Gemini answers across its own app and Google’s surfaces, grounding many replies in live results. We measure whether it recommends you, skips you, or gets you wrong — with the captured answer behind every number.
Gemini visibility — whether Gemini mentions, recommends, omits, or misrepresents your product when a buyer asks for tools, across its trained knowledge and its Google-grounded retrieval.
Gemini is Google’s standalone assistant, available in its own app and woven into Google products. It answers using two ingredients:
For a buyer-intent prompt, Gemini names a shortlist built from what the open web repeats, then describes each option. Because grounding strength varies, a product can be cited in a grounded answer and omitted in an ungrounded one — so we test and report both, just as we do for ChatGPT’s modes.
When grounded, Gemini leans on the indexable sources Google trusts: review sites, “best X” and comparison pages, documentation, authoritative editorial content, and structured product data. The same fundamentals that make a page eligible for Google’s AI surfaces make it eligible for Gemini’s grounding — there’s no special markup and no placement to buy.
There is no recommendation dial in Gemini. There are inputs you can improve and outcomes no one controls.
Answerability — clear, self-contained answers to buyer questions on crawlable pages.
Sources — presence and accuracy on the review and comparison sites Gemini’s grounding pulls from.
Structured data — accurate, standard schema so your facts are unambiguous.
Crawlability — indexable pages, so they’re eligible to be grounded at all.
Inclusion — whether any given answer names or cites you. Output is probabilistic.
Grounding — whether a prompt triggers live retrieval at all.
Training data — what the model already learned is fixed until retraining.
No special markup, no paid placement, no inclusion guarantees.
Illustrative figures for the fictional category “AI sales assistants for seed-stage SaaS.” We note whether each answer was grounded, then whether you appeared.
Reported separately for grounded and ungrounded answers. See the full sample report.
We do not promise a Gemini recommendation, citation, or ranking. We measure your current visibility, distinguish grounded from ungrounded answers, and prioritize fixes to the public evidence Gemini’s grounding can read. Snapshot starts at $99.
Gemini blends what it learned during training with Google-grounded retrieval when it reaches to the web. For buyer-intent prompts it tends to name a shortlist drawn from widely-repeated, indexable sources, then describe each. As with every model, output is probabilistic — there's no dial to set its recommendation.
Related but distinct. AI Overviews and AI Mode live inside Google Search; Gemini is the standalone assistant (app and integrations) that also uses Google grounding. Behaviour and citations can differ between them, so we audit and report Gemini separately from the Search surfaces.
Often, especially when it grounds an answer in live Google results — it can show supporting links. Grounding strength varies by prompt and mode, so we record per prompt whether Gemini cited sources and which ones, rather than assuming every answer is grounded.
No. There is no inclusion or ranking guarantee in Gemini. We measure your recommendation, mention, omission, and misrepresentation rates — each as a numerator over a denominator — and improve the public, citable evidence Gemini's grounding can read.
Like other engines, misrepresentation traces to stale or thin public sources, or to confident generation where grounding is weak. We capture the exact answer as evidence, note whether it was grounded, trace the likely source, and prioritize correcting the underlying page.
Start with a $99 Snapshot, then move into Monitoring for recurring Gemini visibility and competitor movement.