# Which of our pages do AI answers actually cite?

> For the complete documentation index, see [llms.txt](https://attensira.com/llms.txt).
> Every page on this site is also served as Markdown: append `.md` to any URL, or send `Accept: text/markdown`.

> Ask your assistant which of your pages AI answers cite, on which engine and prompt, then check which pages AI crawlers fetched at all. The Attensira MCP chain.

- Canonical page: https://attensira.com/mcp/use-cases/audit-which-pages-get-cited
- Family: mcp-use-cases
- Last updated: 2026-10-04
- Author: Karl-Gustav Kallasmaa (https://x.com/kallasmaa)
- Claims as data: https://attensira.com/claims.json?slug=audit-which-pages-get-cited

**In short:** Ask your assistant to call get_analytics with area=page, days=90 and order=best_first on the Attensira MCP server. It ranks your own URLs by how often AI answers cite them. Open the top few with get_page to see which prompts and engines cite each, then check area=bot for pages crawlers never fetched.

## What it showed on attensira.com (read 2026-10-04)

Asked in Cursor: "Which of our pages do AI answers cite most?"

Call: `get_analytics area=page days=90 order=best_first`

Four pages carry it, and Attensira flags one of them as missing from the sitemap it read.

| Row | Value | Note |
| --- | --- | --- |
| /compare | 61 citations | n = 4,374 answers · ChatGPT and Perplexity |
| / (home) | 60 citations |  |
| /pricing | 58 citations | ChatGPT only |
| /alternatives/profound | 54 citations | flagged as not in the sitemap |

_Our own pages, the www host only, over 90 days._

## Which call answers it?

```json
{ "tool": "get_analytics", "arguments": { "area": "page", "days": 90, "order": "best_first", "limit": 100 } }
```

## What is the chain behind it?

1. **Rank your own URLs by citation activity** Call get_analytics with area=page, days=90, order=best_first and limit=100. A long window, because page-level citations are sparse — most pages collect a handful over a quarter, not over a week.
2. **Open the top few pages** Call get_page with the absolute URL exactly as it appears on the site. It returns which prompts and which models that page earns citations from, which is a better template for the next page than any guess.
3. **Cross-check against crawler activity** Call get_analytics with area=bot and days=90 to see which pages AI crawlers fetched at all. A page that was never fetched cannot be cited, and that is an access problem rather than a content one.

## How do I read what comes back?

| What you see | What it means |
| --- | --- |
| A page cited often across many prompts | A hub. It carries disproportionate weight, so keeping it current matters more than adding a new page. |
| A page crawled but never cited | Reachable and not chosen. That is a content problem rather than an access one. |
| A page never crawled and never cited | Start with access. Nothing can cite a page it has not fetched. |
| Citations concentrated on one model | That surface favours your format. Others may need the page structured differently. |

## What can't this tell you?

- A crawler fetch is not a citation and not a plan to make one. Crawlers fetch pages; models answer prompts; the two are different systems and correlate only loosely.
- AI crawlers do not execute JavaScript, so a client-side beacon has never recorded one. Server-side collection is the only way to see crawler traffic, and the product generates the install instructions for it.

### What did the audit find on our own workspace?

The finding that mattered was not the ranking but the flag. One of the four pages models lean on most, our Profound alternatives page, was missing from the sitemap Attensira read: a page the models chose that our own site was not advertising. That is the kind of result an audit exists for, and the reason to run one before deleting or merging anything.

### Why open the pages that win before the ones that lose?

A page that already gets cited is a working example of what the models want from your domain, on your topics, in your format. That makes it a better template for the next page than any general advice, including ours. Losing pages tell you that something is wrong; winning pages tell you what right looks like on your site.

### Why should the window never change between audits?

Run it on a fixed cadence with the same window. A quarterly audit over 90 days compares like with like; an audit run whenever somebody remembers, over whatever window seemed reasonable that day, produces a series nobody can read. Write down the three pages you expect at the top before you look. Page-level counts are sparse enough that almost any story fits them afterwards, and a wrong prediction is itself the finding.

If the domains in your category matter more than your own pages right now, read [which sources the models cite](/mcp/use-cases/see-which-sources-models-cite) next; if one prompt keeps surfacing, [turn it into a plan](/mcp/use-cases/turn-a-losing-prompt-into-a-plan).

## FAQ

### Why does this recipe use a ninety-day window?

Because page-level citations are sparse. A single page typically collects a small number of citations across a quarter, and a thirty-day window leaves most of your pages at zero for reasons of sampling rather than quality.

### A page is crawled constantly but never cited. What does that tell me?

That access is fine and selection is not. The crawler could fetch it and the models did not choose it, which points at the page's structure and substance rather than at robots.txt or your origin configuration.

### A page is never crawled and never cited. Where do I start?

With access. Nothing can cite a page it has not fetched, so check robots.txt, any bot filtering at the CDN, and whether the page is reachable without JavaScript before touching the copy.

### Why can I not see AI crawlers in my existing analytics?

Because AI crawlers do not run JavaScript, and browser analytics is a script that runs in a browser. By the time the beacon fires, the only visitors left are humans. Crawler visibility requires server-side collection.

### What is the position number on a citation?

A mean index within a citation list per source domain — roughly, how early that domain tends to appear among an answer's citations. It is not a ranking of you against competitors.

## Sources

Each entry below is one claim on this page. `retrieved` is the date the exporter or a human last saw the source say this, and nothing more: it is not a publication date and not a promise of re-verification. `cite` is the anchor on the canonical page where this claim is rendered.

### crawlers-no-javascript

Attensira's MCP recipes state that AI crawlers do not execute JavaScript, so a client-side beacon has never seen one and server-side collection is required.

- source: https://docs.attensira.com/mcp/recipes (docs.attensira.com)
- retrieved: 2026-09-04
- cite: https://attensira.com/mcp/use-cases/audit-which-pages-get-cited#claim-crawlers-no-javascript

> AI crawlers do not execute JavaScript, so a client-side beacon, however written, has never seen one and never will