Discussion starter: an AI search question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- AI Search
Is Gemini app visibility a separate problem from AI Mode visibility, or have you seen the same pages cited in both for the same prompts?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
Niraj Raut
Treat them as related but separate. AI Mode is part of Google Search and draws on Google’s search index and ranking systems, so pages that rank well for the underlying searches tend to be the ones it cites. The Gemini app is a separate assistant product: it can search the web to ground its answers, but it also answers from the model’s own knowledge, behaves differently in conversation and presents sources differently. In practice there is likely to be overlap, particularly for questions where Gemini searches, but you should not assume the same pages are cited in both. Measure them separately, because the data sources differ: Search Console’s generative AI reports cover AI Overviews, AI Mode and Discover’s AI features, not the Gemini app, while Gemini referrals can be seen in analytics.
How the two differ
AI Mode Gemini app Where it lives Google Search Standalone assistant app and web product Main source of answers Google’s search index Model knowledge plus web search when needed Reporting Search Console generative AI reports (impressions); clicks within web search totals Analytics referrals from Gemini; no Search Console report Controls Search generative AI control in Search Console, at site level Google-Extended for use of content in Gemini model training How to check overlap for your content
- Build a prompt set that reflects your audience’s questions.
- Run each prompt in AI Mode and in the Gemini app, repeatedly, recording cited sources.
- Compare which of your pages appear in each, and which competitors appear instead.
- Note where the Gemini app answers without citing sources, which suggests it relied on model knowledge.
What tends to help in both
- Pages that rank well for the searches behind common prompts.
- Clear, specific answers stated in the text.
- Accurate, consistent facts about your brand across your site and third-party sources, which matter especially when an assistant answers from model knowledge.
- Presence in the independent sources that assistants rely on for your category.
Hypothetical example:
A travel insurer runs 40 prompts in AI Mode and the Gemini app. AI Mode cites its policy comparison pages for most questions about cover levels, reflecting their strong search rankings. The Gemini app cites them for about half the same questions, answers others from general knowledge without sources, and in a few cases cites a comparison site instead. For questions about the insurer’s brand, the Gemini app repeats an outdated fact about its excess levels, traced to an old third-party article. The insurer updates the third-party source and its own pages, and tracks the two products separately from then on.
Watch for this:
The controls are different. Opting out of Google’s generative AI features in Search Console affects AI Overviews, AI Mode and AI Overviews in Discover; it is not the same as the Google-Extended control, which relates to use of content for Gemini model training. Understand which control affects which product before changing anything.
Model knowledge versus retrieved sources
When an assistant answers from its model knowledge rather than a search, what it says about your brand reflects what was widely written about you in the data it learned from, which may be months or years old. You cannot update that directly. What you can do is make sure the sources that will inform future models, and that assistants retrieve when they do search, are accurate and consistent: your own pages, major profiles, reference sources and prominent third-party coverage. For brands with outdated facts in assistants’ answers, correcting widely read third-party sources often matters more than anything on the brand’s own site.
Keep prompts identical when comparing
The two products respond differently to phrasing, and the Gemini app may use context from earlier in a conversation. Use identical prompts in fresh sessions, with the same account settings, when comparing them. Otherwise differences in the answers may reflect the test rather than the products.
Reporting
Report AI Mode through Search Console’s generative AI impressions and your prompt tracking, and the Gemini app through analytics referrals and prompt tracking. Keep them in separate lines of the AI visibility report, with a note on overlap from your prompt tests. Combining them into one “Google AI” figure hides the differences that tell you where to act.
Treat AI Mode as an extension of search, where rankings drive citations, and the Gemini app as a separate assistant with its own behaviour; test both with the same prompts, measure them from their own data sources, keep brand facts accurate everywhere, and understand which control applies to which product.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.