It is tempting to add up a column of percentages and divide by the number of entries. Sometimes that works, but often it gives a wrong answer, and the error can be large. The fix is to remember what each percentage is a percentage of.
A percentage always describes a part of some whole. If the wholes differ in size, the percentages carry different amounts of information. Averaging them as if they were equal treats a small group and a large group as equally important.
Class A has 20 students and an 80% pass rate. Class B has 80 students and a 50% pass rate. The tempting calculation is (80 + 50) ÷ 2 = 65%.
Now work from the actual counts:
The correct overall pass rate is 56%, not 65%. Class B is four times bigger, so it should influence the result four times as much. The right method is a weighted average, with each percentage weighted by the size of its group.
You can reach the same answer without converting back to counts. Multiply each percentage by its group size, add the products, and divide by the total size: (80 × 20 + 50 × 80) ÷ 100 = (1,600 + 4,000) ÷ 100 = 56%.
When the groups are exactly the same size, the weights cancel and the simple average is fine. That is why the shortcut sometimes works and sometimes fails, and why it is so easy to trust it when you should not.
There are cases where you really do want every percentage to count equally. If three students each have a final grade percentage and you want the average of those three grades, there is nothing to weight. The key question is whether the percentages describe equal-sized wholes, or whether you are explicitly treating each one as one data point.
Percentages over days need careful counts. Suppose a student attended 18 of 22 school days in one month. That is 18 ÷ 22 = 81.8%. If the next month has 20 school days with 19 attended, the rate is 95%.
The overall attendance is 37 out of 42, which is 88.1%, not the simple average of 81.8% and 95%, which would be 88.4%. The difference is small here, but it grows when the months differ more. To avoid miscounting the school days in the first place, a day counter is a handy way to confirm the dates in a range, though you still need to remove weekends and holidays yourself.
Cashback offers are a common place for this mistake. Say you spend $400 on a card that pays 2% and $100 with a different card that pays 5%. The simple average of the rates is 3.5%, but your actual return is $8 plus $5, which is $13 on $500, or 2.6%.
The 2.6% figure makes sense because most of the money earned the lower rate. To see what a particular purchase returns, use a cash back calculator and then divide the total reward by the total spending.
Percentage changes behave in another surprising way. If a value rises 10% and then falls 10%, the average of the two changes is zero, but you do not end up where you started. Starting at 100, a 10% rise gives 110, and a 10% fall from there gives 99.
The reason is that the second percentage applies to a different base. For growth over several periods, multiply the growth factors instead of averaging the percentages. That is a separate method, but it is useful to recognise when a simple average will not do.
Whenever you need to combine percentages, follow this sequence:
Once you have totals, an average calculator can help check the arithmetic, but it cannot know whether the percentages should have been weighted, so the decision stays with you.
Percentages are not numbers you can combine blindly. Check what each one is a percentage of, convert back to totals where possible, and weight when the groups differ. A few seconds spent on this step prevents answers that look precise and are quietly wrong.