Applying a percent is a multiplication
The fastest way to get percentage adjustments right is to stop thinking of them as "work out the slice, then add it on" and start thinking of them as a single multiplication. A 20% increase multiplies by 1.20. A 20% decrease multiplies by 0.80. A 7.5% sales tax multiplies by 1.075. Once you are working in multipliers, the arithmetic stops being fiddly and the traps disappear.
The reason is straightforward. Increasing by p percent means keeping everything you had, which is 100% of it, and adding p more hundredths: 100% + p% = (100 + p)/100 as a multiplier. Decreasing means keeping (100 − p)% of it. So a 15% discount leaves you paying 85% of the price, and 0.85 × the price gets you there in one step instead of two.
Multipliers also compose. If a product's cost rises 8% and then a 5% surcharge is added on top, the combined effect is 1.08 × 1.05 = 1.134, a 13.4% increase — not 13%. And because multiplication is commutative, applying the surcharge first gives exactly the same answer, which is a useful thing to know when someone insists the order matters. Order matters only when one of the adjustments is a fixed amount rather than a percentage.
The formula and what each part does
For a single adjustment applied n times:
new = old × (1 ± p/100)n
The bracket is the multiplier, and it is the only place the direction enters. Plus for an increase, minus for a decrease. Its value tells you immediately what will happen: above 1 the magnitude grows, exactly 1 leaves the value untouched, between 0 and 1 the magnitude shrinks, and at 0 the value collapses to nothing.
The exponent is the repeat count. Raising the multiplier to the power n is what compounding means: each application acts on the output of the last one, not on the original. Three successive 10% rises give 1.1³ = 1.331, so 33.1% overall, and the extra 3.1 percentage points over a naive 30% is the interest on the interest. The gap widens with both the rate and the count — ten 10% rises give 159.4%, not 100%.
The starting value carries the units and the sign. The multiplier is dimensionless, so dollars stay dollars and kilograms stay kilograms. If the starting value is negative, the multiplier still does the same arithmetic, but the visible direction reverses: an "increase" of 10% applied to −$400 gives −$440, which is a larger debt. The calculator flags this rather than letting the word "increase" mislead you.
Reversing one step is division, not subtraction of the same percent. The undo factor is 1/f, so undoing a 25% rise means multiplying by 1/1.25 = 0.8, a 20% cut. Undoing a 25% cut means multiplying by 1/0.75 = 1.333, a 33.3% rise.
Worked example: a $250 item, up 20%, and back again
A supplier raises a $250 part by 20%. Work it through, then reverse it.
- Convert the percent. 20 ÷ 100 = 0.20.
- Build the multiplier. An increase, so 1 + 0.20 = 1.20.
- Apply it. 250 × 1.20 = $300.
- Read off the change. 300 − 250 = $50, which is also 0.20 × 250. Both routes agree.
- Reverse it. To get from $300 back to $250, multiply by 1/1.20 = 0.8333, a cut of 16.67%. A 20% cut from $300 would land at $240, which is $10 too low — the classic error.
Now compound it. Suppose the supplier applies the same 20% again next year, and again the year after. The multiplier is 1.20³ = 1.728, so the part reaches 250 × 1.728 = $432, a total rise of 72.8% rather than 60%. Year by year: $300, then $360, then $432. Each year's $ increase is bigger than the last — $50, $60, $72 — because the same rate is being applied to a bigger base every time.
The same arithmetic runs backwards for depreciation. A $20,000 vehicle losing 20% of its value each year is worth 20,000 × 0.8⁵ = $6,553.60 after five years, having lost 67.2% of its value in total. Notice it never reaches zero: a fixed percentage decline approaches zero without arriving, which is exactly why tax codes use straight-line schedules for assets that genuinely reach scrap value.
Reading the multiplier
The multiplier is the number to sanity-check. If it is 1.075 you have added 7.5%; if you meant 75% you have slipped a decimal place. A multiplier between 0 and 1 must shrink the magnitude and one above 1 must grow it, so a "discount" that moves the value further from zero means the direction control is wrong.
