Visibility #26: Google Claims Billions of AI Clicks

Visibility Report #26 | July 21, 2026

What matters: Google is putting a large number on AI search traffic, but the measurement layer still does not let publishers verify it.

Operator read: track citations, prompts, paid placements, and destination URLs separately before trusting a blended AI visibility score.

Google says its AI search features send billions of clicks to websites each week. That sounds encouraging, but publishers still cannot isolate those clicks in Search Console or evaluate Google's claim independently. Now we have to win twice: earn the citation, then determine whether it produced traffic anyone can measure.

Google Claims AI Search Sends Billions of Clicks

Google says AI Overviews and AI Mode collectively send billions of clicks to websites every week. The company also says AI-driven clicks are higher quality, but it has not released the underlying data, defined the comparison group, or provided reporting that lets site owners verify either claim. For publishers deciding how much to invest in AI visibility, that missing measurement layer matters as much as the headline number.

So treat Google's figure as a directional signal, not a benchmark. Compare AI-feature exposure, cited URLs, assisted conversions, and referral traffic rather than waiting for a single clean metric. Analytics aren't truth. They're opinions with decimal points. (Because one Google number without a denominator apparently wasn't enough.)

📎 Search Engine Land | Search Engine Journal​

Top Stories Are Appearing Inside Google AI Overviews

Google is rolling out Top Stories modules within AI Overviews, bringing fresh news results directly into generated answers. News publishers may gain another citation surface, but the module can also satisfy more of the query before a reader visits the source. Speed, clear publication dates, original reporting, and well-labeled updates probably matter more when an AI answer is assembling a current-events response.

Operators should monitor news queries separately from evergreen prompts and record which URLs enter both Top Stories and the surrounding AI Overview. What remains unclear is whether inclusion produces incremental traffic or redistributes clicks that previously went to the standard news block.

📎 Search Engine Land​

Semrush Adds a Workflow for Measuring AI Share of Voice

Semrush published a workflow for measuring how often a brand appears across tracked AI responses relative to competitors. Share of voice can help establish direction, but the result depends on the prompt set, engine, location, model version, and collection schedule. Two teams using different prompt portfolios can produce different answers for the same brand.

Use the metric as a windsock, not a GPS. Save the underlying responses, separate mentions from citations, and annotate major prompt or model changes so a dashboard movement does not get mistaken for market movement.

📎 Semrush​

seoClarity Introduces LiveWire for SEO and AEO Intelligence

seoClarity describes LiveWire as an intelligence layer that supplies search and answer-engine data to AI tools. The pitch addresses a real operating problem: general-purpose agents can generate recommendations without current ranking, crawl, citation, or competitive evidence. Connecting those systems could make analysis more useful, although the supplied story does not establish how LiveWire performs against existing exports, APIs, or analyst-led workflows.

Teams evaluating it should use a fixed set of research tasks and score source traceability, freshness, accuracy, and time saved. An answer that sounds polished but cannot show its evidence is still a liability.

📎 seoClarity​

Duane Forrester Opens an Anonymous GEO Tools Survey

Duane Forrester is asking practitioners which GEO and AI visibility platforms they use, trust, and think are missing something. The anonymous survey contains 17 questions, takes less than five minutes, and remains open through July 27. It does not collect email addresses, and Forrester is aiming for 1,000 responses before publishing results in August.

The category is crowded and vendor-claim heavy, so independent practitioner data could add useful signal. If you have used these platforms, this is a chance to contribute before the market gets another comparison based mostly on feature pages.

📎 Duane Forrester Decodes | Take the survey​

More From This Week

Search & AI Visibility

• ​How to Measure AI Search Visibility (Search Engine Journal) - A measurement framework covering prompts, mentions, citations, sentiment, and competitor visibility.
• ​NotebookLM Changes May Expose Sites to More AI Scraping (Search Engine Journal) - A look at how Google's evolving research products may change content access and publisher control.
• ​Google Quantifies AI Search Clicks Without Publishing the Data (Search Engine Journal) - The company supplied a large aggregate claim but no reporting publishers can use to validate it.
• ​Top Stories Roll Out in Google AI Overviews (Search Engine Land) - Fresh news results now have another route into Google's generated answers.
• ​Google Says AI Search Sends Billions of Weekly Clicks (Search Engine Land) - The scale is notable, but site-level AI feature reporting is still missing.

From the Tool Blogs

• ​Six Ways to Automate AEO With Letaido (Ahrefs) - A vendor-specific look at automating prompt monitoring and answer-engine optimization tasks.
• ​LiveWire Adds SEO and AEO Intelligence to AI Tools (seoClarity) - A data layer intended to ground AI workflows in current search evidence.
• ​How to Prove AI Search With Real Tests (seoClarity) - A webinar on designing tests that connect AI visibility activity to observable outcomes.
• ​How to Measure AI Share of Voice (Semrush) - A workflow for comparing brand visibility across selected prompts and competitors.
• ​Does AI Keep Office Hours? (Profound) - Research into whether answer behavior changes by time of day, with implications for sampling schedules.
• ​Profound Finds Different Google AI Products Give Different Answers (Profound) - July 14 research showing Gemini, AI Overviews, and AI Mode should be measured as separate visibility surfaces.
• ​Profound Launches FactCheck for AI Accuracy Monitoring (Profound) - July 14 product update for checking whether AI-generated brand claims are accurate against a source of truth.
• ​Peec AI Adds Advertising Visibility Reports (Peec AI) - July 16 product update that shows ads appearing in ChatGPT responses for tracked prompts.

Agency & Practitioner Insights

• ​How Vanta's SEO Moat Translates Into AI Visibility (Foundation Inc) - A teardown connecting established search demand, authority, and content distribution with AI discovery.
• ​Is Your University Website Ready for AI Search? (Search Influence) - A readiness framework for higher-education sites competing in generated research and recommendation journeys.
• ​Why Landing Pages Matter More in AI-Mediated Search (Search Influence) - Landing-page clarity affects paid performance, organic relevance, and what machines can extract.
• ​SEO-Optimized Landing Pages and AI Ad Performance (Search Influence) - An argument for treating the destination page as part of the automated advertising system.
• ​AI Search in Higher Education (Search Influence) - Practitioner observations from a Harvard professional learning session.

This newsletter is produced collaboratively by Will Scott and Bob, an AI agent. Human oversight, AI efficiency.

So What Do You Do About It?

Create a fixed audit set of 20 high-value prompts this week. Record the engine, surface, response, citations, paid placements, destination URLs, and whether the answer states your core facts correctly. Run the same set again next week; that repeatable evidence will tell you more than a single blended visibility score.