festivals or hackathons you sponsor to support your community, the interviews, the panels, and the rooms full of clients. It’s obvious to you, but largely invisible to AI.

Bring the offline online and feed it to the machines by publishing self-reporting content and linking to the social posts and summary articles others write. That’s a huge win most brands miss.

But it works the other way, too. Your codified source of truth can feed your offline communication, so the story a client hears from you at a conference, in a newspaper, on the radio, or face to face is consistent with the story you’re telling AI on the web.

That matters more than it seems. If the two differ, you lose the person because the gap reads as doubt to a human and as low confidence to a machine.

Clarity and consistency over time, online and offline, is the name of the game.

Organize and codify the five into one source of truth

Once you’ve harvested all five streams, organize and codify them into a single source of truth: a database you build to output whatever format each surface needs, including HTML, schema, MCP, RDF, prose, audio, video, and images.

Organize the data once, centralize it, set up a system that codifies it on the way out, and from there you can distribute it in a few clicks while your digital footprint stays clear and consistent as it grows.

Then distribute it across your digital ecosystem in the format your human audience expects and packaged so machines can ingest it cleanly.

Where you publish affects how much the machine believes you, and the rule is simple: the less of you there is in it, the more it trusts it. You’re working across three tiers.

First-party: You claim 

You publish on your own properties, in your own voice. You state who you are and set the frame. It’s the baseline, and on its own it proves nothing because you wrote it and you published it.

Second-party: You corroborate

Here, you’re still publishing, but across a broader footprint and with other voices in the mix. Two things widen here.

• The platform: In addition to your own entity home website, you publish on platforms where you own the account, such as YouTube, LinkedIn, Medium, and press releases. You’re stating your case the same way you would on your website, just on another property you control.
• The voice: You can publish your own words, or you can publish what a client or user said, such as a review, quote, or case study, on your own site and across those other accounts.

It’s a step up from first-party because the substance is no longer solely your own assertion, even though you’re still the one choosing it and publishing it.

Third-party: They prove you

A third party publishes in its own voice, on its own site or social accounts, or on a neutral platform such as Trustpilot, with no involvement from you. 

Think clients and partners sharing their experiences, journalists, analysts, academics, and the long tail of user-generated content that assistive engines lean on.

It’s the strongest evidence because you had no hand in creating it.

You can’t write that third tier, but you can feed it. Your clients publish because you’ve served them well enough that they want to, so earn it.

Independent publishers can’t see inside your business, so give them something to work with: a client story they can build on, a view into your operation, or data about your business and industry they can cite.

Giving outside parties a true, detailed version of your business to publish is what PR, marketing, and content teams have always done. The only thing that’s changed is that now you do it so machines read the result as proof, not just so humans read it as coverage.

Point all three tiers at the same picture — you, your audience, and the independents — and they align into one answer the machine can’t miss.

Read the grid by how much of you is in the publication.

• First-party is all you. Your words on your own site. It’s pure claim, and the machine treats it as the baseline because you wrote it and you published it.
• Third-party is none of you. Someone else’s words on a platform you don’t control. That’s why it’s the strongest proof.
• Everything in between is second-party corroboration. Your own words carried onto an account you run elsewhere, or someone else’s words that you chose to publish on your own page.

The same review is second-party when you surface it on your site and third-party when the client publishes it on their own account. The words are identical. The weight is different. The difference is determined entirely by who publishes it.

Step back, and you have a powerful loop: You harvest your operations, codify them into a single source of truth, and distribute them across the tiers machines read. Then the machines recommend you, your ICP arrives, and serving them generates the next round of operations to harvest.

Each turn feeds the next, so your digital footprint compounds instead of resetting.

The mirror principle is why this is the whole game

At the moment it recommends a brand, think of the AI engine as an honest broker — the impartial intermediary of business and diplomacy, the one with no stake in who wins, whose entire value to the person asking is that it can’t be bought.

