# AI indexing, Live fetches, AI training

> How many pages AI crawlers requested in the period, by the three purposes.

**Each number is the count of requests made by verified AI crawlers, for that purpose, in the chosen period.**

## Formula

```
AI indexing   = verified requests from indexing crawlers (OAI-SearchBot, Claude-SearchBot, PerplexityBot…)
Live fetches  = verified requests from live-fetch crawlers (ChatGPT-User, Claude-User, Perplexity-User…)
AI training   = verified requests from training crawlers (GPTBot, ClaudeBot, Google-Extended…)
```

Every request counts once. A crawler reading the same page 10 times makes 10 hits. The purpose comes from the
[crawler list](https://survelium.com/docs/ai-traffic/crawlers.md#crawler-list).

The change next to the number is a percentage against the same length of time before:

```
change = (this period − previous period) ÷ previous period × 100
```

## Example

In the last 28 days, OAI-SearchBot requested 120 pages and PerplexityBot 80, all from verified IPs. Another 40
requests came from a "PerplexityBot" on an address that is not Perplexity's:

```
AI indexing = 120 + 80 = 200
```

The 40 are spoofed and do not count. If the 28 days before had 160, the number shows **+25%**.

## What counts and what doesn't

| Counts | Doesn't count |
|---|---|
| Page requests (`GET`/`HEAD`) by listed crawlers, verified | Spoofed requests (the IP is not the operator's) |
| Requests with any status code, 404 or 500 included | Unverifiable and unchecked hits — unless you switch on "Include unverifiable" |
| | Googlebot, Bingbot, SEO tools, monitors |
| | Crawlers removed from the list, even if their hits were received |
| | AJAX calls, form posts and paths excluded in the integration (e.g. `/wp-admin`) |

When there are unverifiable hits, a line under the numbers reads "+N hits from crawlers that cannot be verified are
not counted. Include them" — one click adds them.

## How to read it

- **AI indexing** is the closest to future visibility: indexed pages can be found and quoted when someone searches.
  Compare it with [visibility](https://survelium.com/docs/metrics/visibility.md) and [used as source](https://survelium.com/docs/metrics/used-as-source.md).
- **Live fetches** mean someone asked an assistant something and it opened your page to answer. It is the most direct
  sign you are used as a source right now.
- **AI training** brings no traffic or citations right away. Many sites block it in `robots.txt`; that is your choice.
- A very large number from one crawler often comes from a full recrawl of the site; see **AI crawlers** for detail.
- The numbers only come from pages that run PHP. A page served from a cache that crawlers do not bypass is missing;
  see your platform's install guide.

## Where it appears

- **AI traffic → Overview** — the three numbers, the change and the "AI traffic per day" chart, stacked by purpose.
- **AI traffic → Pages** — the same three columns, per page.
- **AI traffic → AI crawlers** — the hits of every crawler, with its purpose.