Predicting job performance: the science of smarter hiring

Predicting job performance means measuring the things that actually carry across employers: general cognitive ability, work preferences and motivation, and demonstrated behaviour on work resembling the job. Resumes and rapport predict very little, which is why a good hire so often feels like a lucky hire. Combining a validated assessment with a scored, structured interview and a scorecard written before anyone applies is what turns that luck into a repeatable call you can explain afterwards.

Last reviewed September 2026

Why a good hire feels like a lucky hire

Most hiring processes gather evidence that does not travel. A resume records what someone did inside a different company, with a different manager, different constraints and a different definition of success. A confident interview records how well a person performs in a forty-minute conversation about themselves, which is a genuine skill and almost never the job.

When the inputs are weak, outcomes scatter. Some hires land brilliantly, some fail for reasons nobody can articulate, and the hiring manager quite reasonably concludes that hiring is a bit of a lottery. The lucky hire is the one who happened to want the work the role contained. Nothing in the process detected that, so nothing in the process can repeat it.

The cost of the other outcome is easy to underestimate. Recruitment fees are the visible part, and the larger share sits in the ramp-up time, the manager hours, the delayed projects and the churn that follows in the team. Running your own numbers through a cost of a bad hire calculator tends to end the debate about whether better assessment is worth the licence fee.

What actually predicts job performance

Decades of selection research point at a short list of things that do travel between employers. The rank order shifts by role and by study, and the broad picture has been stable for a long time.

General cognitive ability is the strongest single predictor across most roles, and it matters more the more complex and ambiguous the work is. It measures how quickly someone takes in new information and solves problems they have not seen before, which is what the first six months of any role consist of.

Work preferences and motivation predict engagement, discretionary effort and how long someone stays. This is the part the resume cannot show. Two people with identical capability produce different output when one of them is doing work they are drawn to and the other is doing work they tolerate.

Demonstrated behaviour on job-like work predicts well because it is the closest thing to watching the person do the job. Work samples, structured behavioural questions scored against a rubric, and job simulations all sit here.

Years of experience and education predict weakly once a threshold is met. They are the easiest things to read off a page, which is why they dominate screening despite being poor at the task. We have written separately on why a resume does not predict performance.

Nothing here predicts perfectly, and that is worth saying plainly. The aim is to raise the hit rate from something close to chance to something you would defend in front of a board.

Which tests predict job performance

Buyers searching for a job performance test usually want to know which instrument to trust. The honest answer is that no single one carries a hiring decision, and the combination matters more than the choice. Here is how the common methods compare.

Whichever you pick, ask the vendor what the instrument was validated against, on which population, and when. A published psychometric assessment carries technical documentation. A personality quiz repackaged for hiring does not. If the shortlist is the stage you are trying to fix, it is also worth reading how assessments compare with video interviewing, because the two get sold as the same purchase and answer different questions.

How Compono predicts team dynamics and reduces bias in hiring decisions

Two separate mechanisms are doing the work here, and they are worth pulling apart.

Team dynamics come from measuring the team you already have. Plotting every member on the work personality wheel shows which of the eight kinds of work the team leans into and which one is nobody's instinct. A candidate profile can then be read against that shape rather than against a vague idea of seniority, so the prediction is about how this person will change the team rather than how well they will present in the room. Managers get the likely friction points in advance, which is a far better conversation to have in week one than in month six.

Bias reduction comes from the order of operations. Compono Hire assesses every candidate on the same dimensions before the shortlist is built, so the first comparison is between scores rather than between documents. Criteria are set in advance, results are recorded, and the reasoning survives long enough to be audited. The fit boundary is worth naming: this improves the quality and the defensibility of a decision between the candidates in front of you, and it does nothing about a pool that was narrow before anyone applied. Sourcing and job ad wording are separate problems with separate fixes.

No assessment removes bias entirely. Humans still run the interviews and still make the offer. What structure does is reduce the number of moments where an unexamined preference is the deciding input.

How science-backed shortlisting and culture fit analysis work together

Shortlisting on capability alone produces candidates who can do the job. Shortlisting on fit alone produces candidates who will enjoy the job. Neither by itself predicts someone who will do it well for three years.

