Welcome to Citations.io — measuring AI visibility, not rank
Why we built Citations.io and how sampled answers + citation evidence replace traditional rank tracking in the age of generative search.
The shift from links to answers
For twenty years, SEO measured one thing: did a link to your site appear, and how high. That model breaks the moment a user reads an answer instead of a list.
When someone asks ChatGPT, Perplexity, Gemini or Claude about a category — accounting software, running shoes, a London plumber — the engine synthesises an answer. Your brand is either named in that answer or it isn't. There is no page 2.
What Citations.io measures
We sample real prompts across the four engines that matter, on a daily cadence, and track:
- Mention rate — how often your brand appears in answers for the prompts you care about
- Recommendation rate — how often the engine actively recommends you
- Citation coverage — which sources the engines cite to support those mentions
- Source influence — which domains are doing the work of putting you in front of buyers
- Competitor visibility — the same metrics for every brand named alongside you
Everything is grounded in verbatim sampled answers, not modelled scores. You can open any number on the dashboard and read the answer it came from.
Why sampling, not scraping
There is no public API for "what did ChatGPT say last Tuesday". Citations.io samples a representative prompt set, captures the full answer, and computes share-of-voice with confidence labels so you know whether a 12-point shift is signal or noise.
This is the AI-era equivalent of SERP tracking — and the methodology is in the methodology guide.
What's next
More field notes coming: how to pick prompts that matter, what "good" looks like by category, and the difference between AEO and GEO in practice.