Entries tagged measurement also appear on the methodology page. Entries tagged console and site describe the product and this website.
site
1 set of tokens, 1 type scale and 1 figure frame across every public page and every console screen. Prices, quotas and plan facts are read from 1 catalogue, and every product figure is the real component fed by a dated run.
console
Every run is now checked for pages the engines cannot read, sub-queries the brand loses, and third-party pages that feed the answers. Each finding names the queries it affects, the engines that ran them and the page cited instead.
console
The AI SERP table, the why-them drawer, the query list, the alert rules and the settings screens read the stored run rows. Every word on screen comes from the message catalogue.
console
Signing in works by e-mail link, and a workspace can hold several people with read-only guests. Alerts fire on a lost core mention, a new competitor in the top 3, a drop in share of answers, and a model change.
measurement
ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews are sampled with the market's location set. Cited pages, the sub-queries each engine ran and the brands named in the text are read out of the stored answer.
measurement
Queries are written in the market's language or imported from Search Console. A run creates 1 sample per query, engine and repeat, and keeps the raw answer behind every number.
site
The repository holds the API, the worker and this website, with 1 message catalogue as the source of every string. The database covers projects, markets, queries, runs and samples.