AI Overviews Now Appear on 83% of Branded Searches

The Visibility Report #37 | Week of September 29-October 5, 2026

Ahrefs found AI Overviews on 82.91% of the US desktop branded search results it checked on September 29. A search for your own name can now put an AI-generated description and third-party sources ahead of the website you control.

Google also told publishers to manually fact-check AI-generated copy and metadata, while Shopify made eligible product catalogs available through AI shopping channels by default. Brand visibility, content review, and commerce are all moving into systems where the website is only one source among many.

AI Overviews have moved into branded search

Ahrefs analyzed 232.8 million US desktop Google result crawls from July through September. The dataset included 92.5 million branded-keyword crawls and 140.2 million non-branded crawls. On September 29, AI Overviews appeared on 82.91% of the branded results. The September monthly average was lower at 73.3%, up from 61.3% in July.

A separate Ahrefs panel followed exact-name searches for 100 major brands. On September 30, 63 usable brand-name results showed an AI Overview, and 48 of those appeared in the first position. Wikipedia, YouTube, and LinkedIn were the most common outside domains cited in that smaller panel.

Those samples cover US desktop results and Ahrefs' own query sets. Google can change triggers and placement quickly, so 83% is a measured point in time rather than a universal rate. Still, branded search can no longer be treated as protected territory. Review the answer, the description of the company, the cited sources, and the pages still visible after the AI result.

Source: Ahrefs study​

Google now calls manual fact-checking of AI content critical

Google updated its guidance for sites using generative AI on October 1. The document says AI output may contain inaccuracies and that publishers should manually fact-check and review all AI-generated content before publication. The same review applies to title elements, meta descriptions, structured data, and image alt text.

Google separately added a warning against deceptive authorship information. Its people-first content guidance now names AI-generated headshots, invented names, and false credentials used to imply human expertise. The document calls those practices deceptive and a signal of a low-quality page.

This is guidance, not a new ranking-system announcement. The operating change is clear enough: route machine-written body copy and metadata through the same factual review, and audit author profiles for claims that a reader cannot verify. A clean article paired with an invented expert still creates a trust problem.

Sources: Google's generative AI content guidance | Google's people-first content guidance | Search Engine Journal​

Shopify can place eligible products inside AI shopping channels by default

Shopify says its agentic storefronts are active by default for eligible stores. Products can appear in ChatGPT, Google AI Mode and Gemini, Microsoft Copilot, Meta, and other connected channels through Shopify Catalog or Google Merchant Center. Merchants can manage the channels under Sales channels > Agentic.

Direct checkout is available for qualifying stores on some channels. When a buyer completes the purchase inside an AI app, ordinary browser-based analytics may not fire. Shopify still records the order in its admin, so channel-level order data matters more than a referral-session report for those transactions.

Default enrollment is distribution, not demand. Check which channels are enabled, confirm that product titles and attributes are accurate, and compare Shopify orders with analytics before judging the program. A zero in web analytics does not prove that an in-app checkout produced no sale.

Sources: Shopify Help Center | Slobodan Manic in Search Engine Journal​

One visibility score hid six different versions of ClickUp

Foundation ran 21 prompts twice across ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, and Google AI Mode, producing 252 answers about ClickUp. The brand appeared in 117 of 120 unbranded discovery answers, a 97.5% presence rate. That strong average hid large differences in position, category language, recommendation, and cited sources.

For one buyer question comparing ClickUp, Asana, and monday.com for a 100-person company, ClickUp received none of the 12 default recommendations. It also lost all 12 user-experience comparisons, even though five of the six surfaces described it positively on average. Other prompts about software consolidation and project execution produced much stronger results.

This is an agency case study about one brand, not a market benchmark. It shows why a mention-rate average needs a diagnostic layer. Track whether the brand appeared, where it appeared, how it was described, which product was recommended, and what evidence was visible. Then send the problem to the right team. A false product concern needs different work from a weak category association.

Source: Foundation ClickUp model-drift study​

From Practitioners

• ​Search Influence: Paula French's senior living playbook connects AI-search eligibility with consistent care-level information across the website, directories, reviews, and Google Business Profile. Published October 2.
• ​Lily Ray: Branded AI Overviews can give third-party sites prominent impressions for searches that traditionally belonged to the brand. Her post is interpretation of the Ahrefs/DemandSphere trend, not separate factual proof.
• ​Mark Traphagen: Prompt tracking should be treated as sampling. Stable topic sets and repeated observations are more useful than reading one response as a ranking.
• ​Slobodan Manic: Shopify's agentic dashboard mixes referrals and in-app commerce, so merchants should read order records and channel settings alongside web analytics.
• Community correction: An apparent Google AI Overview tracking parameter spread through X and LinkedIn before Search Engine Roundtable updated its report. The parameter came from a browser extension, not Google. No Google tracking rollout is included in this issue.

From the Tool Blogs

• Ahrefs: A workflow for diagnosing AI visibility problems separates missing mentions, weak recommendations, inaccurate descriptions, source gaps, and access problems. Published September 30.
• seoClarity: Prompt tracking as a sampling method recommends stable topic clusters, varied intent, repeated observations, and business-outcome checks. Published September 29.
• Lumar: A diagnostic checklist for generative-search eligibility starts with retrieval and technical access before content teams rewrite a page. Published September 29.
• Semrush: How old positioning persists in LLM answers covers conflicts between current owned messaging and years of outside descriptions. Published September 29.
• RankZero: A setup guide for AI mention tracking explains what to record across answer engines and how to establish a baseline. Published October 2.
• Profound: The half-life of an AI citation is 11 days reports vendor research across 883,000 pages and 1.19 million page-engine lifecycles. Published September 30; treat the findings as platform research, not a universal benchmark.
• Peec AI: The key to prompt tracking analyzes seven months and 120 million chats to compare prompt breadth, repeated runs, and the noise in visibility scores. Published September 29; this is vendor research.

What to watch

• Whether the branded AI Overview surge persists after Google's late-September rollout changes.
• Whether Google exposes cleaner reporting for branded AI Overview appearances and external clicks.
• Which Shopify AI channels add fees, new eligibility rules, or clearer direct-checkout attribution.
• Whether prompt-tracking vendors publish reproducible methods for separating real movement from sampling noise.

What to do next

• Capture a weekly sample of your highest-value branded searches by country and device.
• Record the AI description, cited domains, visible owned page, and any material error.
• Add titles, descriptions, structured data, alt text, and author profiles to the AI-content review checklist.
• For Shopify, document every enabled AI channel and reconcile its orders against web analytics.
• Split AI visibility reporting into presence, position, description, recommendation, and source type.

Your brand can rank first and still lose control of the first explanation a searcher sees. Audit the answer, the evidence behind it, and the business result separately.

The Visibility Report | Will Scott​
This newsletter is produced collaboratively by Will Scott and Bob, an AI agent. Human oversight, AI efficiency.
​Subscribe at thevisibilityreport.com​