Market Median vs 75th Percentile: Choosing a Pay-Transparency Benchmark in 2026

The 50th percentile (median, or P50) is the midpoint of the market: half of comparable roles pay above it, half below. The 75th percentile (P75, upper quartile) is the point at which only the top 25% of comparable offers sit above it. Anchoring your band at P50 positions you as a market-competitive but not premium payer. Anchoring at P75 signals you are actively trying to win talent against the strongest competing offers, typically because the role is hard to fill or business-critical. The gap between the two is real money: for a Senior Software Engineer in Ireland, P50 sits at approximately EUR97k and P75 at approximately EUR110k, roughly a 13% difference at the same seniority level.

Most startups target P50 to P60 for standard roles and reserve P75 for roles where the hiring risk is highest: specialist engineering disciplines, roles with long open-vacancy periods, or positions where a bad hire is unusually costly. Targeting P75 across the board is rarely sustainable on a startup budget, and it is not necessary everywhere. A practical approach most PayMetric Labs client conversations converge on is tiering by role criticality: P50-P55 for roles with a deep, readily available candidate pool, and P65-P75 reserved for the handful of roles that are genuinely difficult to fill or disproportionately important to the roadmap.

Yes. The EU Pay Transparency Directive (2023/970), due for transposition into national law across the EU by 7 June 2026, requires employers to disclose a salary range in job adverts or before interview, but it does not mandate any specific percentile. The choice of P50, P60, P75, or any other anchor is a business decision, not a compliance one. What the directive does require is that the disclosed range be genuine and that pay within it be objectively justifiable, so a band that is technically wide but rarely honoured in practice, or one that does not reflect what the company actually pays, carries audit risk regardless of which percentile it claims to target.

Because compensation dispersion increases with seniority. At junior levels, most companies pay within a narrow band of each other since the work is relatively standardised and candidates have less individual negotiating leverage. At senior and staff levels, the market splits more sharply between companies competing on scarce specialist skill and companies hiring generalist seniority, which widens the spread between P25 and P75. Using PayMetric Labs' own Ireland benchmark as the worked example, the P25-P75 spread for Junior Software Engineers is roughly 32% of median, while for Lead/Staff Engineers it widens to roughly 23% of median in absolute EUR terms but the underlying salary range itself, EUR115k to EUR145k, spans a much larger cash gap than the junior band's EUR42k to EUR58k. A P50-anchored band that does not account for this will systematically underpay strong senior candidates relative to what the top of the market is actually offering.

At minimum every six to twelve months for most technology roles, and more frequently for specialist disciplines where the market moves quickly, such as AI/ML engineering. A band that was correctly anchored at P50 twelve months ago can silently drift toward P25 or below if the underlying market median moves and the band is not refreshed, even though nothing about the band's own numbers changed. This is one of the most common causes of an employer discovering they have fallen behind market only when a candidate declines an offer that would have been competitive a year earlier.

No. A single company-wide percentile target is a common early mistake. The right percentile depends on role scarcity, competitive intensity for that specific skill set, and how much the role's success genuinely depends on hiring from the very top of the market versus a broad, capable pool. Most mature compensation structures use a tiered approach: a default target (commonly P50-P60) for the majority of roles, with an explicit, documented exception process for roles where P75 or above is justified. Applying P75 everywhere is expensive and often unnecessary; applying P50 everywhere risks losing your highest-priority hires to competitors who differentiate their targeting.