Know who arrives from an AI, and what they buy

The script logs AI bot visits, not human visitor data. Identification happens only when a person gives you their email through your own forms, at which point the anonymous history that preceded it is merged into that record, the same way any CRM or marketing automation tool works. On Cloudflare, installation is a single OAuth click, verified on both the apex domain and www, with no snippet to paste. On other stacks the script is a standard tag. Two conditions sit outside the script itself and are worth checking first: the root URL has to be served over HTTPS, and a firewall or robots rule must not block the CiteMeBot/1.0 user agent, otherwise your pages are only partly indexed. If a Content Security Policy blocks the tracker, Citeme detects that automatically and tells you, rather than leaving you with a month of empty charts and no explanation.

Citeme records conversions on your site with their declared value, purchases and signups, attributed over a 30 day window. That number is an estimate: it reflects what happened on the site, not what was eventually invoiced. CRM verified revenue is the stricter figure. When HubSpot, GoHighLevel or Attio is connected, deals that your team actually marked as won flow back into the Impact tab, counted per currency. Leads that came from AI are tagged automatically in the CRM, so the attribution survives inside the tool your sales team already lives in. The two figures leave out different things. The estimate ignores refunds and deals lost later on; verified revenue ignores everything not yet moved to won, so it lags a long sales cycle by as long as that cycle lasts.

It is a probable link, not a certainty, and the product says so. AI assistants rarely pass a clean referrer, so no tool can promise a deterministic chain from a specific typed question to a specific visitor. What Citeme does have is both halves of the picture. It runs your prompts against the ten surfaces it tracks and knows which answers cite which of your pages, and it sees the landing page and timing of the visit. Matching the two gives a probable attribution that gets stronger the more distinctive the landing page is. A visit to a niche comparison page cited in exactly one answer is close to certain. A visit to the homepage is a weaker signal, and it is presented as such. Two limits bound the method. The citation half is only as fresh as the last audit, which is why a rhythm of at least one every two weeks is recommended. And a page that sits behind a login, is marked noindex or is absent from your sitemap is never indexed, so no answer can be matched to it.

Four, and conflating them is the most common mistake in AI analytics. A training crawl is a model harvesting content for a future version, valuable for long term presence but worth nothing this quarter. An AI search index pass is an engine building its retrievable index. A real time retrieval is an engine fetching your page to answer a question being asked right now, which is the signal that a citation is live. A human visitor arriving from an AI answer is the only one of the four that can buy anything. Adding the four together produces a single AI traffic line that mixes machine reads with people, which is how such a number grows while nothing converts. Citeme shows which bot came, why, and on which pages, with a live world map of the human visits.

No. Without a CRM you still get the chain up to the conversion: which prompts earn traffic, which pages receive it, who those people are once they identify themselves, what they did across sessions, and the value of purchases and signups recorded on the site over a 30 day window. What the CRM changes is where the number comes from. Instead of a figure Citeme computed from site events, you get one your sales team confirmed, per currency, on deals they marked as won. If you sell online and your conversions carry their real value, the site side is usually enough. If you sell through a pipeline, the site figure will overstate or understate the outcome depending on your win rate, and the CRM is what closes that gap.