A recruitment agency or consulting firm director often opens their activity report on a single line, the number of placements this month. The figure is reassuring, it feeds the annual targets and compares one consultant to another. It stays silent on what matters, though: how long each role tied up the team, where the candidates who signed came from, whether the hires last past the probation period, and whether you start from a blank page on every assignment or build on the interviews already run.
This article is for the leader who wants to steer beyond volume, whether you run a recruitment agency or an IT services firm. It walks through five metrics in the order they read, each with a definition, a market benchmark where one exists, and a way to calculate it without dedicating a full-time role to it. The last one, the talent-pool reuse rate, is the one almost nobody tracks, and it is often the most revealing.
Three rules before you put down a single figure
A badly defined metric steers you worse than a missing one, because it gives a false impression of control. Three precautions apply to the five indicators that follow.
First, write down the numerator and the denominator in black and white. "Time to fill" means nothing until you have fixed the starting point (role opened or job posted?) and the end point (offer accepted or actual start?). Two teams measuring different things cannot be compared.
Next, prefer the median to the mean for anything that is a duration. Recruitment durations are stretched upward by a few roles that drag on for months, and a mean lets itself be pulled by those extreme cases. The median, the point that splits your hires into two halves, describes the team's daily reality better. It is, in fact, the measure France Travail uses in its statistics on filled job postings.
Finally, segment by role type. A global time to fill aggregates managers and junior profiles that have nothing in common. France Travail reports a median of 67 days to hire a manager or supervisor in 2023, against 37 days for an unqualified employee, a thirty-day gap that the overall mean erases.
Metric 1 · Time to fill
This is the time that passes between the moment a role is opened and the moment a candidate accepts. It is distinct from time to hire, which starts from the application of the future hire and stays shorter, since it ignores the period when the posting is still looking for takers. For an agency, this duration is also your sales cycle, every week gained on a fixed-fee assignment weighs directly on the margin.
To measure it, date the role opening and the accepted offer for each role, calculate the median by job family, and track its trend month after month rather than in absolute value.
On the market benchmark side, France Travail reports for 2023 a median of 47 days across all the job postings filed with it, half of hires completing in less time. It is a median across all profiles, and it rises to 67 days as soon as you restrict it to managers and supervisors, as seen above. One and the same indicator, two scopes, two values: proof that a duration can never be read without specifying what it covers.
Metric 2 · Cost per hire
Add up everything a hire consumes, the direct costs (job advertising, tool licences, any fees) and the indirect costs, often the heaviest and least tracked, the time recruiters and managers spend on sourcing, interviews and qualification. For a recruitment agency or an IT services firm, it is your internal production cost of a placement that you have to isolate here, distinct from what you bill the client.
Two benchmarks help place the order of magnitude, keeping in mind that they describe the point of view of the company hiring for itself. On the direct-employer side, some estimates put the average cost of a hire in France between 5,000 and 8,000 euros for a standard profile, a range that does not always reflect all indirect costs (vacant role, team overload, temporary drop in productivity), so the full cost can exceed it markedly. On the services-market side, engaging an agency generally represents 15 to 25% of the candidate's gross annual salary, around 18 to 20% on average and 25 to 35% for executive search. That last percentage is the price paid by the client, not the agency's internal cost, and confusing the two distorts any profitability calculation.
A sustainable calculation comes down to four steps, once a quarter: estimate the average time spent per role (recruiter hours plus manager hours), value it at a loaded hourly rate, add the direct costs, and divide by the number of placements. An order of magnitude tracked over time beats an exact calculation done only once.
Metric 3 · Source of hire
This metric answers a simple question, where do the candidates who sign come from. Job posting, referral, direct approach, network, internal talent pool, each channel has a different yield, and tracking it avoids continuing to fund a channel that does not convert.
A classic trap is to look only at the volume of applications per channel. The APEC offers a more accurate reading, a channel's effectiveness is calculated by relating the share of candidates hired through it to its usage rate. A heavily used channel that produces few hires is less effective than a discreet channel that converts strongly. To place usage, the APEC reports that publishing a posting and mobilizing the network were, for 80% of companies, the two channels that most often made it possible to finalize a manager hire in 2023. The point to keep for later, the APEC counts the application pool (former candidates, profiles in reserve) among the channels in its own right, which most teams forget to measure.
In practice, at each placement, record the real originating channel of the signed candidate, then calculate for each channel the ratio between its share in hires and its share in the effort put in. It is this ratio, and not the raw volume, that tells you where to reinvest.
Metric 4 · Quality of hire
This is the most useful and most elusive metric, because there is no single market standard to measure it. The ISO/TS 30411:2018 standard offers a framework, assessing quality either by the new hire reaching their performance objectives, or by the fit of their skills observed after hiring. Lacking a common reference, a widespread approach on the HR side combines two horizons, a short-term composite (time to hire, candidate conversion rate, candidate and manager satisfaction) and a long-term composite (probation validation, retention, performance sustained over time).
