Recruitment KPIs for agency desks: the numbers that predict billings
The recruitment KPIs that matter on an agency desk follow one funnel: jobs taken on, CVs sent, first interviews, offers, placements, fees billed. Count each stage weekly, watch the ratios between neighbouring stages monthly, and you can see a thin quarter coming weeks before it shows up in the billings.
- Most published recruitment metrics (cost per hire, time to hire, quality of hire) are written for in-house teams. An agency needs funnel counts and the ratios between them.
- Three kinds of KPI: activity (what you did), conversion (how well it worked), outcome (what you billed). Manage weekly on the first, coach on the second, pay on the third.
- Set targets backwards from the billing number. The calculator below turns an annual target into CVs to send per week.
Why agency recruitment KPIs differ from in-house metrics
Search for recruitment metrics and you mostly get advice for corporate talent teams: cost per hire, time to hire, quality of hire, offer acceptance. Those measure a cost centre. An agency desk is a sales pipeline whose product happens to be people, so it needs pipeline numbers.
The in-house benchmarks are still worth knowing, because they describe your client's side of the table:
Sources: SHRM 2025 Recruiting Benchmarking data brief (2,371 US respondents; medians; time to fill is 45 days for executive roles) and the CIPD Resourcing and talent planning report 2026 (1,014 UK respondents; cost medians come from a subsample of 223 to 263 who could estimate within ±20%, and include agency fees where paid).
Two things follow for an agency. First, the client's 44 days is the envelope you work inside, and you control only the front of it. Second, roughly seven in ten UK employers in that CIPD survey don't measure time to hire, so an agency that reports its own speed and ratios is often handing the client the only hiring data they have. The same report found 41% of organisations that hired had sourced through a recruitment agency. Your numbers are part of theirs.
The agency funnel, from jobs on to fees billed
Every permanent desk runs the same six stages. Here is one quarter on an example desk at a fictional agency, Northwind Talent. Bar widths are to scale, which is why the bottom looks alarming. It always does.
The funnel gives you the vocabulary for everything else. "CVs sent" is what US staffing calls submittals, so CV-to-interview ratio and submittal-to-interview ratio are the same measure. "Placements" means starts, not accepted offers; the difference between the two is where a lot of forecast billing quietly disappears.
Activity, conversion and outcome KPIs
Recruiter KPIs come in three kinds, and most arguments about KPIs are really arguments about which kind to manage on.
What you did
Jobs taken on, CVs sent, interviews arranged, client meetings. Fully in the consultant's control and visible this week. Easy to inflate, so never use alone.
How well it worked
The ratio between two neighbouring stages. This is where skill shows up, and where coaching should point. Needs a few weeks of data before it means anything.
What you billed
Placements, fees, gross profit, repeat clients. The only numbers the business banks, and the slowest to move. By the time they dip, the cause is two months old.
A workable rule: weekly meetings look at activity, monthly reviews look at conversion over a rolling window, and commission is paid on outcome. Agencies that manage on activity alone get plenty of CVs sent and a falling CV-to-interview ratio. Agencies that look only at billings find out about problems when it is too late to fix the quarter.
Recruitment metrics and formulas, in one table
Definitions matter more than targets. Pick one definition per metric, write it down, and make sure the whole team counts the same way. The worked examples use the desk above.
| KPI | Formula | Example desk | Kind |
|---|---|---|---|
| CV-to-interview ratio submittal-to-interview | CVs sent ÷ first interviews | 96 ÷ 32 = 3 : 1 | Conversion |
| Interview-to-offer ratio | First interviews ÷ offers | 32 ÷ 10 = 3.2 : 1 | Conversion |
| Offer-to-placement rate offer acceptance | Starts ÷ offers made × 100 | 8 ÷ 10 = 80% | Conversion |
| Interview-to-placement ratio | First interviews ÷ placements | 32 ÷ 8 = 4 : 1 | Conversion |
| Fill rate | Jobs filled ÷ jobs taken on × 100 | 8 ÷ 24 = 33% | Conversion |
| CVs per job | CVs sent ÷ jobs worked | 96 ÷ 24 = 4 | Activity |
| Time to shortlist | Date shortlist agreed − date job taken on | median 3 working days | Speed |
| Time to submit | Date first CVs sent − date job taken on | median 4 working days | Speed |
| Time to fill | Date offer accepted − date job taken on | median 34 days | Speed |
| Fee per placement | Fees billed ÷ placements | £72,000 ÷ 8 = £9,000 | Outcome |
| Billings per consultant | Fees billed ÷ fee-earning headcount | £72,000 a quarter | Outcome |
| Client repeat rate | Clients billed this year who were also billed last year ÷ clients billed last year × 100 | 14 ÷ 22 = 64% | Outcome |
| Fall-off rate drop-outs in rebate period | Placements that left inside the rebate period ÷ placements × 100 | 1 ÷ 8 = 12.5% | Outcome |
Three counting rules save a lot of arguing later. Count first interviews only in the ratios, or a client who runs four rounds will flatter your numbers. Count a placement on the start date. And for fill rate, decide what happens to jobs the client cancels or fills internally: leaving them in the denominator is harsher and more honest, since you spent the hours either way.
