You publish a page, AI bots crawl it hard for weeks, and it still doesn't show up in a single answer. That gap is normal. It's the citation lag, and understanding it changes how you read your own data.
What the citation lag is
The citation lag is the delay between an AI bot crawling your page and the engine actually citing it in answers. A page moves through three stages. It gets crawled (trained on or fetched), it gets indexed (added to the engine's retrieval layer), and only then does it get cited at answer time. The lag sits between the first stage and the last, and it commonly runs to weeks.
This maps directly to the AI crawler lifecycle: train, index, answer. Crawling is the entry point, not the finish line.
What practitioners are reporting
Field data shared by the CEO of Promptwatch on LinkedIn in June 2026. Single-vendor, self-reported, so directionally credible rather than definitive. Described the pattern in concrete terms:
- Crawl precedes citation by weeks. GPTBot crawl volume on a page spiked before OpenAI's search index picked the page up and citations began.
- Explainers overtook comparisons. Broad comparison guides were cited fast but plateaued; focused single-topic explainer pages with deep context broke through later. Reportedly climbing from roughly 250 to about 1,700 daily citations.
- Google rank didn't predict ChatGPT citation. A page's position in Google search did not correlate with how often ChatGPT cited it.
Treat the numbers as one practitioner's data, not a law. The shape of the finding. Crawl first, citation later, explainers durable. Lines up with how retrieval systems are known to work.
Why heavy crawling with zero citations is a good sign
Brands often panic when AI bots hammer a new page that earns no citations. That reaction reads the signal backwards.
Heavy bot crawling with no citations yet is a leading indicator, not a failure. The crawl spike is what happens before the engine starts citing you. Judging a page dead in week two is judging it before the pipeline has run.
The implication for measurement: track crawl activity and citations as two separate signals on a timeline, not one. Rising crawl with flat citations early on is the expected mid-pipeline state. Flat crawl and flat citations weeks in is the real warning sign, that's a discoverability or relevance problem, not a patience problem.
What it means for what you publish
If citation follows crawl by weeks, the strategy writes itself:
- Publish explainers as the backbone. Deep, single-topic pages start slower but climb higher and hold. They're built for durable citation, not a quick spike. (This is also why AI loves a focused, comprehensive page over a thin one.)
- Use comparisons for quick wins. They get cited fast; just don't expect them to keep climbing.
- Don't chase Google rank as a proxy. Ranking #1 in classic search doesn't guarantee the AI citation. They're scored differently.
- Give pages a quarter before you judge them. Pair the lag with the 3-month citation cliff: you wait weeks to start getting cited, then citations decay after roughly a quarter without a refresh. The window is real but finite.
The practical takeaway: measure the crawl-to-cite pipeline as a timeline, be patient through the lag, and weight your calendar toward explainers. Watching crawl activity turn into citations over the weeks that follow. Across every engine. Is exactly what Buffy Intel tracks.