Combining them is straightforward in principle. Candidates complete the assessment before the resume review, which puts measured evidence ahead of inference. Each candidate is scored on organisation fit, job fit and personality fit alongside skills and qualifications, so the shortlist is ranked on a weighted total rather than assembled from whoever the reader warmed to. The interview then tests the specific things the scores raised.

An illustrative example with invented numbers shows the mechanics. Say the scorecard weights capability at 40 per cent, cognitive ability at 30 per cent and measured fit at 30 per cent. A candidate scoring 8, 6 and 9 out of 10 lands on 7.7. A candidate scoring 6, 9 and 7 lands on 7.2. The point is not the decimal. It is that the conversation shifts to whether the weightings are right for this role, which is a question the team can actually answer, rather than whose impression carries more weight in the room.

Culture fit analysis earns its place only when it measures preferences and behaviour rather than background or shared interests. Unmeasured, fit collapses into affinity and becomes a similarity filter with better branding. Measured against a defined culture, two candidates from entirely different backgrounds can both score well, because what is being compared is how they work.

The eight work personalities and the work each is drawn to

Compono groups work preferences into eight types. Each describes the kind of work a person is motivated to do rather than what sort of person they are.

• Doer. Finishes things, prefers momentum to deliberation.
• Auditor. Methodical and detail-driven, holds standards and compliance.
• Helper. Builds cohesion, notices when people are struggling.
• Advisor. Thinks through implications and shapes direction with judgement.
• Pioneer. Starts things, comfortable with risk and ambiguity.
• Campaigner. Persuasive and outward-facing, sells the idea and builds the network.
• Evaluator. Weighs options logically and tests the assumptions others skipped.
• Coordinator. Organises people and work so delivery becomes predictable.

A high-growth business short on follow-through does not need another Pioneer, however well that person interviews. It needs the work nobody is currently drawn to. The free work personality overview explains what each type looks like day to day.

Building a scorecard so job success is predicted, not discovered

The practical version of all this fits on one page, written before the role is advertised.

1. List what the job actually contains. The activities, in rough order of how much time they take. Not the title, not the org chart.
2. Name the capabilities and the cognitive demand. What must be true on day one, and what can be learned in ninety days.
3. Describe the work preferences that suit it. Which of the eight kinds of work this role is mostly made of.
4. Set the weightings and get them agreed. In writing, with the hiring manager, before the first application arrives.
5. Assess every candidate the same way, early. Ahead of the resume sift, so the ranking is built on measurement.
6. Score the interview before the debrief. Each panellist records their scores independently, then the panel discusses the gaps.
7. Check the prediction at six and twelve months. Compare scores against performance and retention, and adjust the weightings. This is the step that makes the next role easier, and the step almost everyone skips.

Frequently asked questions

What is the best predictor of job performance?

General cognitive ability is the strongest single predictor across most roles, and it gets stronger as the work becomes more complex. On its own it is incomplete, because it says nothing about whether a person is motivated by the work or likely to stay. Pairing it with a work preference assessment and a scored structured interview gives a far better prediction than any one method alone.

Is there a test that predicts job performance?

No single test carries a hiring decision. Cognitive ability tests, work personality assessments, work samples and structured interviews each predict a different slice, and the combination is what raises accuracy. Ask any vendor what their instrument was validated against, on which population, and when.

Why does a good hire so often feel like luck?

Because the evidence most processes collect does not travel between employers. A resume describes results produced under someone else's conditions, and an interview measures how well a person talks about themselves. When the inputs are weak, outcomes scatter, and the hires that land well are the ones who happened to want the work the role contained.

How do you predict team dynamics before someone starts?

Measure the team you already have first. Mapping every member's work personality shows which kinds of work the team leans into and which are nobody's instinct, and a candidate profile can then be read against that shape. It gives managers the likely friction points in advance rather than six months in.

Can predicting job performance with data reduce bias?

It reduces the number of moments where an unexamined preference is the deciding input, which is a real gain and not the same as removing bias. Consistent criteria set in advance, the same assessment for everyone, scores recorded before discussion, and a quarterly review of selection rates by stage are what make the improvement durable.

What are the eight work personality types?

Doer, Auditor, Helper, Advisor, Pioneer, Campaigner, Evaluator and Coordinator. Each describes the kind of work a person is naturally motivated to do, and a team performs best when the work it depends on is covered by someone who is drawn to it.