Scope and assumptions
Percentage difference treats the two numbers as peers (no 'before/after'); use percent change when one is the baseline. Illustrative.
Two numbers with no time order between them — last year's sales versus this region's, your score against a peer's — don't have a "before" to measure change from. For those, percent change is the wrong tool. Percentage difference treats both as equals and splits the gap across their average, which is exactly why it always lands positive and reads the same whichever way you write the pair.
The symmetric core
Percentage Difference = |A − B| ÷ ((A + B) ÷ 2) × 100
The vertical bars mean "ignore the sign" (always positive). The base is the average of the two numbers, not either one — because when neither is "the original," picking one as the base would be arbitrary and would flip the answer around depending on order.
Seeing it in a table
| A | B | |A−B| | Average | % Difference | |---|---|---|---|---| | 40 | 60 | 20 | 50 | 20 ÷ 50 × 100 = 40% | | 100 | 120 | 20 | 110 | 20 ÷ 110 × 100 = 18.18% | | 50 | 50 | 0 | 50 | 0% | | 80 | 100 | 20 | 90 | 20 ÷ 90 × 100 = 22.22% |
Notice 100 vs 120 is a 20% change (base 100) but only an 18.18% difference (base 110) — two different questions with two different numbers.
Why the average, not one of the values
When neither number is "the original," choosing one as the base is arbitrary — 40 vs 60 would give 50% if you picked 40, or 33% if you picked 60. The average is the neutral, order-independent choice, so 40 vs 60 and 60 vs 40 are identical. That's the whole point: peer comparisons shouldn't depend on which you listed first.
Choosing difference over change
- Same product, two stores: no "before," use % difference (e.g., $40 vs $60 charger).
- Price last year vs this year: clear order, use % change.
- Two lab measurements: peers, use % difference to report spread.
- Target vs actual: a target isn't "before," but "how far from target" reads cleanly as % difference.
Comparison Errors in Percentages
- Using percent change by reflex — if there's no "before," the average base is the fix.
- Dropping the absolute value — difference is never negative; a sign means you subtracted backward.
- Dividing by the wrong number — base is (A+B)/2, not A or B alone.
- Forgetting the ÷ 2 — the average is (A+B)/2, not just A+B.
- Mixing it with percentage-point difference — different unit, different meaning.
Where to go deeper
This page anchors the family of advanced percentage operations. Follow a spoke for the case you need:
- Successive percentage change — Successive Percentage Change
- Percentage of a percentage — Percentage of a Percentage
- Ratio → percent — Convert Ratio to Percent
- Proportion → percentage — Proportion to Percentage
- Percentage error — Percentage Error Calculation
- Percentage uncertainty — Percentage Uncertainty
- Continuous growth — Continuous Percentage Growth
- Percentage points vs percent — Percentage Points vs Percent
- Reverse a percentage increase — Reverse Percentage Increase
FAQ
Can percentage difference exceed 100%? Yes. With 10 and 100, the gap is 90 and the average 55, so 90/55 × 100 ≈ 163%. It isn't capped at 100 the way a part-of-whole percent is, because neither number is a fixed total.
Is 'relative difference' the same thing? Not always. Some fields define relative difference as |A − B| ÷ min(A,B) or ÷ max(A,B), dropping the ÷2 average. That gives a different number, so check which convention your audience uses before quoting a figure.
What if the two values aren't really peers? The average base only makes sense when both numbers describe the same kind of quantity with no natural order. Comparing a grain of sand to a mountain with this formula produces a number that's technically valid but meaningless — reach for a ratio or log scale instead.
How many decimals should I report? Match your input precision. If the inputs are whole numbers, one decimal of % difference is usually honest; padding to 163.6363% implies a precision the data doesn't have.
Does it work on percentages themselves? Careful — comparing 40% and 60% as raw numbers gives 40% difference, but that's 40 percentage points apart and only a 50% relative change. Keep the units straight.
Sources: Khan Academy, Comparing quantities; Maths is Fun, Percentage change.
Disclaimer: Illustrative examples for learning the formula. Real statistical reporting may define "difference" differently by field; confirm the convention your audience expects. Informational only.
Sources and references
- Khan Academy — Comparing quantitieshttps://www.khanacademy.org/math
- Maths is Fun — Percentage changehttps://www.mathsisfun.com/numbers/percentage-change.html
Maintenance and corrections
The named author or publishing brand is responsible for this article. Worked examples are illustrative, not user testimonials, professional advice or approval by the institutions cited. Maintenance responsibility and review methods · Report an error