Very little that is independent. One of the few published data points is a Bullhorn survey of 1,400 recruitment professionals, in which 40% of firms said they submit three to four candidates for every hire, 24% four to five, and 14% one to two (Bullhorn, drawing on its 2018 trends survey). Self-reported figures like this tend to describe jobs that filled. Count every CV on every job, including the ones you lost, and a contingent desk will sit well above that. Your own trailing twelve months, split by client and by consultant, is the benchmark that counts.
What a bad ratio is telling you
A ratio on its own is trivia. Its value is that each one points at a specific conversation in the process. When a number moves, this is where to look first.
High CV-to-interview ratio
The client is rejecting on paper. Either the brief was taken thinly and you're guessing, or the right people are being presented badly: raw CVs, no summary, no case for why this person. Check both before sending more.
High interview-to-offer ratio
Candidates look right on paper and wrong in the room. Usually a screening or prep problem: motivations and salary not tested, or the brief missed the thing the hiring manager cares about most.
Low offer-to-placement rate
Counter-offers, competing processes, salary surprises at the end. The fix sits earlier than the offer: ask about counter-offers at first call, and re-check expectations before every interview stage.
Low fill rate
You are working jobs you can't win: too many agencies on them, no access to the hiring manager, no urgency. This is a job qualification problem, and the cure is taking on fewer, better vacancies.
Rising fall-off rate
Placements leaving inside the rebate period cost you the fee and the client's patience. Look for candidates who were talked into the move, and for clients whose onboarding you never asked about.
Good ratios, low billings
The funnel is healthy and too small. That is an activity problem at the top: not enough jobs on, or an average fee that has drifted down. Ratios can't show this; counts can.
One caution on reading these. Ratios on small numbers jump around. A consultant who gets two offers from two interviews one month and none from six the next has not become worse at the job. Use a rolling eight or twelve weeks, and look at the direction before the level.
Time to submit, time to shortlist and time to fill
In-house teams measure time to hire and time to fill. An agency owns only the front of that clock, so it needs its own speed metrics, measured from the day the job comes on.
Time to shortlist measures sourcing: how fast you can find and qualify people. Time to submit adds the step after that, when chosen candidates are turned into something a client can open: interview notes written up, CVs put into the agency's format, contact details removed. On a contingent job with three agencies on it, time to submit is the speed metric that decides who gets the interview slots.
The stretch between "shortlist chosen" and "CVs sent" rarely gets its own timestamp, and it is often where a day goes missing. A shortlist agreed at 3pm that goes out at 11 the next morning lost most of a working day to reformatting. If your ATS records both dates, measure the gap for a month before deciding whether it is a problem.
Presentation also feeds the first conversion ratio. A client deciding between two similar candidates on paper will interview the one whose CV is easier to read and comes with a reason. Our guides to candidate presentation and the candidate profile template cover what that document should contain.
Temp and contract desks: margin, hours and headcount
A temp or contract desk has the same funnel at the front and a different business at the back. Revenue recurs weekly instead of landing once, so the outcome KPIs change.
| KPI | Formula | Why it matters |
|---|---|---|
| Gross profit per hour | Charge rate − pay rate − employment on-costs | The unit the whole desk is built from |
| Gross margin % | Gross profit ÷ charge rate × 100 | Shows discounting; compare by client |
| Weekly gross profit | Hours billed × gross profit per hour, summed across contractors | The contract equivalent of billings |
| Contractors out | Headcount on assignment at week end | Leading indicator of next month's GP |
| Net contractor change | Starts − finishers in the week | A desk can place weekly and still shrink |
| Extension rate | Assignments extended ÷ assignments reaching end date × 100 | The cheapest GP you will ever earn |
| Hours billed vs expected | Timesheet hours ÷ contracted hours × 100 | Catches missing timesheets and short weeks |
On-costs vary by country and engagement model (employer taxes, holiday pay, pension, umbrella or payroll fees), so define gross profit once with your finance lead and keep it fixed. When one agency says "margin" and means markup on pay rate, and another means percentage of charge rate, the same contract can be quoted as 25% or 20%.
For a mixed agency, convert both sides to gross profit so desks can be compared: a perm fee is all gross profit on the day it is invoiced, while a contractor at £6 an hour for 40 hours is £240 a week for as long as the assignment runs.
Setting KPIs for recruitment consultants: work backwards from billings
Activity targets picked from the air ("ten CVs a week, because that's what we've always said") get gamed or ignored. Targets derived from the consultant's own billing goal and their own ratios are much harder to argue with. The arithmetic is short:
- Placements needed Annual billing target ÷ average fee. Then add back expected fall-offs, since a refunded fee doesn't count.
- Offers needed Placements ÷ offer-to-placement rate.
- First interviews needed Offers × interviews per offer.
- CVs to send Interviews × CVs per interview. Divide by working weeks for the weekly number.
- Jobs to take on Placements ÷ fill rate. this is the one people forget
Put your own numbers in. The defaults are the example desk, not a recommendation.
