What you can see while a run is happening, what you get once it is over, and what happens to people's answers after that.
Nothing has to be exported for you to see what is happening. Four things arrive, in this order.
The host console counts submissions as they land, question by question. The projector view shows the same thing sized for a room, and nothing on that screen can change the session.
A question is open, then locked, then revealed โ three separate states, so you decide when a distribution goes on the wall instead of it appearing the moment someone answers.
Once a run is over, its results page lists every open-text answer in full. Themes and sentiment are an optional pass over them, started by you because it spends an AI credit; if that pass fails you still have every answer.
Every answer, question by question, with the totals beside them โ and a CSV or Excel file of either scope when the numbers belong somewhere else.
For you as well as for everyone else. While a survey is still collecting there is no running tally anywhere in the product, and an aggregate export of it is refused โ a partial result changes who you chase and when you stop. It unseals once the session is ended, closed or archived. A live vote is the opposite by design: its tally is revealed in the room, by you, when you decide.
Whether the window is open is worked out from its opening and closing times every time it is read, not from a flag some background job sets. So the deadline holds even when that job is late โ and an answer submitted after the closing time is refused rather than quietly counted.
There is one way out and it is a file, so there is no picker to get wrong.
Files
CSV and Excel, offered side by side rather than behind a picker. Detail is one row per answer; aggregate is the summary. The results page also prints without the app's navigation wrapped around it.
The audit log
Reopening a question, exporting, and the membership changes around a session are each recorded as one line, with who did it and when. It is what lets a results page say a run was reopened rather than quietly showing different numbers than the person who read it yesterday.
Participant data from a finished run is deleted after a set number of days. A number of days on its own invites the wrong assumption in both directions โ that your questions go too, or that nothing goes โ so here are both halves.
Deleted
Who took part, their email addresses, their answers and drafts, and any themes drawn from them โ those quote people directly.
Kept
Your questions, and each run's totals โ so a finished run still reports how many took part after the answers behind that number are gone.
A session that is still open is never touched. The clock starts when a run finishes, so an async window left open for a term is not counting down against itself.
Deletion means what it says: those answers are gone rather than hidden, and no later export can bring them back. If a run matters, take the CSV or the Excel file while it is still there.
Everyone in a workspace can see the retention window, because a collaborator running sessions is the person who gets asked how long the data is kept. Changing it is an owner's decision, set in days within a fixed range โ ask us and we will set it for you; there is no screen for it at the moment.
Two pages that decide what the results described here will look like.