Benchmarking data is easy to buy and easy to misuse. The failure mode is consistent: a percentile figure is taken as a fact about the market rather than a summary of a sample somebody else assembled for a different purpose.
Three things are usually missing. The first is matching. Surveys match on job title, and titles are not comparable across ownership models. A head of finance in a founder-owned business and a head of finance in a fund-backed platform share a title and almost nothing else.
The second is composition. Base salary is the smallest part of the decision in many of the markets we work in. Bonus deferral, long-term incentive, carried interest and the vesting a candidate walks away from are what determine whether an offer is accepted. Benchmarking base alone produces offers that are expensive on paper and declined in practice.
The third is timing and context. Compensation data describes the past. If two competitors have restructured since the data was gathered, or a fund has closed and is building a team, the market has moved and the table has not.
A usable benchmark therefore states its sample size for every figure, matches on remit rather than title, includes total reward, and explains what drives the range: scarcity, ownership, location, or the scope of the role itself.
The practical consequence of getting this wrong is not only cost. It is a year of declined offers, an internal pay structure distorted by one exception made under pressure, and a retention problem created by the correction that follows.
Base salary is the smallest part of the decision in many markets. Benchmarking it alone produces offers that are expensive and still declined.