Much as a travel agent carries every airline or a mortgage broker has the whole market on screen, the honest broker carries every brand in your category and recommends whoever it judges to be the best solution for the person in front of it. That impartiality is the entire reason the buyer trusts it, and it’s the reason the engine recommends your competitor without a flicker of disloyalty: it was never on your side, it’s on the buyer’s.

This is good news once you see it the right way. An honest broker wants to recommend you. It can only recommend what it clearly understands and trusts. So you don’t have to trick a rigged system. You have to give the broker the organized facts: hand it the most detailed, best-corroborated picture of who you are, what you do, who you serve, and why you’re the best fit for them, better and more often than your competitors manage.

Brief it better than they do, build a clearer and more convincing case, and on merit, you become the name it reaches for at the top, middle, and bottom of the funnel. The brands losing today mostly aren’t outbid. They’re out-briefed because the only picture the broker ever had of them was thin.

The briefing comes from your digital footprint. The broker forms its view of you from the world’s view of you: the reviews, the coverage, the corroboration scattered across the market, and what it shows about you is its opinion of the world’s opinion of you. That’s the mirror principle.

You can try to flatter the broker, trick it, lean on it, and it might even work for a while, but an honest broker reads the world, so the thing that holds is changing what the world can see. Do that, and you’re not manipulating anything. You’re showing the broker proof: something that was always true, just underrepresented, or simply never visible.

Briefing the honest broker is the whole job. Harvest the five streams, organize and codify them into one source of truth, distribute them across the tiers the broker reads, and you’ve handed it the fullest, truest, best-corroborated picture of you in the market, at the one moment that pays: when someone is in the market for what you sell and the broker is about to be asked who it recommends.

Do that consistently enough, across everything the broker reads, and you stop merely briefing it and start training it. That’s the finale: how you train the honest broker rather than just brief it, what an under-briefed broker costs you, and when paid placement pushes you versus when it backfires.

For now, hold onto one idea: the broker recommends based on what the world shows it, so change what the world shows it. AI is laying foundations now that will hold for years, and the incumbent wins, so don’t leave it until tomorrow.

This is the 17th piece in my AI authority series.

• Part 1, “Rand Fishkin proved AI recommendations are inconsistent, here’s why and how to fix it,” introduced cascading confidence.
• Part 2, “AAO: Why assistive agent optimization is the next evolution of SEO,” named the discipline.
• Part 3, “The AI engine pipeline: 10 gates that decide whether you win the recommendation,” mapped the full pipeline.
• Part 4, “The five infrastructure gates behind crawl, render, and index,” walked through the infrastructure phase.
• Part 5, “5 competitive gates hidden inside ‘rank and display’,” covered the competitive phase.
• Part 6, “The entity home: The page that shapes how search, AI, and users see your brand,” mapped the raw material.
• Part 7, “The push layer returns: Why ‘publish and wait’ is half a strategy,” extended the entry model.
• Part 8, “How AI decides what your content means and why it gets you wrong,” covered annotation, the last gate where you’re alone with the machine.
• Part 9, “Why topical authority isn’t enough for AI search,” opened the competitive phase proper with topical ownership.
• Part 10, “The funnel flip: Why AI forces a bottom-up acquisition strategy,” named the process.
• Part 11, “The framing gap: Why AI can’t position your brand,” exposed the gap between evidence and recommendation.
• Part 12, “The 10-gate AI search pipeline: Find where your content fails,” showed you how to find (and repair) your F grades in the AI engine pipeline.
• Part 13, “The delegation boundary: How AI decides which brands win,” mapped how delegation moves between user and engine across Search, Assistive, and Agent modes.
• Part 14, “The funnel query pathway: A framework for measuring AI visibility,” built the measurement instrument.
• Part 15, “The micro-macro shift: How to measure AI visibility now that precision is gone,” moved measurement from micro precision to macro trend.
• Part 16, “How SEO turns customer success into AI-readable proof,” put SEO inside post-sale operations.
• Up next: Why Google isn’t dying: paid and organic just collapsed onto every AI surface.