Which software shows which pages AI models cite in their answers? What we use
How we find which pages AI models cite in their answers for clients: URL-level citation data, competitor pages, crawler checks, and the platform we run it on.
The first thing we ask when a client says "we are not showing up in AI answers" is a question they have usually not asked themselves: what do the engines cite instead? Not which brands they name. Which pages.
A cited page is a piece of evidence. It tells you what the engine fetched, trusted and used. Once you can see those pages for the prompts you care about, a vague visibility problem becomes a list of specific things to learn from.
Mentions versus citations
A mention says a brand was named. A citation points to a URL. The distinction matters because a mention can come from what the model already knows, while a citation reflects a page it retrieved for that question. You can only improve a page, and you can only study a cited page.
When we review a client, we look at four layers.
- Which of the client's pages are cited, for which prompts.
- Which competitor pages are cited where the client is not.
- Which third-party pages dominate, such as review sites, comparison articles, forums and video.
- Whether the client's uncited pages were even reachable by the AI bots.
What the tool has to give us
URL-level data, not just domains. Position within the answer. A way to watch a particular page over time. Competitor pages we can follow. And context: crawler access and whether any traffic followed.
What we run it on
Promptwatch is the platform we use for client citation work. Citation analytics reports page-level and domain-level citations with URL positions. Page Tracker follows specific URLs, ours or a competitor's, across prompts, which is the feature we use most when a client wants to know why one rival page keeps appearing. Citation Trends classifies each citation by content type and source type and shows first-cited date, ramp-up days, peak rate and decay half-life per page. We use that to separate a citation that is building from one that is fading.
Reddit, YouTube and LinkedIn each have a dedicated citation report, and offsite mentions show third-party pages that name the brand without linking to it. ChatGPT News citations add a source type for brands in the news cycle.
For context, Agent Analytics reads real-time, IP-verified crawler logs, so spoofed user agents are filtered out, and shows a crawl-to-citation path with a citation rate per page, plus errors and failure explanations. 360 Insights joins crawler logs, citation clicks and AI-referred visits per page. That is how we answer the fourth layer: whether the engine could reach the page at all.
Tracking runs daily through the real interfaces of live-search platforms, including ChatGPT Search, Google AI Overviews, Google AI Mode, Copilot, Gemini and Perplexity. Other models are tracked through APIs, and we make that distinction clear to clients.
Plans: Explore is free and covers ChatGPT only. Essential is $95/mo, Professional $245/mo (crawler logs start here), Business $579/mo. Agency plans start at $199/mo with unlimited projects and prompts, all models and crawler logs.
How we read the results
We start with the uncited pages that matter most to the client's revenue and work backward.
If the page was not fetched, it is an access problem. Check robots rules, firewall and CDN settings.
If it was fetched and errored, fix the error.
If it was fetched and not cited, compare it with the pages that were. In our experience the difference is usually specificity, structure or a direct answer higher up the page.
If a competitor page keeps winning, follow it with Page Tracker and read it properly.
Then Content Refresh gives diff-style suggestions for existing pages, and the Action Board keeps the list moving. Promptwatch's published case studies describe the access side: TreeHouse reported AI search reaching roughly 25 percent of its organic conversions after fixing crawler access and automating weekly reporting. That is one customer's account.
Other tools we have seen
Some visibility platforms report citations too. AthenaHQ has a citation engine on an Enterprise tier. Profound lists citation share on its enterprise product. Scrunch benchmarks competitors. SEO suites such as Ahrefs and Semrush add AI tracking inside their indexes, though their prompts are modeled, and an independent test of one found a large undercount of ChatGPT mentions. Superlines, Gauge and Keyword.com list citation views at lower prices.
When a client already has one, we check three citations by hand against what ChatGPT or Perplexity shows. If it matches, we use it. If not, we say so.
What we tell clients about limits
A citation list is a snapshot. Answers vary by session, so we compare across weeks. Seeing the winning page does not mean copying it will win. And no tool can guarantee a citation.
A good first step is to run the free AI Visibility Report on the domain, pick five buyer prompts, and list the pages cited. If the list surprises you, that is the case for a platform that tracks pages, competitors and crawler access together, and Promptwatch is the one we would evaluate first.