A rank tracker tells you where a link sits in a list of ten. An answer engine doesn't return a list of ten — it returns a paragraph that names two or three companies. WellCited asks the engines your buyers' questions, keeps the paragraph, and tells you whether your name is in it and which page earned the citation.

No. We never ask for CMS access, we don't install a script, and there is no plugin. WellCited generates the file or the markup, you paste it in, and then you press verify — at which point we re-crawl the public URL and tell you what we actually found there.

ChatGPT, Perplexity and Gemini. Starter samples ChatGPT and Perplexity; Growth and Agency sample all three. We also check whether nine AI crawlers — including GPTBot, PerplexityBot, ClaudeBot and Google-Extended — are allowed to read your site at all, because that check is free and it is the one that most often comes back broken.

Adoption is not universal and we will not pretend otherwise. llms.txt is a proposed convention for giving a model a clean, linkable map of your site, and it costs one static file to publish. We generate it because it is cheap, harmless and occasionally decisive — and because writing it forces the structure that the checks which definitely matter, like schema and answer-shaped headings, also reward.

One credit is one prompt answered by one engine. Twenty-five prompts across two engines is fifty credits. Starter includes 200 credits a month, Growth 600, Agency 2,000, and scan packs are available if you want to run something extra without changing plan. Plan credits refresh each period; pack credits never expire.

Yes, and it needs no account. It runs seven heuristic checks against your live page — no language model is involved, so it costs us close to nothing and we have no reason to gate it. The score is never hidden behind a signup, because a score you cannot share is a score nobody sees.

Yes — three on Starter, five on Growth, ten on Agency. You name them, and every scan records when an engine reached for one of them instead of you, which prompt it happened on, and which of their pages was cited. That last part is usually the most useful thing on the screen.

No. There is no demo to book, no seat count to negotiate and no discovery call. Sign up, paste a domain, run a scan. If something breaks, email us and a human answers — that is the only conversation on offer.

Yes. Every free audit publishes a page at its own address, and that page shows the score, the grade, all seven checks and the crawler table to anyone who has the link. The address contains a random token, so nobody can enumerate the audits or harvest the domains that have been run — but a link, once shared, is a link. Reports are never deleted for scoring badly: the number is the product, and hiding it would defeat the point of publishing it.

You get a new scorecard at a new address, and the previous one stays where it is. Only the newest audit of a domain is marked canonical: the older pages keep working for anyone who already has the link, they carry a banner pointing at the current report, and they tell search engines to index the current one instead. That is why a re-audit never breaks a link you have already shared.

Six of the seven checks are things you can change in an afternoon — a robots.txt line, an Organization block, a description, an llms.txt, a sitemap, a heading rewritten as the question it answers. The seventh, content extractability, is a rendering decision and can take a sprint. Re-run the audit as soon as the change is live: it fetches the page fresh every time, so the new number reflects what is actually being served, not what we remember.

They keep working. Everything WellCited generates is plain text or plain markup that lives on your own server — an llms.txt file, a JSON-LD block, a heading rewritten in your own CMS. There is nothing to disconnect, no script to remove, and no part of it stops functioning when the subscription does. What you lose by cancelling is the weekly re-scan and the verification, not the work.