How to Use the Promptwatch Page Tracker to Prove a Page Gets Cited

Brand-level AI visibility is the number on the slide. Page-level citation data is the number that settles arguments. Someone wrote the comparison post, someone owns the pricing page, and eventually someone asks whether either one shows up in ChatGPT answers. A mention chart can't answer that. A per-URL watchlist can.

The Promptwatch Page Tracker is that watchlist. You add URLs, and each row shows how many AI responses cited the page, how many distinct tracked prompts those citations came from, when it was last cited, and, for pages on your own domain, the AI crawler visits and AI click-throughs behind those citations. We rank Promptwatch first in this directory partly because of this view: it joins prompt tracking, crawler logs, and visitor analytics on the same row. A tracker that only reports mentions has nothing to put in the crawl and click columns.

Everything below comes from Promptwatch's Page Tracker documentation and its use cases page, both read on October 15, 2026.

Add the pages

Open Prompts in the left rail, then Page Tracker, and click Add pages. Paste URLs one per line, up to 100 per batch. The dialog flags invalid lines as you type. URLs are normalized and pages you already track are skipped, so pasting the same sitemap section twice does no harm.

The faster path is from citation data you are already reading. Rows in the citations tables have a crosshair icon that adds or removes that URL in one click. When a page catches your eye in a citation table, that is the moment to start watching it.

You can track any URL. Your own pages, a competitor's comparison post, a review site that keeps showing up in your category answers. Only your own domain gets the crawl and click columns, because those come from your crawler logs and your visitor analytics.

The columns, ranked by how often we read them

1. Responses. How many times the URL appeared as a source in AI responses during the date range. This is citation volume and the column you will sort by most.
2. Prompts. How many distinct tracked prompts cited the page at least once. This is the column people skip and shouldn't. A page cited by 15 prompts is structural to how models describe your category. A page with many responses from one prompt is one answer template away from disappearing.
3. Last Cited. "Never" means no citation in the range. On a key landing page, a Last Cited value that keeps getting older is the early warning.
4. Citation Crawls. A sparkline of requests from the bots that fetch pages to cite them (ChatGPT-User, OAI-SearchBot, Perplexity-User and similar). It needs connected crawler logs.
5. Citation Clicks. A sparkline of visits from AI platforms. It needs visitor analytics.

Every column follows the date range, which defaults to the last 30 days and covers at most 90. Click a row to open the details sheet with tabs for the page content, the prompts citing it, and the actual responses. That is where "cited 40 times" turns into "these 12 prompts cite it, here are the answers."

Plan check before you rely on the sparklines

Crawler logs are listed from Professional ($245/mo, 25M crawler logs) and Business ($579/mo, 100M). Essential ($95/mo) lists visitor events (200K) but no crawler-log allowance, so on Essential expect the Citation Crawls cell to show a dash. Agency plans include crawler logs on every tier: Kick-off $199/mo (10M), Growth $399/mo (25M), Scale $799/mo (100M). Explore is free with 10 ChatGPT prompts, which is fine for a first look at citation counts and too small for anything else.

Four ways to use it

Proving a new page works. Add the URL on launch day. The documented order is crawls first, then citations, then clicks. If crawler logs are flowing and Citation Crawls stays at zero after a couple of weeks, the problem is discovery: check robots rules and whether the page is reachable. If crawls arrive but Last Cited stays on "Never," the problem is the content.

Watching key landing pages. Pricing, top product pages, the best comparison posts. Add them once and sort by Responses in a weekly review. When citations fade, the row tells you where to look. Crawls still healthy points to content. Crawls gone points to something technical.

Surviving a migration. Add the old and new URLs before the switch. After it, the old row should fall and the new row should rise. If answers keep citing the old URL for weeks, your redirects aren't reaching the sources models search, and you found out from a table instead of a quarterly dip.

Benchmarking a competitor page. Track the rival page that keeps winning your prompts. Its Prompts and Responses counts are the bar. Open its responses and note what it offers that yours doesn't.

What to do this week

Start with five of your most important pages, one competitor page, and anything you published this month. Pages that models were already citing show history from day one, because counts come from citation data Promptwatch has already collected. New pages start at zero and fill in as prompts run. Remove rows you stop caring about; removing a page from the watchlist leaves its citation data in place.

Promptwatch review: Promptwatch. Product: promptwatch.com. Directory: all tools.

FAQ

Can I track pages I don't own?

Yes. Any URL works. Crawl and click sparklines only appear for your own domain.

Why does a new page show zero?

Counts fill in as tracked prompts run. A page that was already being cited shows history immediately.

How far back can the table look?

The default range is 30 days and the maximum is 90.