What a share of the total actually is
An item's share is its value divided by the total, multiplied by 100. The only decision is what counts as the total. Leave the total field blank and the sum of the items becomes the total, which is right for a breakdown that is complete by construction: four sales channels that between them are all of the sales. Enter a total instead and the items are measured against something larger, which is right when the list is a subset. Four hundred people out of a thousand surveyed said yes; the four hundred is not the total, the thousand is, and the difference shows up as an unaccounted remainder rather than being quietly absorbed.
Getting that wrong is the most common error in a percentage breakdown, and it is invisible in the output, because both versions produce a tidy column of percentages. Only one of them answers the question that was asked.
Why the column adds to 99.9 and what is done about it
Round three equal thirds to one decimal and you get 33.3 three times, which is 99.9. Round 16.666 and 83.333 to whole numbers and you get 17 and 83, which is 100, by luck. The error is a property of rounding each number in isolation, and it does not go away with more decimal places, it just gets smaller and less visible.
When the items do add to the total, this page uses the largest remainder method, also called Hamilton apportionment: floor every percentage at the chosen precision, count how many units are missing from 100, and add one to whichever rows lost the most in the flooring. It is the same procedure used to allocate legislative seats to population shares. The result is the set of rounded numbers closest to the true values that still adds to exactly 100, and the column labelled "rounded on its own" shows what each figure would have been without it, so you can see how far anything moved.
| Item | Exact | Rounded alone | Adjusted |
|---|---|---|---|
| A | 33.333% | 33.3% | 33.4% |
| B | 33.333% | 33.3% | 33.3% |
| C | 33.333% | 33.3% | 33.3% |
| Total | 100% | 99.9% | 100% |
With three identical values there is no principled way to choose which row gets the extra tenth, and the method takes the first. That arbitrariness is real and it is why a table footnote reading "figures may not sum due to rounding" is a legitimate alternative to forcing the sum. Use the adjusted column when the total has to read 100, use the unadjusted column when each individual figure has to be defensible on its own, and do not do both in the same table.
Splitting an amount rather than measuring one
The second mode runs the arithmetic backwards. You give it an amount and a set of weights, and it divides the amount in proportion. The weights can be percentages that add to 100, or a plain ratio such as 3, 2, 1, or raw values like headcount or square footage; only the relative sizes matter. Splitting $1,000 five to three to two gives $500, $300 and $200, and splitting $100 three ways gives $33.34, $33.33 and $33.33 rather than three impossible amounts ending in a third of a cent.
The same largest remainder logic runs at the cent, so the shares always add to the amount you entered. That matters for anything that has to reconcile: a cost allocation across departments, a shared invoice, a rebate divided among accounts. A split that leaves the books a cent short is a split someone else has to fix.
Reading the output honestly
Percentages of a total flatten the size of the underlying numbers, which is what makes them useful and what makes them misleading. A category that grew from 2 percent to 4 percent doubled, and may have grown by eleven dollars. A category that fell from 40 percent to 35 percent may have grown in absolute terms while something else grew faster. This page keeps the raw values in the table beside the percentages for that reason, and if you carry only the percentages into a slide, carry the total with them.
Questions people ask
Why do my percentages add to more than 100?
Because the total you entered is smaller than the sum of the items. That is usually one of two things: a total typed from a different source than the item list, or items that overlap, where the same underlying thing has been counted in two categories. Overlapping categories genuinely can exceed 100 percent of a population, which is why survey questions allowing more than one answer are always reported with a note saying so. If the categories are meant to be exclusive, the arithmetic is telling you they are not.
Should I force the column to add to 100 percent?
It depends on what the table is for. If a reader will add the column, or if it feeds a chart where the segments have to fill the ring, force it and say in a footnote that shares were adjusted for rounding. If each figure will be quoted individually, or compared against the same figure computed elsewhere, do not force it, because the adjustment moves one or two rows away from their true value by the smallest displayed unit. What you should not do is force the sum silently and then let someone cite an adjusted figure as an exact measurement.
What format does the item list accept?
One item per line, with a name and a number in either order. Commas, colons, tabs, dollar signs and a trailing percent sign are all tolerated, so "Online: 1,250", "Online $1250" and "1250 Online" all parse the same way. A line with no number in it is ignored rather than treated as zero, and a line with only a number gets an automatic name. Negative values are dropped, because a share of a total is not defined for them and a mixed-sign list produces percentages that look plausible and mean nothing.
Can I use this to split a bill?
For an unequal split by weights, yes: put each person on a line with their weight and the bill in the total field. For an even split with a few people paying fixed amounts, the tip and bill-splitting pages on this site handle the rounding conventions that go with a restaurant bill, including who absorbs the leftover cents. This page always resolves the leftover by giving it to the largest fractional shares, which is right for an accounting allocation and slightly odd as a rule among friends.