How to Tell If a Candidate Can Actually Do the Job

Resumes and interviews measure how well someone describes their ability. They do not measure whether the person has it.

To test the ability itself you need evidence of the work: a realistic work sample, structured questions about decisions the candidate actually made rather than hypotheticals, and reference conversations that ask about specific situations instead of general impressions.

That gap between describing and doing is why confident interviews still produce disappointing hires, and why adding more interview rounds rarely fixes it.

Method	What it reveals	What it misses
Resume	Where someone has been, and what they claim	Whether they caused the outcomes listed
Unstructured interview	Communication, likeability, self-presentation	Almost everything about execution
Structured behavioural interview	How they reasoned through real past situations	How they perform under current conditions
Work sample or realistic task	Whether they can do a version of the actual job	How they operate over time, with a team
Reference check, done specifically	Corroboration of claims, patterns over time	Anything the referee did not observe directly

The pattern is that no single method is sufficient, and the two most commonly used, the resume and the unstructured interview, are the two weakest at the thing you care about.

An interview tests someone's ability to describe work under time pressure to a stranger. Some jobs genuinely require that skill. Most do not.

The predictable consequence is that you select for polish. Candidates who present well advance, candidates who do the work well but describe it plainly do not, and the correlation between those two groups is weaker than most hiring managers expect.

The other problem is consistency. When different interviewers ask different questions and score against different implicit standards, you are not comparing candidates against each other. You are comparing each candidate against whoever happened to interview them that day.

The useful version is short, realistic, and resembles something the person would genuinely do in the role.

Good: a two-hour version of a real problem your team has faced, with the context you would actually give an employee, discussed afterwards so you can hear the reasoning.

Bad: a multi-day unpaid project, a puzzle unrelated to the job, or a task so sanitised that it tests nothing.

The discussion afterwards matters more than the artefact. What you want to know is not only whether the output was good, but how they decided what to do, what they would change with more time, and what they noticed was ambiguous in your brief.

Ask about decisions the candidate actually made, not what they would do hypothetically. Hypotheticals test imagination, and the answer is almost always the textbook one.

Useful shapes:

• Tell me about a time this went badly. What did you do next?
• What was the hardest call you made in that role, and what made it hard?
• What would you do differently if you ran that project again?
• What part of this job do you think you would find hardest?

That last question is the most informative on the list. Candidates who cannot name anything difficult about the role either have not thought about it seriously or are not being straight with you.

Most reference calls produce nothing because they ask for impressions and get politeness.

Ask about specifics instead. What did this person actually own? What did they do when a project slipped? Where did they need support? Would you hire them again into that same role, and what would you change about how they were set up?

A referee who answers narrowly and factually is telling you more than one who is enthusiastic in general terms.

The strongest version of all this is not a clever technique. It is running the same evaluation for every candidate for a role, so the comparison means something.

That is harder than it sounds. It requires agreeing what good looks like before you interview anyone, asking comparable questions, and recording the reasoning somewhere it can be reviewed rather than relying on what people remember feeling a week later.

Companies that do this consistently make noticeably fewer mis-hires, not because any one interview is better, but because the decision rests on evidence that can be compared and revisited.

This is the part of hiring The Chum Effect is actually built around. Every candidate we send is interviewed live by a person before you see them, and those interviews are recorded, so you review what the candidate actually said and how they reasoned rather than reading our summary of the conversation.

The point is not the recording itself. It is that the evidence behind a recommendation is available to you, that every candidate for a role is evaluated the same way, and that you can go back and check a decision rather than relying on a memory of how someone came across.

If your current process produces candidates who interview well and disappoint later, tell us about the role. More on how this works across staffing solutions, embedded recruitment squads and direct placement.