Gumshoe runs Visibility Audits (formerly known as reports) at the base domain level. You cannot scope a Visibility Audit to a single landing page or subpage. Entering a deeper URL in the brand website field will not narrow the audit, because Gumshoe reads the base domain from whatever you type.
If you are looking at a specific page, tell us which one and what you are trying to learn about it. The right answer depends on whether you want to know if that page is being recommended, whether a model read it, or how well it is built for AI.
Why can't Gumshoe audit just my subpage?
Because the AI models do not work that way either. If you enter google.com/chrome, Gumshoe audits Google, not Chrome. The models behave the same: ask one for a recommendation and it names a brand and points to the domain, rather than treating a subpage as a separate entity.
You can see this in the models' own answers. Ask about a Disney movie and the response will often cite theme park pages from the same domain, because the model treats everything under disney.com as one brand. Gumshoe reflects that behavior rather than working around it, so what you see matches what your buyers see.
How do I focus an audit on one product or service?
Set the product category or focus when you create the project. Each project is scoped to one brand, one product category, one location, and one language, so the category is what narrows the audit, not the URL.
Using the example above, you would not audit google.com/chrome. You would create a project for the brand Google with a product category of web browsers. That is the only lever for narrowing a Visibility Audit to one product or service.
To track several products or services for the same brand, create a separate project for each one.
How do I tell whether a specific page is being recommended?
Use the Mention breakdown tab:
- Open the Competitive Leaderboard
- Click your brand
- Open the Mention breakdown tab
This shows the raw URLs the models used when recommending your brand, so you can see whether a specific subpage is coming up or only the broader domain.
You can also export this data. The JSON export carries the full URL attached to each brand mention, and the conversation export CSV gives you the brand's domain for each mention.
How do I tell whether a model read a specific page?
Use the Citations page. It shows the exact URL and page an AI model consulted to answer the question it was asked, so you can confirm whether a model pulled information from a subpage.
Citations come from models that perform live web search and report their sources. Models answering only from training data do not return citations.
Citations are not the same as recommendations. A citation is a page a model read to build its answer. A recommendation URL, shown in Mention breakdown, is where a model pointed when it named your brand.
Can I see which prompt led to a specific URL?
Yes. The Citations page and the Conversations page both show the persona, the prompt, and the model tied to each URL, so you can trace any cited page back to the exact question and audience that produced it.
What if I want a technical review of one page?
Run a Technical Audit. It analyzes a single URL, including a product or landing page, for schema, layout, metadata, and crawlability, and returns an AI Optimization Score with improvement guidance. Technical Audit is the page-level tool. The Visibility Audit stays at the brand level.