Category
PendingHow the category was drawn, and what puts a store inside it rather than next to it.
Data
No benchmark is publishable yet.
Publishing one takes a sample large enough to mean something and an aggregate no merchant can be read out of.
What a published figure will carry
- The category, and its edges
- Stores measured
- Engines sampled
- The measurement window
- Rubric version
Neither condition is true today, so this page holds no numbers.
Every figure will be read across these
Nothing to list
No category benchmark is published.
This is the index a published set would fill. It stays empty until a category clears all three conditions below — and some categories will stay unpublished for a long time, because in a thin category the average names the merchants in it by implication.
The bar
Stated before we have to meet it.
Publishing standards written after the data is in are not standards, they are justifications. These three were written while the page is still empty.
- Enough samples to draw a line through 10 minimum
- Anonymous by construction Aggregate
- Comparable, versioned, dated Rubric ver.
The same rule, inside the app
“Not enough data yet”
Below ten samples this is what a merchant sees instead of a score band. It is not a loading state and does not resolve into a number on its own — it is the product declining to say something it cannot support.
- A single store’s score Prompts × engines × cycles
- Measurement window 21 days
- Reputation battery 12 questions
Disclosure
What a published benchmark will carry.
Six things, printed next to the figure rather than in a footnote — so you can judge whether it applies to a store like yours before taking it seriously.
Stores measured
PendingHow many stores are in the aggregate. If the number is small it is printed small, not rounded into respectability.
Engines
3 or 5Which of the five were sampled. Plans differ — three on Presence, five on Authority and Omnipresence — so a mixed sample has to say so.
Window
21 daysThe window behind the figure. Scores are computed over a rolling 21-day window, and the questions fire on a rotation rather than all at once.
Rubric version
VersionedThe readiness rubric the figure was computed under, because a version change can end comparability with anything published before it.
What it is not
Not a tableNot a league table of named stores, not a forecast, and not a causal claim about what any single change would have done.
Questions
Answered from what is already on this page.
Why does this page have no numbers on it?
Because the sample is not large enough yet, and in a thin category an average names the merchants in it by implication. Both conditions have to clear first.
What counts as enough samples?
A store’s score is prompts × engines × cycles over thirty days. The app draws no confidence band under ten samples; a category aggregate is held above that bar, never below it.
Will this be a league table of named stores?
No. A benchmark here is a category aggregate that no individual store can be read back out of, and no merchant is named without agreeing to it.
Which assistants are sampled?
Three on Presence, five on Authority and Omnipresence — ChatGPT, Perplexity, Gemini, Claude and Grok. A mixed sample has to disclose that it is mixed.
Can I compare a figure published later with one published now?
Only within a rubric version. Readiness is scored against a versioned rubric, and when a revision ends comparability the series restarts with it.
Is there any benchmark I can get today?
Your own. The free audit asks the assistants about your store and shows you what came back, which is the only figure that needs no category average.
Meanwhile
One benchmark exists today: yours.
You do not need a category average to find out whether the assistants can name your store.