How to Calculate Percentages — Four Questions, Not One

Four questions wearing the same word

Nearly every percentage problem is one of four questions, and naming yours is most of the work.

• What is 15% of 80? Multiply, then divide by 100: 80 × 15 ÷ 100 = 12.
• What percent of 80 is 12? Divide first: 12 ÷ 80 × 100 = 15.
• Something moved from 80 to 92, so how much is that up? Difference over the starting value: (92 − 80) ÷ 80 × 100 = 15.
• What does 80 cost with 15% off? Keep the rest: 80 × 0.85 = 68, and 12 comes off.

The same three numbers turn up in all four, which is exactly why they get swapped. Two habits keep them apart. First, say the sentence out loud before touching a keypad; the word "of" points at the number you multiply or divide by. Second, guess the size of the answer before you work it out. A percentage under 100 always gives you something smaller than the number you started from, so an answer larger than 80 in the first question means you divided where you should have multiplied. The percentage calculator here keeps all four boxes on screen for the same reason: you pick the question rather than a mode.

The answer depends on what the percentage is of

Every percentage is a percentage of something, and that something decides the answer. 20 is 40% of 50. Turn the same pair around and 50 is 250% of 20. Nothing changed except which number sits under the division line.

A percentage with no stated base is unfinished. The line "costs went up 30%" leaves out 30% of what: last month, last year, or the budget. The same gap is why two true statements can look like a contradiction. Getting 18 out of 24 is 75%; the same 18 answers out of 30 questions is 60%. The score did not move, the denominator did.

The same trap bites when percentages get averaged. A group of 10 people at 90% and a group of 40 at 60% do not average to 75%. Weight them: (10 × 90 + 40 × 60) ÷ 50 = 66%. Percentages can be averaged directly only when the groups behind them are the same size, and they rarely are. When somebody hands you a percentage with no denominator attached, the useful question is not whether it is a lot, but a lot of what.

Up 50% then down 50% does not bring you back

A 50% rise followed by a 50% fall does not return you to where you started, and it is the most expensive percentage mistake there is. Start at 100. Up 50% gives 150. Take 50% off 150 and you land on 75, not 100. Both steps used the same 50, but the second one measured it against a bigger number.

The rule generalizes: rise x% then fall x% and you finish below the start by x² ÷ 100 percent of it. Up and down 10% costs you 1%. Up and down 50% costs you 25%.

Recovery is lopsided too, in the direction people underestimate. Going from 150 back to 100 is a 33.3% drop, not a 50% one. Something that lost half its value has to double, a 100% gain, just to break even. A 25% loss needs a 33.3% gain; an 80% loss needs 400%. Whenever a percentage moves twice, check whether the base moved between the two steps. If it did, the two percentages are not measured on the same ruler, so adding or subtracting one from the other tells you nothing.

Percent and percentage point are different words

When the percentage itself changes, there are two honest ways to describe it and they produce different numbers. A rate that moves from 3% to 5% has risen by 2 percentage points, and it has also risen by 67%, because 2 is two thirds of 3. Neither is spin. They answer different questions: points measure the gap between two percentages, while percent measures the change relative to where the rate started.

The shorthand "pp" exists because "the rate went up 2%" is genuinely ambiguous. It could mean 3% to 5%, or 3% to 3.06%. In interest, unemployment, approval, market share and battery level, the difference is worth spelling out.

A quick way to keep them apart: subtract to get points, divide to get percent. From 3% to 5%, subtracting gives 2 points and dividing gives 5 ÷ 3 ≈ 1.67, a 67% increase. The smaller the underlying rates, the further apart the two readings drift. A move from 0.2% to 0.4% is a rise of 0.2 points and a rise of 100%, and both sentences are true, so the choice between them is itself a statement.

Discounts multiply, they never add

Stacked discounts do not add up. 30% off followed by another 20% off is not 50% off, it is 44% off. Multiply what is left after each step: 0.7 × 0.8 = 0.56, so you pay 56% and save 44%. On an item of 100 the path is 100 → 70 → 56.

Order makes no difference, because multiplication does not care about it: 0.8 × 0.7 is the same 0.56. That is worth knowing when a checkout applies coupons in an order you did not choose. Tax and fees join the same chain from the other side, so 30% off with 10% tax added is 0.7 × 1.1 = 0.77 of the original.

Working backwards is division. If you paid 56 after both discounts, the original was 56 ÷ 0.56 = 100. If you paid 68 after a single 15% cut, the original was 68 ÷ 0.85 = 80. It also shows why a discount above 100% is not a thing: at 100% off you already pay nothing, and past that the shop would be handing money over. The calculator on this page turns down rates outside 0 to 100 instead of printing a negative price as though it were a refund.

Doing it in your head

Two moves cover most mental percentages. 10% is the number with the decimal point shifted one place left, and 1% is two places. Everything else gets built from those. 10% of 46 is 4.6; 5% is half of that, 2.3; 1% is 0.46. So 15% of 46 is 4.6 + 2.3 = 6.9, and 18% is 4.6 + 4.6 − 0.92 = 8.28.

The second move is the flip. a% of b always equals b% of a, because both come out as a × b ÷ 100. Some questions are far friendlier reversed: 4% of 75 is awkward, but 75% of 4 is 3. 16% of 25 becomes 25% of 16, which is 4.

Round to a convenient percentage, then correct. 19% is 20% minus 1%; 24% is 25% minus 1%; a third sits near 33%. And keep a sanity check running: below 100% the answer is smaller than the number you started from, and 50% is exactly half. If you want the arithmetic itself to get quicker rather than merely correct, the mental math test here drills the same multiplications and divisions against a clock.

Why a spreadsheet shows a different percentage

A cell formatted as a percentage displays a number one hundred times larger than the value it holds, and that single fact explains most percentage bugs in a sheet. The value 0.1 shows as 10%. Apply percentage formatting to a cell that already contains 10 and it reads 1000%, because the format changed and the value did not.

Formulas read the value, not the display. Multiplying by a cell that shows 10% multiplies by 0.1, so dividing by 100 on top of that leaves you with a hundredth of the answer you wanted. Pick one convention per sheet: either store rates as decimals and format them as percentages, or store whole numbers and divide by 100 inside the formula. Never both at once.

Two more things look like errors and are not. Change is (new − old) ÷ old, so a row whose old value is 0 has no rate of change at all, and the cell is right to complain. And a column of percentages rounded to whole numbers often adds up to 99% or 101%, because the total is computed from the full values while the screen shows the rounded ones.