Skip to content
  • There are no suggestions because the search field is empty.

Main Agent Analytics use cases

The dashboard shows you what AI agents are doing on your site. This article is about what to do with that, the questions Agent Analytics is best at answering, and how to read the data to answer them.

Each use case below points you to the specific tab, KPI, or filter to look at.

1. Find out which pages AI is reading — and which it's ignoring

Where to look: Pages tab, or Top pages visited by engines on the Overview.

Sort through your visited URLs and check that your most important pages — product pages, pricing, docs, cornerstone content — are actually being crawled. Then flip the question: search for a page you expect to matter and confirm it's on the list at all.

  • A key page with few or no visits is a crawl gap. It may be blocked, hard to reach, thin on content, or poorly linked internally.
  • A minor page with heavy AI traffic tells you what engines currently think your site is "about" — which may not match your priorities.

💡 Pair this with the GEO Audit. If a page is missing from Agent Analytics, the Audit will often tell you if AI crawlers can reach it.

2. See which engine favors which content

Where to look: the per-engine color split on Top pages, and the Engine filter.

Different engines have different appetites. One may lean toward your blog, another toward your documentation. Filter to a single engine and watch how the "top pages" list reorders itself, that reordering is the insight. It tells you where each platform is most likely to pull you into an answer, and where you'd need to strengthen content to win a different engine.

3. Spot real, user-driven interest (your strongest visibility signal)

Where to look: Agents page visits by types, and the On-Demand AI Fetcher category.

On-demand fetches (ChatGPT-User, Claude-User, Perplexity-User, and similar) happen because a real person asked an AI assistant something and it came to your page to answer. A rise in on-demand fetches on a specific page is one of the clearest signals that the page is winning live AI-assistant traffic, not just being crawled in the background.

Track this per page over time to learn which content is actually being used in answers, then make more of it.

4. Compare engines and track share over time

Where to look: Agent visits over time set to %, grouped by engine.

The share view answers "who's paying attention to me, and is that changing?" A steadily rising Perplexity share or a falling ChatGPT share is a trend worth noticing — it tells you where your visibility momentum is heading.

5. Decide what to allow and what to block

Where to look: the Agents tab and the three categories.

Once you can see who's visiting, you can make deliberate choices:

  • Allow Search Index bots — they keep you eligible to be cited.
  • Allow On-Demand Fetchers — they serve live users your content.
  • Weigh up Training / Data Scraper crawlers — they don't drive near-term visibility, so some teams block them.

The exact robots.txt and firewall rules live in the AI Crawlers Reference Guide.

⚠️ Blocking is a trade-off, not a free win. Blocking a search crawler can remove you from that engine's answers entirely. 

6. Close the loop: from crawl → visit → citation

Where to look: Agent Analytics alongside your Brand Reports and GEO Audit.

Agent Analytics is one part of a three-step story about your AI visibility:

  • GEO Audit — can AI crawlers reach your site?
  • Agent Analytics — do they actually visit, and which pages?
  • Brand Reports — are there mentions and citations inside AI answers?

Read them together and you can trace a problem to its source.

A worked example

Say a cornerstone product page is getting plenty of Search Index visits but almost no On-Demand fetches, and your Brand Report shows it's rarely cited. Reading the tools together, the access is fine (it's being indexed), so the gap is relevance: the page isn't being pulled into live answers. That points you toward improving the content itself, clearer structure, better-answered questions, stronger topical coverage, rather than touching robots.txt or crawlability.

That's the goal of Agent Analytics: not just to show you the traffic, but to tell you which lever to pull next.

Related articles