Most AI assistants do not crawl the web themselves. They read from a search index that somebody else built. That single fact decides everything about how your pages get in front of them.
If your page is not in the index an assistant reads from, nothing else matters — not how well written it is, not how well structured it is, not what it scores on any readiness test. It cannot be quoted, because it cannot be found.
Who reads from what#
| Assistant | Reads from | How your page gets in |
|---|---|---|
| ChatGPT search | The Bing index | Push to Bing — instant, free, set up here |
| Microsoft Copilot | The Bing index | Same as above — one setup covers both |
| Google AI Overviews | The Google index | Push to Google — set up here |
| Gemini, when it searches | Google Search results | Same as above — being in the Google index is the prerequisite |
| Perplexity | Its own crawler, blended with third-party indexes | No direct push exists. Being in Bing and Google helps; the rest is being crawlable |
| Claude, when it searches | Third-party search results | No direct push exists |
Two setups cover the assistants you can actually reach: one for Bing, which carries ChatGPT search and Copilot with it, and one for Google.
Why waiting doesn't work#
Search engines find new pages by crawling. They come back to your site on a schedule they decide, based on how often you have historically changed and how important they think you are. For a new or small site, that can be days or weeks.
That was tolerable when the only consequence was ranking a bit later. It is not tolerable now. When someone asks an assistant about your product this week, a page that gets indexed next month is not late — it is absent from the answer.
Pushing changes that. Both Bing and Google accept a direct notification: this URL is new, or this URL changed. No waiting for a crawl schedule. The engine still decides whether to index the page — pushing is not a guarantee — but it decides now instead of eventually.
What CrawlReady does with this#
Turn indexing on and CrawlReady watches your site for changes and pushes them for you. It picks up new and changed pages from five places without you doing anything:
- Your sitemap, when a URL appears in it or its
lastmodmoves forward - The page content itself, when we detect it actually changed
- Your deploys, if you call our API or connect a Vercel or Netlify webhook
- AI crawler traffic, when a bot requests a page your sitemap never declared
- Direct calls, from CI, our CLI, or the Indexing API
Everything funnels through the same guard rails, which matter more than they sound. Engines deprioritise sites that resubmit unchanged pages, so CrawlReady will not send a page again just because a signal fired — the content has to have actually changed, and even then not more than once a day. A page submitted five times that still is not indexed stops being resubmitted, because at that point more submissions are not the answer.
Reading the results#
The Indexing page in your dashboard answers three questions:
Are pages going out? The submitted count, and a banner if something is blocking them. The most common blocker is an unpublished key file, which stops submissions silently.
Are they getting in? Coverage, from Google. Google is the only engine that reports back which URLs made it into the index; Bing and the others have no equivalent, so pages sent only there show as not checked rather than as a guess.
How fast? Median time to index — from the moment a page was submitted to the moment Google confirmed it was in. This is the number that tells you whether any of this is working.
When a page won't index#
Coverage will eventually show you pages in one of two unhappy states, and they mean different things:
Crawled, not indexed. Google fetched the page and chose not to index it. Submitting again will not change that answer. This is a signal about the page — thin content, near-duplicate of another page, or nothing that distinguishes it.
Discovered, not indexed. Google knows the URL exists but has not fetched it. Usually a crawl-budget or site-quality signal rather than a verdict on the page.
For either, the Diagnose action on the row runs a full readiness scan against that URL, which is where the rest of CrawlReady comes in. A page an engine keeps declining is a content or crawlability problem, and that is a fixable thing — resubmitting is not.
Decisions#
- Two setups, not six. IndexNow covers Bing and everything downstream of it; Google needs its own connection. No other assistant accepts a direct push, and we do not pretend otherwise.
- Coverage is reported only where it is real. Google reports back; nothing else does. Pages sent only to Bing show as not checked, never as a guess.
- Pushing is not a promise of indexing. The engine still decides. What pushing removes is the wait before it decides.
Frequently Asked Questions
Which search index does ChatGPT use?
ChatGPT search reads from the Bing index, and so does Microsoft Copilot — one setup, pushing to Bing, covers both.
Does Gemini use its own search index?
No. When Gemini searches, it uses Google Search results, so being in the Google index is the prerequisite, the same as it is for Google AI Overviews.
Can I push pages to Perplexity or Claude directly?
No direct push exists for either. Perplexity uses its own crawler blended with third-party indexes, and Claude uses third-party search results when it searches — for both, being in Bing and Google helps, and the rest comes down to being crawlable.
Does pushing a page guarantee it gets indexed?
No. Both Bing and Google still decide whether to index the page after a push — pushing is not a guarantee. What it removes is the wait: the engine decides now instead of on its own crawl schedule, which can otherwise take days or weeks for a new or small site.
