The promise is that data will tell you who to hire. It won't, and the firms selling that version tend to be selling a product rather than describing a practice.
What data does well is something less exciting and considerably more useful: it tells you where your process is losing people, and whether your plan is achievable.
Where it genuinely works
Diagnosing where a process leaks
Conversion rates at each stage (approach to response, response to interview, interview to offer, offer to acceptance) locate a problem far faster than opinion does. A search producing plenty of interviews and no offers has a calibration problem. One producing offers that get declined has a package or proposition problem. These are different fixes and the numbers tell you which you have.
Testing whether a plan is achievable
If you need eight technical hires in a region within six months, the honest question is whether the qualified population supports it. That's arithmetic: population size, realistic engagement rate, historic conversion. Doing it before the plan is signed off is much cheaper than discovering it in month five.
Benchmarking the offer
Offer-decline reasons, gathered systematically rather than anecdotally, are the fastest route to knowing whether your package is competitive. Most businesses have this data and don't record it in a form they can use.
Timing an approach
Movement in a market is not random. Post-acquisition integration periods, bonus cycles and site restructures create windows when otherwise settled people become open. Tracking those events is more predictive of who will move than any scoring model applied to a CV.
Where it goes wrong
Small numbers. Specialist hiring generates too few data points for most statistical claims. A firm making twelve technical hires a year cannot derive reliable predictors of success from its own history, and confident-looking conclusions drawn from that sample are noise.
Modelling the past
A model trained on who has succeeded historically will reproduce the historical profile, including its biases. If every previous plant manager came from two competitors, the model will confidently recommend the same two, which is a description of the past, not a prediction.
Measuring what's easy instead of what matters
Time-to-hire is easy to measure and frequently optimised at the expense of quality. Quality of hire is hard to measure and usually isn't. This is how a process ends up efficiently producing the wrong outcome.
Assessment tools presented as prediction
Structured assessment genuinely improves consistency. Tools claiming to predict performance from behavioural or inferred data should be held to the same evidential standard as any other hiring instrument, and in several jurisdictions face increasing regulatory scrutiny.
A practical starting point
You don't need a platform. Four things recorded consistently will outperform most tooling:
- Stage-by-stage conversion for every search
- The stated reason for every offer decline
- Source of hire, honestly recorded
- A simple retention and performance check at twelve months
That is enough to tell you where you're losing people and whether the plan is realistic. Everything beyond it is refinement.
Common questions
Can predictive analytics tell you which candidate will perform best?
Not reliably in specialist hiring. The data volumes are too small to support that kind of claim, and models trained on historical hires tend to reproduce the historical profile along with its biases. Analytics is far more useful for diagnosing process problems and testing whether a hiring plan is achievable.
What hiring metrics are actually worth tracking?
Stage-by-stage conversion rates, the stated reason for every offer decline, honestly recorded source of hire, and a retention and performance check at twelve months. These four identify where a process is losing people and whether a plan is realistic.
Why is time-to-hire a risky metric to optimise?
Because it is easy to measure while quality of hire is not, so processes tend to optimise for speed at the expense of the outcome that matters.
Want a read on your hiring data?
We're happy to look at conversion and decline patterns on a search and tell you what they suggest.
Book a Confidential Briefing →