What a real software price benchmark actually needs to include

"You're paying 20% above the market average" is a sentence that sounds convincing and, most of the time, means almost nothing — because a single average hides more information than it reveals.

A median, not an average

An average gets distorted by a handful of extreme contracts — one huge account that negotiated a crushed rate, or one tiny one paying full list price. The median (the middle value of the distribution) resists those outliers far better, and reflects what a "typical" company actually pays.

Two reference points, not one

The median tells you if a price is in line with the market. It doesn't tell you if that price is good. For that you need a second reference point — typically the first quartile, the price paid by the best-negotiating quarter of companies. The same price gap reads very differently depending on which one it's measured against.

A sufficient sample size, or nothing at all

Comparing a price against 4 or 5 similar contracts isn't a benchmark, it's an anecdote dressed up as data. A minimum threshold (15 comparable companies is a reasonable convention) has to be met before showing a gap at all — below it, the right answer is to say so clearly rather than invent a number.

The right level of comparison: plan against plan

The same software can cost 3x more depending on the pricing tier (Business vs. Enterprise, for instance). Comparing a price without knowing its plan means comparing different things under the same name — a common trap in benchmarks built too quickly.

Comparable criteria: company size and industry

The price a 200-employee healthcare company gets has little to do with what a 2,000-employee tech company pays. A real benchmark filters its comparison panel on both criteria before calculating anything — otherwise the resulting number compares situations that have nothing in common.