Results

Results are unavailable until the poll reaches results published. That applies to everyone, including the person who created the poll.

Before publication

• The results page returns an authorization error, for participants and organiser alike.
• No results link or button is rendered anywhere.
• No page contains hidden result markup waiting to be revealed.
• The totals are not even calculated. Whatever asks for them is refused while the poll is unpublished, so no page can leak a number by mistake.

What the organiser can see during voting is participation: how many people were invited, how many joined, and how many submitted a ballot. Turnout is not secret; what people chose is.

Publishing

Either automatically, the moment voting closes, if the poll was configured that way — or manually from the poll's management page.

Publication is announced to everyone who can see the poll, in their own language, through the in-app inbox and by email.

What the results page shows

The winner, or the tied winners with a note explaining that the tie is real and has not been broken arbitrarily.

A summary distinguishing three different numbers, which are easy to confuse:

A table of every option with its score. For ranked polls that is Borda points and first-place counts; for approval polls it is the approval count and the share of ballots. Each row also carries a bar showing its score as a proportion of the winner's, so the shape of the result is readable without doing arithmetic across rows.

For ranked polls, a distribution table showing how many ballots put each option at each position. This is where you can see whether a winner was broadly liked or narrowly divisive.

For the organiser, the tie-break breakdown: which position separated each pair of otherwise level options, or that nothing could.

Anonymity

Anonymous results are the default, and the results page says which mode is in force.

• Anonymous. Totals only. Nothing links a person to a choice.
• Not anonymous. The page additionally lists who took part and whether they voted — never how they voted. Individual ballots are never shown to anybody but their own owner.

Participants are named from their account, so a renamed profile is reflected here. Anyone who took part before accounts were required appears by the display name they chose, labelled plainly as such rather than presented as a member of the platform. No identity token is ever rendered anywhere.

Reproducibility

Results are recalculated from the stored ballots every time the page is requested. Nothing is cached and no total is stored, so the numbers cannot drift from the ballots that produced them. Ask for the same poll twice and you get the same answer.

CSV export

The organiser can export the final result as a CSV, streamed rather than assembled in memory:

• A metadata block: poll title, voting method, ballots submitted, identified participants, invited count, publication timestamp.
• A row per option: position, subject, description, score, first places or share of ballots, and whether it was tied.

The export is authorized like everything else — only the owner, and only after publication. Column headings follow the interface language.

Withdrawing results

An organiser can withdraw published results. The poll returns to closed and the results page becomes unavailable again.

Two honest caveats. First, withdrawing also switches off automatic publication — otherwise the poll would republish itself moments later. Second, people who already saw the result have already seen it. Withdrawal removes access; it cannot un-ring the bell.