Percentage Difference Calculator

Percentage difference compares two values with no reference point

Percentage difference answers one question: how far apart are these two numbers, expressed as a share of how large they are on average? It needs no baseline and no direction. Two lab readings, two supplier quotes, two salary offers: the gap is measured against the average of the pair, so swapping the order of the values changes nothing.

That neutrality is the reason to reach for this measure. Percent change needs a starting point, and it is not neutral. The same two numbers give different answers depending on which one you treat as the start, which is fine when a start is real, such as a price last year, and misleading when it is arbitrary. Percentage difference removes the choice by dividing the absolute gap by the average.

The limitation travels with the benefit. A percentage difference measures the gap relative to the pair, not relative to any outside standard. A 10 percent difference between 95 and 105 is 10 percent, and a 10 percent difference between 9,500 and 10,500 is also 10 percent, yet the second gap is 1,000 units instead of 10. Use the percentage to compare across differently sized numbers, and keep the underlying values beside it so the size of the gap is never lost.

The formula, worked in three steps

The calculation is:

Percentage difference = |A - B| / ((A + B) / 2) x 100

Step one takes the absolute gap, so the order of A and B never matters. Step two finds the average of the two values. Step three divides the gap by that average and multiplies by 100.

Take 110 and 100. The gap is 10. The average is (110 + 100) / 2 = 105. The gap over the average is 10 / 105 = 0.095238, and times 100 that gives 9.52 percent. Every number on this page comes out of that same three-step path, and the calculator runs it for whatever pair you type.

One case has no answer. If the two values sum to zero, the average is zero and the division breaks down. A pair such as 10 and -10 gives a gap of 20 and an average of 0, so the percentage difference is undefined, and the calculator says so instead of printing an infinity. Report the raw gap in that case, or shift both values by the same amount if a shared baseline makes sense for the comparison.

Percentage difference, percent change and percent error

Three related measures get mixed up often, and the fix is to look at what sits in the denominator. Percentage difference uses the average of the two values, so it needs no reference. Percent change divides by a starting value, so the direction of the comparison is built into the answer. Percent error divides by a known true value, which only makes sense when you have a standard to compare with.

Read across one row: the same pair of numbers gives 9.52, 10.00 and -9.09 depending on the question. Percent change carries a sign because it describes movement from a start, and it can be negative. Percentage difference does not, because it describes a distance. Percent error belongs to measurement work, where a reference value is known and you want to know how far an instrument read from it.

The NIST/SEMATECH e-Handbook of Statistical Methods and the NIST Physical Measurement Laboratory treat these comparison questions as measurement problems, with separate chapters on how close two results are to each other and on how far a result sits from a reference. That framing is useful for everyday use too: decide first whether you are comparing two peers or grading one reading against a standard, and the right denominator follows.

Why the average sits in the denominator

The average is the only base that treats both values as equals. Dividing by A alone makes B the measured quantity and A the yardstick, which quietly promotes one number above the other. Dividing by the geometric mean is another symmetric choice, and it is defensible for ratios, but it returns a different number for the same pair and is harder to explain to the person who asked the question.

There is a useful bound on the result. For two positive values the percentage difference runs from 0 to 200 percent. It reaches 200 only when one value is zero, since the average is then half the other value, and it hits exactly 100 when one value is three times the other: for 100 and 300 the gap of 200 sits over an average of 200. That bound is a sanity check, because a tool reporting 250 percent for two positive numbers has a formula problem rather than an unusual dataset.

The table below shows the same pair measured three ways, which is the quickest way to spot which base a report used.

The percentage difference sits between the two directional figures, which is a quick way to recognize it in a report. If a source quotes 20 percent for 80 against 100, it used 100 as the denominator. If it quotes 25 percent, it used 80.

Worked cases you can check by hand

Four situations where the measure earns its keep, each with the arithmetic written out so you can confirm the result.

The quote case is where a percentage difference saves an argument. A $840 gap sounds large until you see it is under 10 percent of the average quote, which is the point at which a second supplier is worth calling. The measurement case shows the opposite habit: a 5.4 g spread sounds negligible on a 500 g part, and 1.08 percent confirms that reading.

Rounding matters when you report the number. The calculator shows two decimal places because that is the precision the inputs usually support. If you write the figure into a report, round once at the end, and state the base you used so the next person can reproduce it.

Traps that trip people up

The first trap is mixing percentage points with percentages. A satisfaction score moving from 70 percent to 78 percent is a rise of 8 percentage points, which is also a relative change of about 11.4 percent, and the two are not interchangeable. Percentage difference sits between those framings: the same move is 8 divided by 74, which is 10.81 percent.

The second trap is a zero in the data. If either value is zero and the other is not, the percentage difference is a clean 200 percent, and that is a real result rather than a glitch. It tells you the gap is twice the size of the average, because the average is half the larger number. It stops being meaningful only when the two values cancel out.

The third trap is comparing percentages that came from different bases. If one source says a price rose 12 percent and another says 9 percent, and the two were computed against different references, neither number tells you the difference between the two prices. Recomputing both against a common base is the only fix.

The fourth trap is false precision. Reporting 9.5238 percent for 110 and 100 implies knowledge the inputs do not support. Two decimal places is enough, and for large values even one decimal is plenty.

A fifth trap is a borrowed definition. The phrase percentage difference is used loosely for percent change in many reports and blog posts, so when you meet a figure, check which denominator the writer used before you quote it next to your own.

What the result does and does not tell you

A percentage difference is a scale-free summary of a gap. It is the right number when you compare two options of different sizes, check whether a set of measurements agrees, or size up how much a bargaining position moves the price. It is the wrong number when you need the actual gap, when one value is a genuine reference point such as a target or a budget, or when the two numbers are not the same kind of thing.

Two habits make the number more useful. State the base in the sentence, not only in the calculation, so a reader knows the average was the denominator. And pair the percentage with the raw gap, because a 4 percent difference is a rounding detail on a small number and a large amount on a big one.

For the neighbouring tools, the percentage change calculator handles movement from a starting value, the percent error calculator measures a reading against a known true value, and the percent increase calculator and percent decrease calculator give the one-directional figures. When you need the middle value that the average represents, the average calculator computes it directly, and the percentage calculator covers the everyday part-of-a-whole work that sits next to this one.