A percentage decrease tells you how much something shrank relative to where it started. Traffic fell from 1,600 visits to 1,240; a price dropped from $80 to $60; headcount went from 50 to 44. The absolute drop is simple subtraction — the percentage puts that drop in proportion, which is what makes a fall of 360 visits meaningful rather than merely large.
Subtract the new value from the old value, divide by the old value, then multiply by 100. As with an increase, the starting figure is the denominator, because it is the thing the change is being measured against.
When you need this
Reporting a decline honestly. Churn, falling traffic, a shrinking budget line. The percentage is what makes it comparable to previous periods and to other metrics.
Quantifying a price drop. A component that fell from $80 to $60 has dropped 25%. If your bill of materials has twenty such components, that percentage is what you multiply through.
Measuring efficiency gains. Page load time down from 4.8s to 3.1s is a 35% reduction. Error rate down from 2.4% to 0.9% is a 62.5% reduction — note that you can take a percentage decrease of a percentage, which is exactly where percentage points start to matter.
Worked examples
1,600 to 1,240. The difference is 360. Divide by the old value: 360 ÷ 1,600 = 0.225, so a 22.5% decrease.
80 to 60. The difference is 20. 20 ÷ 80 = 0.25, so a 25% decrease. Worth noting that going back the other way — 60 to 80 — is a 33.3% increase, not 25%. Same two numbers, different baselines, different answers.
50 to 44. The difference is 6. 6 ÷ 50 = 0.12, a 12% decrease.
Falling to zero. From 30 to 0 the difference is 30, and 30 ÷ 30 = 1, so that is a 100% decrease. A decrease can never exceed 100%, because you cannot lose more than all of something. If a calculation hands you a 140% decrease, you have either swapped the values or you are dealing with a figure that went negative — a profit turning into a loss, for instance — where percentage change stops being a sensible way to describe what happened.
Where people go wrong
Using the new value as the denominator. 360 ÷ 1,240 gives 29%, which is the increase needed to get back to 1,600, not the decrease that occurred. The two are never equal and the gap widens as the change gets bigger.
Expecting a matching increase to restore the original. It will not. Drop $100 by 20% and you have $80; raise $80 by 20% and you have $96, not $100. To reverse a 20% fall you need a 25% rise. This asymmetry is the single most useful thing to internalise about percentages, and it is why percentage rises and falls don’t cancel out.
Stacking successive decreases by adding them. Two consecutive 10% cuts leave you at 81% of the original, a 19% total reduction — not 20%. Multiply the factors (0.9 × 0.9 = 0.81) rather than summing the rates. The same logic governs stacked discounts.
Describing a percentage-point fall as a percentage fall. If a conversion rate goes from 8% to 6%, that is a fall of two percentage points and a 25% relative decrease. Reports that say “conversion fell 2%” when they mean two points are ambiguous at best. See percentage points versus percent.
The related calculators
If you know the starting figure and the rate and want the result — “what is $80 after a 25% cut?” — that is the decrease by a percentage calculator. If you know the discounted price and want to work backwards to the original, use the reverse percentage calculator.