Gemini Brand Visibility: Mentions and Sources

Separate Gemini Apps from AI Overviews and AI Mode

Begin every measurement plan with a product-and-surface map. Gemini Apps are Google's conversational applications. A user submits a prompt and may receive generated content, related links, or displayed sources. AI Overviews and AI Mode are features inside Google Search. Google states that the two Search features may use different models and techniques, and their responses and links can vary.

A model family does not make these surfaces interchangeable. A Gemini Apps answer can be influenced by conversation history, account state, connected services, or location. A Search feature depends on Search eligibility, query behavior, and Google's search systems. A developer's use of the Gemini API is a third context with its own configuration and grounding choices.

Create a row for each surface before collecting data. Do not publish “Google AI visibility” as a blended percentage unless the denominator defines exactly which products, queries, markets, and observations it includes.

Build a reproducible Gemini brand visibility prompt panel

Design prompts around real decisions. Keep branded fact checks separate from unbranded discovery, category recommendations, comparisons, implementation questions, and validation questions. A branded prompt such as “What does ExampleCo do?” tests entity accuracy; it should not inflate a discovery metric.

Each observation should retain a stable prompt ID, exact wording, intent class, Gemini Apps surface, device, signed-in state, market, language, date, conversation state, answer text, brand mention, characterization, displayed source-or-related links, and reviewer notes. Repeat prompts under a declared rule because generated answers can vary. Preserve the raw answer before applying labels.

Define metrics only after the data model. Brand mention rate can use eligible responses as its denominator. Owned displayed-link rate counts responses that display an owned URL. Accurate mention rate counts responses whose material brand claims pass a documented fact rubric. Report sample size beside every rate and maintain a smaller exploratory cohort without silently changing the core trend set. Use the AI prompt tracking framework to preserve versions, repetitions, and evidence consistently.

Review Gemini sources and related links carefully

Google's current Gemini Apps help says sources and related links may appear within or below a response, and that not every response includes them. When a Sources control is available, record every displayed destination, its position or nearby claim, whether it is owned or third-party, and whether the page actually supports the statement.

The official page groups sources and related content together, so a displayed destination should not automatically be classified as evidence supporting the nearby claim. Track at least four outcomes independently: mention without a displayed link, mention with an owned displayed link, mention with a third-party displayed link, and no qualifying mention.

Then review claim support. Open the destination and compare its visible passage with the nearby Gemini Apps statement. A link near a brand name does not by itself show that the destination supports every capability, condition, or comparison in the generated answer.

Measure Google Search AI features as a separate channel

Google Search Central describes AI Overviews and AI Mode as Search features that surface supporting links. To be eligible as a supporting link, a page must be indexed and eligible to appear in Google Search with a snippet. Google says there are no additional technical requirements or special schema required for these features, and meeting the requirements does not guarantee crawling, indexing, or serving.

As documented on July 15, 2026, Search Console's dedicated Generative AI performance report reports impressions from AI Overviews and AI Mode with page, country, date, and device dimensions. Google is rolling it out to a subset of properties, and a site may also need enough generative-AI impressions before the report appears. The documented dimensions do not expose queries, clicks, or an AI Overviews-versus-AI Mode split.

Use the ordinary Search performance report for broader Web-search query and click analysis, but do not relabel those rows as dedicated generative-AI feature measurements. Keep both Search datasets separate from Gemini Apps observations. The Google AI Overviews optimization guide covers Search eligibility and page improvements in depth.

Understand Googlebot and Google-Extended as different controls

Google says Googlebot controls site access for Google Search, including AI Overviews and AI Mode. Search preview controls such as nosnippet, data-nosnippet, max-snippet, and noindex can limit what appears from a page in Search, subject to Google's documented behavior.

Google-Extended is different. Google's crawler documentation describes it as a standalone product token that lets publishers manage use for training future Gemini models and for grounding in Gemini Apps and Grounding with Google Search on Vertex AI. Google also states that Google-Extended does not affect inclusion or ranking in Google Search.

Record crawler-policy changes as governance decisions, not routine content tweaks. Use the AI crawler access guide to separate crawler roles before changing policy, then test the actual robots response and annotate the date. A permissive setting does not guarantee that Gemini Apps will use, mention, or display a link to a page.

Read AIsee's Google 86.9 as one report diagnostic

AIsee places a Google provider diagnostic beside other configured provider results in a dated AI visibility report. That comparison can help a team choose where to inspect brand facts, prompts, sources, and website readiness. It is not a census of Google users or a direct feed from every Google product.

Use the value as a triage signal. The evidence for a Gemini Apps visibility claim must still come from preserved Gemini Apps observations, while evidence for AI Overviews or AI Mode must come from Google Search measurement and page-level review. The AIsee visibility audit documents the full report context.

Improve the source, preserve the test, and avoid causal claims

Start with a canonical brand fact record: official name, domain, category, products, audience, supported capabilities, markets, leadership where relevant, and review dates. Make those facts visible on stable pages and keep organization, product, author, policy, and contact relationships consistent. Cite primary documentation close to volatile claims.

1. Capture a dated AIsee diagnostic and a versioned Gemini Apps prompt baseline.
2. Classify material inaccuracies, missing facts, source gaps, and ambiguous category language.
3. Confirm that priority pages are accessible, canonical, internally linked, and supported by visible evidence.
4. Approve a bounded content, entity, technical, or authority action with a human owner.
5. Release through the site's normal review and deployment process.
6. After discovery time, rerun the same prompts and inspect Search evidence separately.

Annotate product updates, prompt changes, website releases, and major external coverage. A later answer may improve after a page change, but an observational before-and-after comparison does not establish that the page caused the result. Report the association and preserved evidence without promising a Gemini mention or Google Search link.

Frequently asked questions