Course › Module 10 · Recruiting analytics and KPIs

The funnel model every recruiting report is built on

Module 10, Lesson 1  ·  6 min read ·  Updated 21 September 2026

Module 10 · Lesson 1

Every hiring report is built on a funnel underneath. Getting the funnel right is most of the work. The numbers on top are mostly arithmetic once your stages mean something.

The recruiting funnel and where people drop outA narrowing funnel from reached down to started, with the drop between screened in and first interview highlighted as usually caused by delay.ReachedScreened inFirst interviewFinal stageOfferAcceptedStartedthe biggest dropis usually latencyMeasure inbound and outbound as separate funnels
The funnel. The biggest drop is usually latency, not quality.

Your numbers depend on your setup

Module 8 said stages must be positions, not activities. This is where that pays off.

A stage called "waiting on hiring manager" makes every percentage meaningless, because a rate between two stages now includes something that is not a step.

If your numbers look odd, check the stage design before you check the data.

The basic funnel

StageGetting to the next one tells you
Reached — applied or contactedWhether your targeting and your advert are working
Got through screeningWhether your rules match the people you are attracting
First interviewWhether screening predicts anything, and whether booking is slow
Final stageWhether your interviews are set at the right level
OfferWhether your bar and your pool fit each other
AcceptedWhether your pay, process and pitch are competitive
StartedWhether anything goes wrong between accepting and day one

Most teams skip that last row, and it catches a real problem: people who accept and then do not start, usually because a counter-offer arrived during a long notice period.

Two funnels, not one

People who applied and people you approached behave completely differently. Applications give you a huge top and a low pass rate. Approaching people gives you a small top and a high pass rate, because you checked them before contacting them.

Combining them gives you an average that describes neither, and hides what is actually happening. Report them separately and compare each against itself over time.

Numbers that look good and tell you nothing

Not usefulUseful instead
Applications receivedWhat share got through screening — volume without quality is a cost
Interviews heldInterviews per offer
"Candidates in the pipeline"Candidates by stage, with how long they have been there
Messages sentReply rate per message in the sequence
Jobs worked onJobs filled, and how long they stayed open

The pattern: counting activity describes effort. Percentages describe whether the effort worked.

Reading the shape

What you seeUsual cause
Lots of applications, few getting throughThe advert attracts the wrong people, or your rules are too narrow
Good screening, poor interview resultsYour screening rules do not predict what interviewers look for
Plenty reach the final stage, few offersYour bar is above your pool, or the panel cannot agree
Offers turned downMoney, speed, or the process itself
A big drop between any two stagesCheck how long booking took before deciding anything about quality

That last row is the most common wrong diagnosis in hiring. A drop between screening and interview is usually delay, not quality — people left because it took eleven days to book a call.

Small numbers

Most teams do not hire enough for percentages to be stable. With twelve hires a year, one unusual search moves every number you report.

Two rules keep you honest. Always show the count next to the percentage. And never explain a change based on a handful of people. "40% offer acceptance" means something very different when it is 2 out of 5 versus 80 out of 200.

How often to report

  • Weekly: only things you can act on — who is stuck, what needs doing, who is going cold.
  • Monthly: funnel percentages, where people came from, hires against plan.
  • Quarterly: time to hire, cost per hire, quality measures, capacity.

Reporting a quarterly number every week produces noise that people then try to explain, which is worse than not reporting it.

Start here

Before adding any number, make sure every candidate is in the right stage and every rejection has a reason. Clean data with three numbers beats messy data with twenty.