TL;DR: Percent difference compares two values to their average, not to one of them. The formula is |a − b| ÷ ((a + b) ÷ 2) × 100. That makes it symmetric: the difference between 40 and 60 is the same percent whether you start at 40 or at 60. Percent change, by contrast, divides by the starting value, so it gives two different answers depending on direction. Choosing the wrong one is the most common error in this calculation.
The formula, and why the average is there
Percent difference answers the question “how far apart are these two numbers, relative to their typical size?” The typical size of two values with no preferred reference is their mean, so the mean goes in the denominator.
Worked example. Compare 40 and 60. The absolute difference is 20. The average is 50. So 20 ÷ 50 = 0.4, which is 40 percent.
Note that 20 is exactly half of 40 and a third of 60, so percent change would give 50 percent or 33.3 percent depending on direction. Percent difference collapses both into one figure because it has nowhere to put a direction.
The absolute value in the numerator is doing the same job: it removes order. If you swap the two numbers, the answer must not change, and without the absolute value it would.
Percent difference versus percent change
This distinction is the whole point of having two formulas, so it is worth stating precisely.
Percent difference is used when neither value is a baseline — two measurements of the same thing by different methods, two products, two candidates. The denominator is the average.
Percent change is used when one value is the starting point and the other is what happened later. The denominator is the starting value, and the sign carries meaning: an increase or a decrease.
The test is whether order matters. If swapping the two figures should change the answer, you want percent change. If it should not, you want percent difference.
A useful sanity check: percent difference can never exceed 200 percent, because the numerator can never exceed twice the average. If your calculation returns 300 percent, you used the wrong formula.
The other near-misses
Three more measures get confused with percent difference, and each answers a different question.
Percentage point difference is the simple subtraction of two percentages: 45 percent versus 52 percent is a difference of 7 percentage points. No division at all.
Relative difference uses one value as the reference rather than the average, which makes it identical to percent change with a different name.
Ratio expresses the comparison as a quotient without converting to a percent: 60 to 40 is a ratio of 1.5 to 1.
If the two values are percentages themselves and you are asked how far apart they are, the answer is almost always in percentage points, not percent difference. Reporting a rise from 45 to 52 percent as “15.5 percent difference” is arithmetically defensible and practically misleading.
Worked cases that show the trap
Two thermometers. One reads 98.6, the other 99.4. Difference is 0.8, average is 99.0, so percent difference is about 0.8 percent. Small, which is the right impression.
Two thermometers, one faulty. One reads 100, the other 150. Difference is 50, average is 125, so 40 percent. If you had used percent change from 100, you would report 50 percent, which overstates the gap relative to a neutral midpoint.
Two lab methods measuring the same sample. Results of 12.0 and 15.0. Difference 3, average 13.5, so 22.2 percent. This is the standard framing in method-comparison work, because neither result is privileged.
A price before and after. 80 rising to 100. Here order does matter, so the correct measure is percent change: 25 percent. Using percent difference would give 22.2 percent and understate the increase as experienced by the buyer.
That last case is the practical rule in one line: if one value is a baseline, percent change; if both are peers, percent difference.
Where percent difference is the right choice
The measure belongs in comparison contexts where no value is a reference.
Comparing two instruments or two methods measuring the same quantity. Comparing two candidates’ scores in an evaluation with no incumbent. Comparing two suppliers’ quotes for the same specification. Comparing an estimate against a model’s output when neither is treated as ground truth.
It is the wrong choice for anything with a baseline: growth over time, year-on-year change, before-and-after interventions, and discount calculations.
Rounding and presentation
Two decimal places is more precision than most comparisons support, and one is usually right. The bigger issue is consistency: if you report percent difference for one comparison and percent change for another in the same document, readers will assume they are comparable when they are not.
State which measure you used. A single clause — “expressed as percent difference from the mean” — prevents the entire class of misreading, and it costs six words.
If you are comparing several pairs and want to avoid formula slips, a percent difference calculator applies the average-based formula directly, which is worth using precisely because this is the calculation people most often do with the wrong denominator.
The short version
Percent difference is the gap divided by the average of the two values, times 100. It is symmetric, capped at 200 percent, and correct when neither value is a baseline. Use percent change when one value comes first, and use percentage points when comparing two percentages.