Percentage difference answers a subtly different question from percentage change, and using one where you meant the other is a common source of confusion in reports and comparisons.
Percentage change is directional: it has a before and an after, and it measures the change against the starting value. Percentage difference is symmetrical: it compares two values where neither is the baseline, and it measures the gap against their average. Two competing quotes, two sensor readings, two suppliers’ prices, two measurements of the same thing — none of those has a natural “before”, so dividing by either one would arbitrarily privilege it.
The formula is: take the absolute difference between the two values, divide by their mean, then multiply by 100.
When you need this
Comparing two independent measurements. Two labs report 4.7 and 5.1 for the same sample. Neither is the reference, so the percentage difference — 8.2% — is the honest way to describe the disagreement.
Comparing quotes or offers. Supplier A wants $8,400, supplier B wants $9,100. The difference is 8% of the average. Saying “B is 8.3% more than A” is also valid, but it makes A the baseline, which is a choice you should make deliberately rather than by accident.
Quality control and tolerance checks. When a specification says two readings must agree within 5%, percentage difference is almost always the intended measure.
Worked examples
4.7 and 5.1. The absolute difference is 0.4. The mean is (4.7 + 5.1) ÷ 2 = 4.9. So 0.4 ÷ 4.9 × 100 = 8.16%.
8,400 and 9,100. The difference is 700. The mean is 8,750. 700 ÷ 8,750 × 100 = 8%.
40 and 60. The difference is 20, the mean is 50, so the percentage difference is 40%. Compare that with percentage change: 40 to 60 is a 50% increase, and 60 to 40 is a 33.3% decrease. Three different, all correct, all answering different questions. This example is worth remembering precisely because the three numbers are so far apart.
Identical values. Two readings of 12 give a difference of 0 and a percentage difference of 0%, as you would hope. If both values are zero the result is undefined — you would be dividing by a mean of zero — and the honest answer is that percentage difference does not apply.
Where people go wrong
Using it when there is a genuine baseline. If you are comparing this month to last month, last month is the reference and you want percentage increase or percentage decrease. Percentage difference deliberately throws away the direction, which is a feature when comparing peers and a bug when tracking change over time.
Dividing by one of the values instead of the mean. That gives you percentage change, not percentage difference, and the two diverge quickly. For 40 and 60, dividing by the mean gives 40%; dividing by 40 gives 50%. Neither is wrong in isolation — they are answers to different questions — but calling the second one a “difference” invites misreading.
Reporting a direction. Percentage difference uses the absolute value, so it is always positive and carries no sense of up or down. If direction matters to your reader, you have picked the wrong measure.
Comparing values with opposite signs. If one figure is +30 and the other is −10, the mean is 10 and the “percentage difference” is 400%, which is technically computed but practically meaningless. Percentage measures behave badly around and across zero; use absolute differences there instead.
A note on percentage points
When the two values you are comparing are themselves percentages, be careful about how you describe the result. Two conversion rates of 3% and 5% differ by two percentage points, and by 50% relative to the average. Both are true and they will be read differently. Percentage points versus percent covers when each is appropriate.