On implementation, choose two or three signals you can actually collect, for example the six-month probation validation rate, one-year retention, and a manager satisfaction score gathered in a single question. An imperfect proxy tracked regularly beats a perfect index never filled in.
At market scale, France Travail gives a satisfaction benchmark, when a posting is filled, 88% of employers say they are satisfied with their hire, a rate that rises to 93% for a hire ultimately made internally. And quality is decided upstream, the same study reports that 63% of establishments that hired ran into difficulties, of which 93% point to applications, too few (73%) or unsuitable (77%). Measuring quality also means watching the raw material entering the funnel.
Metric 5 · The talent-pool reuse rate
Here is the metric almost no agency tracks, even though it sheds light on the other four. The talent-pool reuse rate measures the share of your placements that come from candidates already present and already assessed in your database before the role opened, as opposed to those from cold sourcing restarted from scratch. The calculation comes down to one division, placements from the internal talent pool over total placements, over a given period.
What we observe in the agencies and IT services firms we work with, without making it a market statistic, matches the order of magnitude described in our reactivation method: a rate often between 5 and 15% when nothing is done to maintain the database, against 30 to 45% achievable once interview data is structured and reusable from one assignment to the next. These candidates are indeed present in the database, but hard to find again, their assessments barely filled in, their labels unused.
The talent pool is not used at 20%. Recruiters reinvent the wheel with every new role.
This metric is a leading indicator. When it rises, time to fill falls (a candidate already met positions themselves faster), cost per hire falls (less cold sourcing), and channel effectiveness improves mechanically since the talent pool becomes a productive channel again. That is why we place it fifth but treat it as the first to move. To understand why this database stays inert in most organizations, see why the talent pool sleeps, and for the step-by-step approach that starts precisely by measuring this database, see reactivating your candidate talent pool.
Instrumenting all this without a Rube Goldberg machine
These five metrics have one thing in common, they assume clean, up-to-date data gathered in one place. That is where the shoe pinches, the information is scattered across the ATS, emails, tracking files and interview notes left in free text. Even at large companies, tooling stays partial, the APEC notes that in 2024 only half of them use recruitment software of the ATS or HRIS type to hire managers, up 12 points since 2021. And the use of AI tools for manager recruitment stays marginal, 4% of the companies concerned in 2024, 11% considering adopting it. In other words, most teams still measure by hand, through exports and spreadsheets.
The first three metrics, time, cost and source, can be drawn from a well-kept ATS, at the price of a few exports. The last two, quality and talent-pool reuse, hit a gap, they need structured interview data, which the ATS almost never captures, since a report most often arrives there as unusable free text.
That is the role of a Hub connected to your ATS, without replacing it. Interview data is structured as soon as the conversation ends and sent back into your ATS candidate profile bidirectionally, which feeds source of hire and makes the talent pool measurable. The database becomes searchable in natural language ("find me the SOC profiles available in Paris already met this year"), with no percentage score and with explainable criteria, which is what the AI Act expects. Transcription is only the entry point, the value lies in the data finally made reusable, and therefore measurable.
Frequently asked questions
Which recruitment metric should you track first?
Start with the three metrics a well-kept ATS already gives you: time to fill (median, by role type), cost per hire and source of hire. Then add quality of hire through two or three simple signals, then the talent-pool reuse rate, which is the most useful and least tracked leading indicator. Five imperfect metrics followed every month beat a perfect dashboard that is never filled in.
Should you use the mean or the median for time to fill?
The median, for any duration. Recruitment durations are pulled upward by a few very long roles, and the mean lets itself be distorted by those extreme cases. The median, which splits your hires into two halves, describes the common case better, and it is the statistic France Travail uses: a median of 47 days in 2023 across all job postings, rising to 67 days for managers and supervisors alone. Also look at the spread, your slowest roles, not just the central point.
How do you calculate the talent-pool reuse rate?
Divide the number of placements from candidates already present and already assessed in your database before the role opened by the total number of placements over the same period. The tricky part is tracing the real origin of the signed candidate, which requires a maintained database and candidate profiles carrying usable interview data. This rate stays low as long as the database is not maintained, and rises markedly once interview data is structured and reusable from one assignment to the next.
Does cost per hire include the agency fee?
No, not for a recruitment agency or IT services firm hiring for its clients. The 15 to 25% of gross annual salary billed is a selling price, not a cost. An agency's cost per hire is its internal production cost, essentially the consultants' time spent on sourcing, interviews and qualification, plus tools and job advertising. For a company hiring for itself, on the other hand, any agency fees do belong in its cost of hire.
Key takeaways
- The number of placements is a lagging indicator, it steers nothing on its own, frame it with the five metrics.
- For durations, think in medians and by role type, the mean masks the spread and the roles that drag on.
- Separate the internal production cost (agency or IT services firm) from the price billed to the client, the 15 to 25% is not a cost.
- Quality of hire has no standard, choose two or three signals you can actually collect rather than a perfect index never filled in.
- The talent-pool reuse rate is the least tracked leading indicator, it pulls time and cost down when it rises.