Salary Benchmarking: How to Use It Without Losing the Offer

A benchmark that lands 10% low costs you twice.

The candidate walks, and the placement you do eventually make is billed against a smaller number. Your fee is a percentage of a salary someone else anchored, and you helped them anchor it.

That second cost is the one firms never price in.

Recruiterflow’s modelling across 2,100+ firms puts revenue per placement as the single highest-leverage number in the business: lift it 10% and revenue per recruiter moves 5.7%, more than any conversion-rate improvement in the funnel (Source: The Economics of Recruiting).

This post covers salary benchmarking for recruitment firms: what it is, why published benchmark data is usually wrong for the specific search in front of you, the moment in the process where it kills offers, and how to use it so it does the opposite.

What salary benchmarking is

Salary benchmarking is the practice of comparing a role’s pay against market data for comparable roles, to decide what it should pay.

The inputs are usually published surveys, job board postings, aggregated self-reported data, and whatever the client paid the last person. The output is a range, and that range then shapes the brief, the screening conversation, and eventually the offer.

It is a useful discipline. It is also, in most of the forms recruiters encounter it, built on data that does not describe the market you are actually working in.

Why the number you pulled is probably wrong

The sample is not your market

Published salary data aggregates across company sizes, funding stages, and industries. A software engineer’s median is computed across a set that includes enterprises paying at the 90th percentile and eight-person companies paying in equity.

Your client sits somewhere specific in that distribution. The median tells you very little about where.

Titles do not mean the same thing

A benchmark keyed to a job title inherits every inconsistency in how companies title people. Vice President at a bank and VP at a thirty-person startup return the same row in a salary survey and describe entirely different jobs.

Seniority is a judgement about scope, budget and reporting line. A salary survey keyed to title cannot see any of those, and pay tracks them far more closely than it tracks what the job is called.

Base is not the offer

Most benchmark data reports base salary. Most candidates decide on the total: bonus, equity, pension, notice period, whether the role is in the office four days a week.

Two offers with identical bases can be twenty percent apart in what the candidate is actually being handed. Benchmarking base alone produces a number that neither side is really negotiating over.

The data is older than the market

Salary surveys report on the past. In a market where a specialism heats up over a quarter, a benchmark compiled last year describes a price that no longer clears.

The tell is simple: if your last three candidates in a discipline all rejected an offer inside the benchmark range, the benchmark is wrong, not the candidates.

The moment benchmarking loses the offer

It is rarely the number that loses the deal. It is when the number enters the conversation.

The failure pattern is consistent. The benchmark is never discussed properly at the start, so the client posts a range they assembled themselves. The search runs. Candidates are screened against that range, and the strong ones quietly self-select out. At offer stage, the client produces the benchmark as a defence of a number the candidate has already decided is too low.

At that point the data is being used to explain why the candidate should want less. Nobody has ever accepted an offer because a salary survey told them to.

Used at the end, a benchmark is a justification. Used at the beginning, it is a map. Same document, completely different outcome.

How to use it without losing the offer

Put the range in the intake meeting

The benchmark conversation belongs in the intake meeting, before a single approach is made, alongside the rest of the calibration on what the role actually is.

If the client’s range is below what the market clears, that is a conversation to have while it is still cheap. Six weeks later, with a candidate they want and a number they cannot reach, it costs the search.

Ask for the number behind the number

The range on the brief is almost never the ceiling. It is the range the client is comfortable saying out loud.

The question that gets the real one:

What would you pay for someone who is clearly the best person you have seen?

Most clients answer honestly, because the framing makes the higher number conditional rather than committed.

Record the answer against the job, not in someone’s inbox. A range that lives only in a recruiter’s memory is a range the firm loses when that recruiter is on holiday, and it is the kind of detail that never reaches the recruitment KPIs anyone reviews at the end of the quarter.

Use it to justify, never to cap

Benchmark data works when it is arguing for a candidate and fails when it is arguing against one.

“Three of the last five people we placed at this level went at 15% above your range, and two of them had less scope than this candidate.”

That is a benchmark doing useful work. It is specific, it is recent, and it comes from placements rather than a survey — the same case you’d make to a client to move a number up rather than hold it down.

Screening candidates out because they are above a published median, on the other hand, removes exactly the people the client would have stretched for.

The benchmark you already own

Every firm is sitting on better salary data than it buys.

Your placement history records what actually cleared: the offer that was accepted, at that level, in that discipline, in that city, at that company size, on that date. Published surveys report what companies say they pay. Your database records what candidates said yes to.

The gap between those two things is where most benchmarking goes wrong, and closing it needs nothing more than a query against work you have already done. Around 71% of placements come from candidates already in the database before the job opened (Source: The Economics of Recruiting), which means the salary history attached to those records is a picture of your own market, not a national average.

Most firms cannot run that query, because the compensation numbers sit in call notes and email threads rather than in a field anyone can filter. A database audit usually finds the data is there and simply unreachable.

Make your own placements the benchmark

Pull your last twenty placements in one discipline. Record the accepted package, the company size, and the month. That table beats any survey you can buy, because it is the only one built from offers that were actually accepted.

AIRA Search reads across the notes, calls and transcripts your team has already recorded, so the salary a candidate mentioned on a screening call is findable even when nobody typed it into a field.

Bring one open search to a demo and we will show you what your own placement data says the role should pay.

FAQs

What is salary benchmarking?

Salary benchmarking compares a role’s pay against market data for similar roles to decide what it should pay. The inputs are usually published surveys, job postings, and internal pay history; the output is a target range used to shape the brief and the offer.

How do you conduct salary benchmarking?

Define the role by scope rather than title, gather data from at least two independent sources, adjust for company size and location, and compare against what comparable roles actually closed at recently. The last step matters most, because accepted offers describe the market more accurately than advertised ranges.

Why is salary benchmarking data often inaccurate?

Most published data aggregates across company sizes and industries, keys to job titles that mean different things at different employers, reports base salary rather than total package, and describes a market that has since moved. Each of those introduces error in the same direction as the others.

How often should salary benchmarks be reviewed?

At least twice a year in stable disciplines, and every quarter in any specialism where you are seeing offers rejected inside the range. A run of rejections at benchmark is the clearest signal that the benchmark has drifted.

Should recruiters share benchmark data with clients?

Yes, at the start of the search rather than at offer stage. Presented during the intake meeting it calibrates expectations while changes are still cheap; produced at the offer, it reads as a reason the candidate should accept less.

What is the difference between salary benchmarking and compensation benchmarking?

Salary benchmarking looks at base pay, while compensation benchmarking covers the full package including bonus, equity, pension and benefits. Candidates decide on the second, which is why benchmarking base alone tends to produce offers that fail late.