Five percentage mistakes that cost people money

Percentages look like the easiest arithmetic there is, which is exactly why the mistakes go unnoticed. Nobody double-checks a percentage, and the errors are systematic rather than random — they lean in one direction, and that direction usually costs money.

Here are the five that show up most often in real invoices, bills and negotiations.

1. Reversing a discount by adding it back

This is the most expensive one, and the most common.

A jacket is reduced by 40% to £60. What did it cost originally?

The instinct is to add 40% back: £60 × 1.4 = £84. That is wrong, and the error is large.

The discount was 40% of the original price, not of the sale price. If the original was x, then x × 0.6 = 60, so x = 60 ÷ 0.6 = £100. The real original was £100, not £84 — a £16 gap on a single item.

The gap grows with the discount:

At 50% off, adding the discount back understates the original by a quarter.

The rule: to reverse a percentage decrease, divide by (1 − rate). To reverse an increase, divide by (1 + rate). Division, not the opposite operation.

This matters in practice for working out whether a sale price is genuine, for reconstructing a pre-VAT amount from a total, and for any negotiation where someone quotes you a percentage off an unstated baseline.

2. Treating successive percentages as additive

Two 10% discounts are not 20% off.

Take £100. The first 10% brings it to £90. The second 10% applies to £90, not to £100, so it removes £9 rather than £10, leaving £81.

That is a 19% total saving, not 20%. Retailers understand this, which is why "an extra 10% off already-reduced prices" sounds better than it performs.

The same applies to increases, and the gap grows quickly:

That last row is worth pausing on. Two 50% cuts do not reach zero — they leave a quarter. Percentages of a shrinking number never reach zero however many times you apply them.

The rule: multiply the factors. A 10% rise then a 5% fall is 1.10 × 0.95 = 1.045, a 4.5% net rise. This also makes it obvious that order does not matter, since multiplication is commutative — a 20% discount then 20% tax gives the same total as tax then discount.

3. Adding a percentage then removing it and expecting to return

Related to the above, but it catches people separately because it feels like it must reverse.

Add 20% to £100 and you get £120. Take 20% off £120 and you get £96, not £100. You are down £4 despite applying the same percentage in both directions.

The reason is the same: the second 20% is calculated on a larger base. 1.20 × 0.80 = 0.96, always, for any starting value.

This appears constantly in tax calculations. If a total of £120 includes 20% VAT, the pre-VAT amount is not £120 minus 20%. It is 120 ÷ 1.20 = £100, with £20 of VAT. Taking 20% off £120 gives £96 and implies £24 of VAT, which is wrong in both figures.

4. Confusing percentages with percentage points

These are different units and the difference is often dramatic.

An interest rate rises from 4% to 6%. That is:

• A rise of 2 percentage points — the arithmetic difference.
• A rise of 50 percent — the relative change.

Both are true. They describe the same event and sound wildly different, which is why the choice between them is frequently editorial rather than mathematical.

The confusion is easy to exploit. "Costs rose 50%" and "costs rose 2 percentage points" describe the identical move from 4% to 6%, and the first will get considerably more attention. When you see a percentage change applied to a figure that is itself a percentage, work out which is meant before reacting.

The convention: use percentage points for the difference between two percentages, and percent for relative change. If a report is not explicit, assume nothing.

5. Averaging percentages with different bases

Percentages can only be averaged meaningfully when their denominators match. They frequently do not.

A student scores 1 out of 2 on the first test (50%) and 10 out of 10 on the second (100%). What is the overall percentage?

Averaging the percentages gives 75%. The actual figure is 11 out of 12, which is 92%.

The averaged answer treats both tests as equally important when one is five times larger. This is the same error behind a common business mistake: averaging monthly growth rates across months with very different starting revenue, and concluding something that is not true of the year.

The rule: to combine percentages, go back to the underlying numbers. Add the parts, add the totals, and calculate the percentage once from the combined figures. Only average percentages directly when the denominators are genuinely identical.

Two habits that prevent most of this

Convert to multipliers. A percentage is easier to reason about as a factor. 20% off is × 0.8. 15% VAT is × 1.15. A 40% increase is × 1.4. Once everything is a multiplier, chains become multiplication and reversals become division, and the errors above become obvious rather than subtle.

Sanity-check the direction and magnitude. A percentage under 100 makes a number smaller. Reversing a discount produces something larger than the sale price. If your answer moves the wrong way, or the size looks implausible, you have almost certainly used the wrong operation — which is nearly always the actual error, rather than the arithmetic.

Rounding, briefly

One more source of small persistent errors: rounding at every step.

Rounding to two decimal places after each operation compounds into a visible discrepancy across a long calculation. It is why an invoice total sometimes disagrees with the sum of its lines by a penny.

Carry full precision through the calculation and round once, at the end, to your currency's smallest unit. Where a total must exactly equal the sum of rounded lines — which is often a legal requirement on invoices — round each line and then sum those rounded values, rather than rounding the total of unrounded lines.

Checking your working

The percentage calculator here offers each of these as a separate mode, because choosing the wrong operation is the actual failure rather than the arithmetic. Every result shows the formula it used, so the answer can be verified rather than trusted — and the reverse-percentage modes show a check line confirming that the answer, run forward, produces your original figure.