Now change one ratio and watch the workload move. Taking CVs per interview from 3 to 4 adds a third to the weekly CV count for the same money. That is the case for coaching on conversion instead of shouting about activity: a consultant who improves one ratio a little gets several hours of their week back.
Use the consultant's own ratios where you have them. A new starter gets the desk average. After two quarters they get their own, and the weekly targets get personal. Two people with the same billing target can reasonably have different CV targets.
A weekly recruitment KPI dashboard
A desk dashboard should fit on one screen and answer two questions: are we doing enough this week, and is it converting. Eight tiles will do. Each shows actual against the weekly target from the calculation above, with ratios on a rolling four weeks.
That mock week is the reason to track ratios at all. Billings look fine because they reflect work done two months ago. The interview count and CV-to-interview ratio say the next two months are thinner, and they point at where: two clients turning down shortlists on paper, which means a call to re-take those briefs this week.
A weekly KPI sheet you can copy
If your ATS or CRM doesn't report this cleanly, a spreadsheet is enough. Paste these headers into row one, with one row per consultant per week. The ratio columns are formulas. If you're still choosing systems, our recruitment agency software guide covers which tools report what.
week_ending,consultant,desk,jobs_on_new,jobs_live,cvs_sent,first_interviews,second_interviews,offers,offers_accepted,starts,fees_billed,fall_offs,median_days_to_submit,client_meetings,notes # Rolling 4-week ratios (calculate on a summary tab) cv_to_interview = SUM(cvs_sent) / SUM(first_interviews) interview_to_offer = SUM(first_interviews) / SUM(offers) offer_to_placement = SUM(starts) / SUM(offers) fill_rate = jobs_filled / jobs_on_new (same quarter) fee_per_placement = SUM(fees_billed) / SUM(starts) fall_off_rate = SUM(fall_offs) / SUM(starts) (trailing 12 months) # Temp / contract desks: add these columns contractors_out,starts_temp,finishers,extensions,hours_billed,gross_profit_week,avg_margin_pct
Five ways KPIs go wrong
- Too many. Past ten numbers per person, nobody is managing any of them. Pick the funnel counts and two ratios.
- Activity with no quality check. A CVs-sent target with no CV-to-interview ratio beside it buys you speculative CVs and annoyed clients.
- Changing definitions mid-year. If "interview" started including phone screens in June, your trend line is fiction.
- Weekly ratios. Too few events to mean anything. Counts weekly, ratios rolling.
- League tables without context. A consultant building a new sector will have worse ratios than one farming three old accounts. Compare people with their own history first.
Questions agency owners ask
What are the most important recruitment KPIs for an agency?
Track one number at each stage of the funnel: jobs taken on, CVs sent, first interviews, offers, placements and fees billed. Then watch the ratios between neighbouring stages, especially CV-to-interview and interview-to-offer, plus fill rate, time to submit and fall-offs inside the rebate period. Temp and contract desks add contractors out, hours billed and gross margin.
What is a good CV to interview ratio?
There is no audited industry standard, and it varies by sector and seniority. Many permanent desks plan around three CVs sent for every first interview and treat anything above five to one as a sign of weak briefs or weak shortlists. Your own trailing twelve months, split by client, is a more reliable benchmark than any published figure.
What is the difference between time to hire and time to fill?
Time to fill runs from the day the vacancy opens to the day an offer is accepted, so it measures the whole process, including approvals and advertising. Time to hire usually starts later, when the eventual hire enters the pipeline, and ends at acceptance. Agencies control only part of either, which is why they also measure time to submit.
How do you calculate fill rate in recruitment?
Fill rate is jobs filled divided by jobs taken on in the same period, shown as a percentage. An agency that took on 24 vacancies in a quarter and placed candidates in 8 of them has a 33 percent fill rate. Count only jobs your agency filled, and decide in advance how to treat cancelled or on-hold vacancies.
What KPIs should a recruitment consultant have each week?
Give each consultant a short list they can influence this week: new jobs taken on, CVs sent to clients, first interviews arranged and client meetings or calls. Review conversion ratios monthly over a rolling window, because weekly ratios jump around on small numbers. Set the weekly quantities by working backwards from the consultant's annual billing target.
What is a good time to hire benchmark?
For in-house hiring in the United States, SHRM's 2025 recruiting benchmarking data puts median time to fill at 44 days for non-executive roles and 45 days for executive roles, measured from requisition opened to offer accepted. Agencies work inside that window, so the useful internal benchmark is days from job taken on to first CVs sent.
How is cost per hire calculated, and what is the benchmark?
Cost per hire is total recruiting spend, internal and external, divided by the number of hires. It is an employer's metric, and agency fees are one of its inputs. SHRM's 2025 data reports a US median of $1,200 for non-executive hires; the CIPD's 2026 UK survey reports medians of £2,000 for most employees and £3,500 for senior managers.
Where to go next
If the ratio you most want to move is CV-to-interview, start with what the client sees: the candidate profile template is a copyable structure for it. For interview-to-offer, a shared interview scorecard makes your screening and the client's interview test the same things. And if you're setting KPIs for a desk that doesn't exist yet, our guide on how to start a recruitment agency covers the first year's numbers.