Poll Result Percentage Calculator – Shares That Add Up to 100%
A poll finishes, you have a list of options and a vote count beside each one, and now you need percentages to post back into the thread. The arithmetic looks like the easiest thing in the world: votes ÷ total × 100. The part that trips people up is what happens after the division, when those numbers have to be rounded for display. This poll result percentage calculator takes the options and raw counts from an X poll, an Instagram Story sticker, a LinkedIn or Facebook poll, a Discord or Slack reaction vote, a YouTube Community post or a show of hands in a group chat, and turns them into a ranked, copy-ready results block whose percentages sum to exactly 100%.
Why separately rounded percentages do not add up
Take the smallest awkward poll there is: three options on one vote each. Every share is 33.333…%, and rounding each one on its own to a single decimal prints 33.3 + 33.3 + 33.3 = 99.9%. Nothing is wrong with any individual figure and the column is still wrong. Readers notice, screenshot it, and reply asking where the missing tenth went. The same thing happens in the other direction: six options can each round up and produce 100.2%.
The fix is the largest-remainder method, also called the Hare quota — the same apportionment rule used to turn vote totals into whole parliamentary seats. Instead of rounding each share independently, the calculator decides how many indivisible display units exist (1,000 tenths of a percent at one decimal place) and hands them out. Each option first takes the whole units its share earns outright, and the few units left over go to the options with the largest unrounded remainders. Applied to the three-way tie, the leftover tenth goes to the first option and the result is 33.4 / 33.3 / 33.3 — exactly 100.0%.
Adjustments are shown, never hidden
Whenever reconciliation moves a row away from its own naive rounding, that row is marked with a small +0.1pp or −0.1pp badge, and the unrounded share is printed beside it to four decimal places. Nothing is quietly nudged behind your back: you can see which option collected the leftover unit and check the figure yourself. Ties on the remainder are settled by the higher vote count and then by position in your list, so the same poll always produces the same percentages no matter how many times you run it.
Ranks, margins and what it takes to overtake
The results table is sorted by votes, with a competition rank that lets joint-placed options share a number. The leader's margin over second place is reported in both votes and percentage points, and it is computed from the exact shares before rounding so it can never contradict the reconciled column by a stray tenth. Every non-leading option also gets a votes to lead figure: the smallest number of extra votes that would put it past the current top count, assuming every other count stays exactly where it is. A collapsible target-share panel answers the related question — how many more votes would take an option to 40%, or to any share you name.
Edge cases that usually break poll maths
An exact tie for first does not produce a winner. The page names every option on the joint-top count and sets the margin to zero, because a tie has no margin. An option on zero votes inside a live poll keeps its row and shows 0.0% with an empty bar — a zero is a genuine result and hiding it changes the story. If no votes have been cast at all, no division happens: shares read as a dash rather than 0.0% or NaN. Duplicate labels are kept as separate rows and flagged rather than merged, since merging would silently rewrite your numbers. Counts run to twelve digits and the allocation is done in exact integer arithmetic, so a poll with billions of votes is apportioned without rounding drift.
Getting your numbers in
You can type options and counts row by row, or paste the whole list at once — one option per line, with the label and count separated by a tab, a colon, a comma or just a space. Thousands separators pasted from a spreadsheet are handled, and any line that cannot be read is listed back to you rather than dropped without a word. When the table looks right, copy it as a plain-text results block, copy it as a Markdown table, or download it as CSV. Everything runs in your browser: no votes, labels or totals are sent anywhere.