A page stuck in Google's "Crawled – currently not indexed" status cannot appear in AI Overviews or AI Mode either. Google's AI surfaces draw only from its Search index, and its own documentation says a page must be indexed to be shown there. So a page Google reads and declines to index is invisible to Google's AI answers, not just its blue links. This piece explains what Google said is behind the status, why the usual cause is "commodity content," and what it means for AI visibility.
It pairs with a companion how-to, how to diagnose and fix pages Google crawls but won't index, and it extends the argument in is GEO just SEO?: the indexing quality bar and the AI-citation bar are converging into one bar.
A page Google won't index can't be an AI Overviews source. The quality threshold for the classic index and the quality threshold for an AI citation are turning into the same threshold.
What does "Crawled – currently not indexed" mean?
It means Googlebot fetched (downloaded) your page but chose not to add it to the Search index. The page was reachable and crawlable; Google simply decided not to keep it. This is different from "Discovered – currently not indexed," where Google is aware of the URL but has not crawled it yet.
| Status | What Google has done | Typical cause |
|---|---|---|
| Discovered – currently not indexed | Knows the URL exists; has not crawled it | Crawl-budget or scheduling; Google hasn't fetched it yet |
| Crawled – currently not indexed | Fetched the page; declined to index it | A technical issue hiding content, or a quality judgment that the page isn't useful enough |
Both appear in the Search Console page-indexing report under Indexing → Pages. Some pages should sit in the crawled-not-indexed bucket: non-canonical duplicates, pagination, feed URLs, and parameter variants are normal there and need no fixing. The concern is when pages you want indexed are stuck.
Why does an unindexed page also vanish from AI Overviews?
Because Google's AI surfaces are built on the same index. Google's Search Central documentation states that "to be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet." Being crawled is not enough; the page has to clear indexing.
That single requirement turns a classic SEO status into an AI-visibility problem. If your content sits in crawled-not-indexed, it is not a candidate for AI Overviews or Google AI Mode at all, no matter how well it would answer the query. The index is the gate, and the gate is upstream of every Google AI answer. This is why optimising for AI Overviews has to start with confirming the page is actually indexed.
What did Google say is behind it?
At an April 2026 Google Search Central event in Toronto, a Google presenter discussed why Google crawls a page and then declines to index it, as reported afterwards by SEO analyst Marie Haynes. (The event organisers asked attendees not to attribute quotes to specific Googlers but permitted sharing what was said, so the quotes below are relayed second-hand and should be read as directional paraphrase.) The presenter framed crawling as downloading a page, then: "If we think it's useful we might put it in a database," meaning the index.
Two causes were described:
| Cause | What it looks like | How common |
|---|---|---|
| Technical issue | Google fetches the page but sees little content. Example reported: a robots.txt rule like Disallow: /*?*, meant to block tracking parameters, also blocked the parameterised CSS and JavaScript the theme relied on, so crawlers saw only a heading and boilerplate |
Rarer; worth ruling out first |
| Quality | Google looked at the page and, per the paraphrase, "found it not to be good," often because many pages already cover the same topic and yours adds nothing more useful | The common case |
The presenter said Google now wants to index content that offers two things: first-hand personal experience, and knowledge no one else has. The reasoning given was that generative AI has lowered the threshold for producing content, so when anyone can generate a page on any topic, the bar for what earns a place in the index rises. One relayed line went further, describing an experiment "with seeing which one produces happier users," implying Google sometimes indexes a page temporarily to test user response. Treat that as one analyst's account of a single event, not official policy; the durable takeaway is the direction, corroborated by trade coverage in mid-2026: generic, easily-replaceable content is increasingly left uncrawled-into-index.
Isn't heavy crawling a good sign for AI visibility?
Sometimes, but do not confuse two different signals. Being crawled repeatedly by AI training and retrieval bots such as GPTBot or ClaudeBot, with no citations yet, is usually normal lag, not failure. Crawl volume tends to spike weeks before an engine's index picks a page up, as covered in the crawl-to-cite lag and the dark library effect. You should not panic-edit a page just because bots read it and citations haven't started.
Google's "Crawled – currently not indexed" is a different animal. It is not an AI crawler building a corpus; it is Googlebot's index-quality gate returning a verdict of "not this one." Two distinctions keep this straight:
- Different crawler, different decision. GPTBot reading your page says nothing about Google's index status, and vice versa. They are separate systems.
- Lag resolves itself; a gate does not. Crawl-to-cite lag ends when the engine indexes you. A crawled-not-indexed verdict is a standing exclusion from Google's AI surfaces until you change what caused it.
So the response differs: ride out crawl-to-cite lag; act on a crawled-not-indexed verdict.
What does this mean for GEO?
It collapses two to-do lists into one. The signals that get a page indexed by Google are increasingly the signals that get it cited by any AI engine: it must be reachable and parseable, and it must clear a rising quality bar built on first-hand experience and unique knowledge rather than restated common knowledge. That is the same standard behind entity strength and earning AI citations.
The practical move is to treat your crawled-not-indexed report as an AI-visibility report. Pages sitting there are pages your Google AI answers can never use, and they are a strong hint at what other engines will skip too. The companion piece, how to diagnose and fix pages Google crawls but won't index, walks the checks step by step.
Watching whether your pages are indexed, reachable, and actually cited across ChatGPT, Gemini, Claude, Perplexity, and Google's AI surfaces, over time, is exactly what Buffy Intel is built to measure.