Platform performance
Everything we claim, and how it’s counted.
Software companies quote numbers from case studies chosen for being flattering. These are every workspace, since launch — including the quiet ones. When the sample is small it says so, because a number without its denominator isn’t evidence.
- 7 minfrom asking to the first answerMedian, across panels that got one
- 45%of the people asked actually answeredEvery routed person, every panel
- 1voices in a typical answerMedian memos per answered panel
- 50%of askers said “this would have been a meeting”2 askers, their own tap
- 1disagreements surfaced instead of smoothed overNamed in a merged answer, checked against real respondents
4 answered panels across 1 company, refreshed hourly. Check the numbers.
How each one is counted
- From asking to the first answer
- Median across every panel that got one: the gap between the question being posted and the first memo arriving. Panels nobody answered are not in the median — they have no first answer to measure — so this is a speed figure, not a response figure. The next one is the response figure.
- Of the people asked, who answered
- Every person routed to a panel, over every person who recorded or typed something on it. Nudges do not reset it and unanswered panels count in full.
- Voices in a typical answer
- Median memos per answered panel. The median rather than the mean, so one busy panel can’t lift it.
- Would this have been a meeting?
- A single tap from the person who asked, offered once per panel, reported as the share who said yes — with the number of people who answered the question shown beside it. It is the one outcome figure here, and it is theirs rather than ours.
- Disagreements surfaced
- Counted where the synthesis named two or more people taking positions that cannot both be true, and every name resolved to someone who actually answered. A name the model invented is dropped rather than counted.
Demo workspaces are excluded. Nothing here is self-reported by a customer, and nothing is projected. Why we build it this way →