Why a Percentile Is Not a Percent — and Why That Matters in Compensation

In the world of compensation, it’s common to hear phrases like:

But here’s the problem:

Too many people treat percentiles like percentages, or assume they behave like standard deviations from a mean. That misunderstanding can lead to confusion, unrealistic pay expectations, and flawed decision-making.

Let’s break down why this matters — and how to use percentiles correctly when making compensation decisions.

So, What Is a Percentile?

A percentile is a rank or position in a dataset. It tells you the value below which a certain percentage of the data falls.

• The 50th percentile (aka the median) means half the data is below that value.
• The 90th percentile means 90% of the data points fall below that value.

Notice: it’s not about how much higher the 90th is than the 50th — it’s just saying where it lands in the distribution.

Why This Causes Confusion

Let’s say you’re looking at software engineer salaries:

Now compare that to another dataset:

Both are “paying at the 90th percentile” — but one implies a 2% increase over median, and the other implies a 50% increase.

That’s a massive difference in magnitude, and it’s why relying solely on percentiles without looking at actual numbers can be misleading. If your dataset is small or narrowly sliced, your percentile spreads may be extremely tight — leaving leaders, recruiters, or candidates wondering why the “90th” looks so… average.

But Wait — What About Standard Deviations?

In statistics, there’s a known relationship between percentiles and standard deviations if the data follows a normal distribution (the classic bell curve):

• ~68% of data falls within 1 standard deviation of the mean (roughly the 16th to 84th percentile)
• ~95% within 2 SDs
• ~99.7% within 3 SDs

But most compensation data isn’t perfectly normal. It’s skewed by factors like equity, signing bonuses, job title inflation, or small sample sizes. That’s why treating percentiles as proxies for standard deviation can get you into trouble.

Real-World Implications for Compensation Strategy

Let’s go back to that 90th percentile pay target.

If you don’t check the actual values behind that target:

• You may offer far less than the market expects — despite paying at the “90th.”
• You may blow your budget trying to hit a percentile that’s based on inflated outliers.
• You may create internal confusion when percentiles don’t match expected percent differences.

That’s also why we usually don’t share percentiles with employees or candidates. It's a complex concept, and without understanding the data behind the number, percentiles can be more misleading than helpful.

Best Practices for Using Percentiles in Compensation

1. Look at the actual values, not just the percentile ranks. Understand the spread.
2. Use wide enough data cuts to avoid artificial compression between percentiles.
3. Avoid sharing percentiles externally unless you can explain the underlying dataset clearly.
4. Don’t assume percentiles = percent differences — they rarely do.
5. Double-check your expectations: if leaders expect a 90th percentile value to be “way above market,” make sure the data supports that.

Final Thought: Percentiles Require Context

Percentiles are a powerful tool in compensation — but only when used with the right context.

They’re not percentages. They’re not linear. And they’re not stand-ins for magnitude.

So the next time someone says, “Let’s pay at the 90th,” don’t just nod. Ask:

What does that actually mean — in dollars? In dataset size? In impact?

Because when it comes to pay, precision matters.