The salary-survey market has been running on the same product structure for roughly forty years. A consultancy collects payroll data from member companies, anonymises it, runs descriptive statistics, and re-sells the result back to the contributing companies as the "market view." Mercer and Korn Ferry are the two largest players in Europe. WTW and Aon Radford occupy the next tier. The economics of the model are unusual: the customer is also the supplier.
At SME scale, a typical license is €10,000 to €15,000 per year for a single country and a single industry vertical. For multi-country exposure or premium industry cuts the number scales accordingly — €25k–€60k is the operational corridor for a 150-person SaaS company that wants more than one geography.
The line-item anatomy of what that buys is worth looking at directly, because the gap between what the product is and what most companies assume the product is has widened. The Pay Transparency Directive — by requiring explanation, not just benchmark — has changed which side of that gap a survey product sits on.
The invoice, line by line
The €12,400 buys distributional summaries of someone else's payroll, a methodology document, six analyst hours, and a conference invitation. None of those line items are inaccurate to their price. What they are not is the same as what most companies actually need them to be.
Four things the license does not include
1. Per-row data
The survey is delivered as percentiles of a population. The individual payroll rows that make up the population are never disclosed. This is appropriate confidentiality — but it means the dataset cannot be queried directly, cannot be filtered by combinations of attributes that were not pre-computed, and cannot serve as a microdata source for any internal analysis.
2. Named-company comparators
The percentile says where the "market" sits. It does not say which companies contributed. The 75th percentile for a Senior Software Engineer in Lisbon might be made up of three contributing employers or thirty. Two participating companies cannot identify each other inside the dataset. Useful for confidentiality; less useful when the question is whether a specific peer is paying more.
3. Real-time data
The data has a reference date. The 2026 cycle uses payroll data collected mid-2025, processed Q3-Q4 2025, and released January 2026. By the time it is consumed, the data is 12–18 months old. In a market with 4–7% annual wage growth, the aging effect is non-trivial. The firms apply an "aging factor" to bring the data forward, which is itself a statistical estimate.
4. An answer to "why does this person earn this amount?"
This is the most consequential omission. The percentile says the market range. It does not say why a specific employee sits at the 60th percentile rather than the 40th. The pay-setting rationale required by Article 6 — and the contemporaneous decision record that Article 18 expects to be available — sit on the employer's side. The survey provides the band; the employer is responsible for the explanation of position inside it.
The product the survey is, and the product it is sold as
The diagonal matters. The bottom-left is what the product is. The top-right is what regulators increasingly read it as. The bottom-right is what the PTD has effectively asked employers to produce — and what no survey license, regardless of price, has ever been able to provide.
The economics of the model, briefly
The unusual feature of the salary-survey market is that the customer is also the data supplier. Mercer and Korn Ferry do not collect data through scraping or surveys of employees. They collect it from the participating employers' HRIS exports, in exchange for license discounts. A contributing company pays roughly half what a non-contributing one pays.
This creates a two-sided lock-in. The participating company is incentivised to contribute (to lower its license cost) and is also incentivised to keep using the product (because it has invested in mapping its roles to the framework). The cost of switching to another survey provider is the re-mapping cost. The cost of switching to no survey provider is reputational — comp committees expect "external benchmark" as a phrase to appear in board papers.
The product structure is hard to disrupt because the substitutes are not other survey products. The substitutes are categorical: either a different model of market reference, or a different question altogether.
None of this is a criticism of the survey product as a market reference. It is a useful market reference. The criticism — implicit in PTD's requirements — is that it has been sold and bought as a record of defensibility, which is not what it is. The €15,000 is appropriately priced for what it is. It is mispriced for what most companies were assuming they were buying.
The market reference, broken into two questions
The PTD-era version of the same question separates into two parts that the survey product folded together.
Part 1 — where does the market sit? Answered by salary distributions, increasingly from sources that do not require a €15,000 license: open salary data, recruitment platforms, public payroll disclosures, and employer-pooled real-data networks. The cost of this part of the answer has fallen by an order of magnitude in five years.
Part 2 — why does this person sit where they sit? Answered by the employer's own equal-value structure, pay-setting criteria, and decision records. This is what the PTD requires to be defensible, and what no external benchmark — at any price — can produce. The cost is internal: the architecture itself, not a subscription.
The €15,000 survey is appropriately priced for what it is. It is mispriced for what most companies were buying it to do.
Where the diagnostic starts
The first useful question is not "should we keep our survey subscription?" but "what work is the subscription currently doing that has nothing to do with what we actually need it for?"
A readiness diagnostic surfaces — by axis — what the employer's own pay-setting architecture contains and what is missing. The market reference question is the easier of the two. The defensibility question is the harder one, and it does not sit inside the invoice above.
The market reference is the easy half. The defensibility is the half that matters.
ReadinessCheck™ surfaces — by axis — where the employer's own architecture is in place, where it is partial, and where the rationale that PTD expects to be documented is still verbal. 20 minutes. No salary data.
Start the ReadinessCheck →