Loading…

Revenue figures are estimated from stored plan and credit-pack data, not a live Stripe pull — see the numbers as directional, not accounting-grade.

Per day, UTC. Runs are billed tool calls; active users are accounts that made at least one. Demo tries include blocked repeats.

The 1–5 star rating asked for after a dashboard run (and the stars on each run in “Your runs”). Per day, UTC. “Overall” is the average of every rating so far. Your own and test accounts are left out.

Latest rated runs

Counted from the database, not Google Analytics, so it includes visitors who declined analytics and anyone who only uses Enrichr through Claude. Activated = ran a billed call. Hold GA4's numbers against this.

By acquisition source

Distinct accounts at each step between a first useful result and a payment. A “useful result” is a run that finished with something in it (complete or partial), not just a billed call. Steps are counted independently: someone can see the offer next to a finished result without ever being refused, so each is measured against everyone who got a result. Your own and test accounts are left out.

Answers to “What stopped you from using Enrichr again?”, asked only of people who got a result and have been away two days. They tell apart a one-time look, confusing results, a failure, and price. Your own and test accounts are left out.

From the run ledger: every capture that was started, whether or not it was billed. Failed is a run that produced nothing usable (an error, a bot wall, or a timeout); those are not charged. Queue wait is time spent waiting for a browser, kept apart from how long the work took. Last 7 days against the 7 before; the owner’s own testing is left out.

Slowest and failed runs (last 7 days) — the stages that finished, in order. A run that was cut off stalled in the step after the last one listed.

Counted by the server on every request, because Google Analytics never sees bots and ChatGPT often strips the referrer. Search agents build the index answers are cited from; user agents fetched a page because someone asked the assistant to; training agents collect for models. Referral rows are people who clicked through from an assistant. Last 30 days.

Where assistants send people

The page a visitor landed on after clicking through from ChatGPT or another assistant.

Pages assistants read for users

Fetched while an assistant was answering a real person’s question (for example ChatGPT-User). These are the pages being used as answers.

Most fetched pages

Accounts by plan

Everyone who tried the homepage demo, newest first — including repeat attempts that were blocked. Location comes from the IP address; Signed up matches an account created afterwards from the same IP, so treat it as a good guess rather than proof.

Click an account to see its full profile — usage, keys, and whatever signup context could be discerned.

Banned emails

Blocked from ever signing up again, until unbanned here. Ban and revoke both live on an account's profile — click into an account above to use them.