For the total percent change, compare it against the simple product of rate and count. Apply the same non-zero percent more than once and a compounded increase exceeds n × p, while a compounded decrease lands above −n × p — that is, it removes less in total than repeated simple subtraction would. At a single application the two figures are the same by definition. Four 25% cuts leave 0.75⁴ = 31.6% of the original standing, not zero. This asymmetry is not a rounding artefact; it follows from the fact that each step acts on a smaller base than the last.
A 100% decrease is the boundary case. The multiplier is exactly 0, the value goes to zero, and no percentage adjustment will ever move it again. Beyond 100% the multiplier goes negative, which flips the sign of the value at every application. Arithmetically that is fine, but in almost every practical setting a "120% reduction" means someone has computed a percentage against the wrong base — usually cost when they meant selling price. If your figures are prices, check the definition with the percentage calculator before applying it.
Finally, resist quoting the reverse percentage as if it were the forward one. They differ by a factor of f, and the difference is large for large rates: a 50% rise is undone by a 33.3% fall, and a 50% fall by a 100% rise.
Compounded multipliers by rate and number of applications
| Adjustment | ×1 | ×2 | ×3 | ×5 | ×10 |
|---|---|---|---|---|---|
| +20% | 1.20000 | 1.44000 | 1.72800 | 2.48832 | 6.19174 |
| +10% | 1.10000 | 1.21000 | 1.33100 | 1.61051 | 2.59374 |
| +5% | 1.05000 | 1.10250 | 1.15763 | 1.27628 | 1.62889 |
| −5% | 0.95000 | 0.90250 | 0.85738 | 0.77378 | 0.59874 |
| −10% | 0.90000 | 0.81000 | 0.72900 | 0.59049 | 0.34868 |
| −20% | 0.80000 | 0.64000 | 0.51200 | 0.32768 | 0.10737 |
Read the −20% row across: ten annual falls of a fifth leave 10.7% of the original value, not zero. Read the +10% row: ten rises of a tenth multiply the value by 2.59, well above the 2.00 that adding the rates would suggest.
Where percentage adjustments go wrong
- Undoing a rise with the same percent. A 20% rise needs a 16.67% cut to reverse, because the cut is measured on the larger figure. Divide by the multiplier instead of subtracting the rate.
- Adding rates that should be multiplied. A 10% rise followed by a 10% rise is 21%, not 20%. A 10% rise followed by a 10% cut is −1%, not zero.
- Confusing markup with margin. A 50% markup on a $100 cost gives a $150 price, on which the margin is 33.3%. The markup base is cost; the margin base is the selling price. Two different bases, two different numbers for the same $50.
- Applying a discount to a tax-inclusive figure. Order does not matter between two percentages, but it does matter which base each percentage was defined against. Discounts are usually defined on the pre-tax price.
- Treating a percentage decline as reaching zero. Repeated percentage falls approach zero asymptotically. Anything that genuinely hits zero on a schedule is a fixed-amount reduction, not a percentage one.
- Ignoring the sign of the starting value. "Increase by 10%" applied to a negative balance makes it more negative. The word describes the multiplier, not the direction on the number line.
Related tools and where each one belongs
This calculator goes forwards: you supply a rate and it produces the adjusted value. When you already have two values and want the rate that connects them, that is the inverse problem — use the percentage change calculator, which divides the difference by the earlier figure and also reports the growth factor you would need here.
For a static slice of a fixed amount — a tip, a commission, a share of a budget — nothing is being adjusted, so reach for the percentage calculator instead. It solves the same base-100 identity for whichever of part, percent or whole you are missing.
When the adjustment is defined by a ratio rather than a rate — scaling a recipe by 3/2, converting at a map scale, adjusting a dose per kilogram — set it up as a proportion with the proportion solver. And when you need the rate that produces a known total over several periods, you are taking an nth root of the overall multiplier, which the exponent calculator does with a fractional exponent: the annual rate behind a 33.1% three-year rise is 1.3311/3 − 1 = 10%.
Percentages that stay rational are often easier to check as fractions. A 12.5% cut multiplies by 7/8; a 33⅓% rise multiplies by 4/3. The multiplying fractions calculator chains several such factors exactly, with no rounding drift at all, which is worth doing when a long sequence of adjustments has to reconcile to the penny.