Recruitment analytics

Recruitment analytics shows agencies and in-house talent teams which sources bring candidates who get placed, which funnel stage loses them, and how long roles take to fill. It is built from the stage dates in your applicant tracking system, with time to fill and time to hire defined apart.

Last reviewed

One desk's funnel by source, example

Job boards sent 600 of 760 applications and 4 of 15 placements

Example data
  1. Job boards600 applications4 placed

    0.7 placements per 100 applications

    Applied to shortlisted
    10.0%
    Shortlisted to interviewed
    40.0%
    Interviewed to offered
    25.0%
    Offered to placed
    66.7%
    540 of 600 job board applications stopped at the shortlist
  2. Candidate database120 applications6 placed

    5.0 placements per 100 applications

    Applied to shortlisted
    30.0%
    Shortlisted to interviewed
    50.0%
    Interviewed to offered
    44.4%
    Offered to placed
    75.0%
  3. Referrals40 applications5 placed

    12.5 placements per 100 applications

    Applied to shortlisted
    40.0%
    Shortlisted to interviewed
    62.5%
    Interviewed to offered
    60.0%
    Offered to placed
    83.3%

Blended: 15 placements from 760 applications, 2.0 per 100.

Blended 14.7% shortlist rate hides the 10.0% source
Example data: one invented permanent desk over a quarter, counting candidates by the source recorded on their first application. Placed means the candidate started, which comes after the offer accepted that ends time to fill and time to hire.
Show the numbers
Candidates by stage and source, example data
StageJob boardsCandidate databaseReferralsAll sources
Applications60012040760
Shortlisted603616112
Interviewed by client24181052
Offered68620
Placed46515
Applied to shortlisted10.0%30.0%40.0%14.7%
Shortlisted to interviewed40.0%50.0%62.5%46.4%
Interviewed to offered25.0%44.4%60.0%38.5%
Offered to placed66.7%75.0%83.3%75.0%
Placements per 100 applications0.75.012.52.0

Recruitment metrics, with the formula behind each

The KPI library has finance and sales dashboards built the same way.

KPIFormulaA good readingThe common mistake
Time to fillDays from the role being approved (in-house) or registered by the client (agency) to the offer being accepted Read per role type, against the date the hiring manager or client needs someone in post Starting the clock when the advert goes live. The days spent waiting for sign-off or a job description drop out of the figure.
Time to hireDays from the placed candidate's first application or first approach to their offer being accepted Read beside time to fill. The gap between the two is time spent before the right candidate entered the process. Reporting it under the name time to fill. The two clocks start at different points, and a short time to hire can sit inside a long time to fill.
Stage conversionCandidates who reached the next stage ÷ candidates who entered this stage, per role type and per source Steady stage by stage, so a fall points at one stage and one source Dividing by everyone who applied in the month. Candidates still waiting for a decision count as lost, so the latest month looks worse than it is.
Source qualityPlacements from a source ÷ applications from that source × 100 Read with the cost of each source beside it Crediting the last source recorded. If a returning candidate's source field is overwritten when they reapply, the source that first found them loses the placement.
Offer acceptance rateOffers accepted ÷ offers made, by the month the offer was made Steady by role type, with each decline given a recorded reason Stopping the count at acceptance. A candidate who accepts and then takes a counter-offer from their current employer has to be recorded as a failed start.
Fill rateRoles filled ÷ roles taken on, by the month the role was taken on Rising on the role types the desk has chosen to focus on Counting roles the client cancelled or filled itself as the desk's misses. Those need a different fix from roles the desk couldn't fill.
Fee per placement and margin (agency)Permanent: fee invoiced, net of any refund under the rebate period. Contract: (charge rate − pay rate − employer on-costs) per hour worked. Read per consultant and per client, with permanent and contract kept apart Quoting contract margin before employer on-costs. For contractors on the agency's own payroll, this row counts employer's National Insurance, holiday pay and pension as on-costs.
Cost per hire (in-house)(External costs, such as agency fees, advertising and assessments + internal costs, such as recruiter time and recruitment software) ÷ hires in the period Read per role type, so an expensive specialist hire doesn't hide inside an average Counting agency fees and adverts only. Roles the in-house team fills itself then look free, and the team's own cost never appears.

The source that sends the most applications can be the one that places the fewest people.

Where recruitment data lives, and where it meets HR data

The applicant tracking system or recruitment CRMCandidates, roles, stages and sources. If it stores only each candidate's current status, stage conversion and time to hire can't be rebuilt, so stage history is kept, or captured daily from now on.
Job boards and the careers siteViews and applications, counted by each board on its own terms. Without a source tag that survives the application form, the board's count and the ATS count drift apart.
Duplicate candidatesThe same person can apply through a board and be found in the database in the same month. Until records are matched on email address and phone number, they count twice.
Back office and financeTimesheets, invoices, fees and refunds. Fee per placement needs each invoice joined to its placement, and refunds joined back when a candidate leaves early.
The HR system and payroll, for in-house teamsStart dates, probation outcomes and leavers. Joined on an employee ID, they show which sources bring hires who pass probation and stay a year, reported only as counts and rates by source and role type. This join is where recruitment analytics meets HR analytics and people analytics.

Analytics services for recruitment agencies and talent teams

Stage history comes from your ATS, and data monitoring and support checks each morning that its data arrived and tells you when it did not.

Conversion is read stage by stage and source by source, so a leak in the funnel points at one stage and one source you can act on.

AI in recruitment analytics

AI-led analytics answers "which source placed the most candidates for finance roles this quarter?" from the desk's own data, and every answer shows its query. We stop at reporting and never score, rank or reject candidates. The AI for recruitment page shows the step before this one, matching candidates across job boards before any source gets the credit.

Questions from agencies and talent teams

Do you build recruitment dashboards for agencies?

Yes. A desk dashboard shows each consultant's open roles, stage conversion by source, time to fill against the client's need date, and permanent fees beside contract margin, all built from stage dates in the recruitment CRM.

Can you set up tracking for a careers site?

Yes. The tags carry each application's source and track the form's steps, so a visit joins to the candidate record in the applicant tracking system. Names, CVs and anything else a candidate types stay out of the tags.

Can AI help a recruitment desk with its reporting?

Yes, on the desk's own numbers, once stage, source and role type are columns in its data and source comes from each candidate's first application. Every answer shows its query, and the work doesn't score, rank or reject candidates.

What is the difference between time to fill and time to hire?

Time to fill starts when the role is approved or registered, so it shows how long the client waited. Time to hire starts when the person eventually placed first applied or was approached, so it shows how fast the process moved. Both end at offer accepted.

How long should we keep candidate records?

As long as your retention policy says, with a lawful basis. You set that with your data protection adviser, against the ICO's draft guidance on recruitment and selection, which is under review. The reports use counts and stage dates, so deleting records on schedule doesn't break them.

Does your ATS keep stage history?

On the call we check whether your ATS export keeps enough stage dates to rebuild your funnel by stage and source.

Book a call

Free 30 min call