Movement, not a number that drifts.
Anyone can hand you a score. The useful question is what changed since last week, and whether you need to do something about it.
What changed since last time
Every run is compared with the previous run on the same engine. You get the prompt that stopped naming you, not a score that quietly slid.
Who was named instead
Only rivals you declare are counted, so share of voice means something. Different spellings fold into one name rather than splitting a competitor in two.
How early you appear
Position is distance into the answer, normalised against its length. Being named in the first sentence is not the same as the last.
Which pages were cited
The exact URLs each engine leaned on, split by engine. The most directly actionable list here. Those are the pages worth being on.
Your keys, your spend, no markup.
You bring your own API key for each engine, so runs happen on your account at cost. Before a batch runs, the app shows how many prompts and engines it covers and roughly what it will spend. A monthly cap stops it there.
Free and invite-only, which is a deliberate limit rather than a launch offer. Each account tracks a small number of brands against a fixed set of stored prompts, so the queue and your bill both stay predictable.
What it deliberately does not do.
Worth knowing before you start, rather than discovering later.
- No Guessing your competitors. Names are declared. Automatic entity resolution that gets it wrong quietly poisons every number after it.
- No Sentiment as a headline metric. It is a signal read from the sentence around a mention. Useful, but not precise enough to lead with.
- No Crawling the open web. It reads your own site to suggest prompts. Nothing further.
- No Promising a stable baseline. Providers change models under the same name. When that happens mid-window it is marked on the chart, because a move there is the model, not you.