The Job Interview Has Never Worked. Why Are We Still Using It?
Decades of research confirm that unstructured interviews are poor predictors of job performance. Organisations know this. They continue regardless. The persistence of a demonstrably flawed practice tells us something important about what hiring is actually for.
The opening tension
The research has been clear for fifty years: the unstructured interview predicts job performance at roughly the same level as a coin flip. The structured interview is meaningfully better. Almost no organisation uses it consistently. This is not an oversight.
The question
If unstructured interviews are demonstrably poor predictors of performance, why do organisations continue to use them — and what does the persistence of an ineffective practice reveal about what hiring decisions are actually optimising for?
The debate
The Psychologist — Occupational psychologist and assessment expert
The evidence here is among the most robust in applied psychology. Work sample tests and structured interviews with behavioural anchors significantly outperform unstructured conversations as predictors of job performance. This has been known since at least the 1980s. The persistence of the unstructured interview is not ignorance — practitioners are aware of the research. It is a revealed preference for something other than predictive validity. The question worth asking is: what are those other things, and are they legitimate?
The Economist — Labour economist
Responding to The Psychologist
The economic framing is useful here. Validated assessment tools cost more upfront — to design, administer, and defend. Unstructured interviews cost almost nothing and create no documented trail that can be audited for bias. If the organisation's implicit goal is minimising selection cost and legal exposure while maintaining hiring manager autonomy, then the unstructured interview is rational for those goals — just not for the goal of selecting the best person for the role. The "broken" system is perfectly optimised; it's just optimised for the wrong thing.
The Curator — Expert in organisational knowledge and judgment
Responding to The Economist
I want to defend what the interview is actually trying to do, even poorly. The implicit theory of the unstructured interview is that you are not just predicting task performance — you are assessing whether this person will navigate the politics, build the relationships, and read the social situation of this specific organisation. Those things are genuinely important and genuinely hard to test with work sample tasks. The interview's problem is not that it is trying to assess the wrong things — it is that it is a terrible instrument for assessing the right things.
The Contrarian — Critical thinker and institutional analyst
Responding to The Curator
The Curator is too generous. "Culture fit" — which is the primary thing unstructured interviews are actually assessing — is not a coherent construct. It is a container for affinity bias. You fit the culture if you remind me of people I already like and trust. The interview persists not because it is an imperfect instrument for a legitimate purpose, but because it is a socially legitimate mechanism for decisions that would be embarrassing if made explicitly. No one would defend "hire people who remind me of myself" — but the interview allows that decision to be made without being named.
The Psychologist — Occupational psychologist and assessment expert
Responding to The Contrarian
The Contrarian has named something important. The research on affinity bias in interviews is clear: interviewers systematically rate candidates higher who share their demographic background, educational background, and communication style — even when the interview is nominally structured. The solution is not to add more structure to the same conversation but to separate the selection criteria from the interview, validate them independently, and hold multiple interviewers accountable to the same rubric. Most organisations are not willing to do this because it constrains hiring manager discretion.
The Economist — Labour economist
Responding to The Psychologist
I want to end with a slightly uncomfortable prediction. The organisations that have adopted systematic, validated hiring — structured interviews, work samples, blind screening — have not found the process to be costless. They discover that what they thought they were selecting for (leadership potential, strategic thinking, culture fit) does not predict the outcomes they care about nearly as well as what validated tools measure (specific cognitive abilities, specific behavioural patterns in relevant situations). The uncomfortable implication is that many years of hiring decisions were optimising for proxies rather than outcomes. Valid hiring is a mirror, and not everyone likes what they see.
What the debate revealed
The debate revealed that the persistence of the unstructured interview is not ignorance of the evidence — it is resistance to what the evidence implies. Valid selection processes require defining what you are actually selecting for, being explicit about the criteria, and submitting those criteria to scrutiny. The interview, for all its predictive failure, allows the organisation to maintain productive ambiguity about what it actually wants. "Culture fit," "leadership potential," "executive presence" — these categories survive precisely because they cannot be objectively measured. The interview is the mechanism by which subjective preferences are laundered as professional judgment. Fixing hiring requires fixing the willingness to be explicit about criteria — which is a problem of organisational honesty, not measurement technology.
Practical implications
- Structured interviews with consistent questions, clear rubrics, and multiple independent assessors outperform unstructured interviews by a measurable margin — but require upfront investment in defining what good looks like.
- Work sample tests and job-relevant simulations are the highest-validity selection tools available and are dramatically underused, primarily because they require more work from both candidates and assessors.
- The persistence of "culture fit" as an interview criterion is particularly problematic — it is the category under which most affinity bias operates and the one most resistant to definition.
- AI-assisted structured screening is generating genuine improvement in some contexts and introducing new sources of bias in others — the key variable is whether the AI is optimising for defined outcomes or learned patterns in historical data.
- Organisations serious about hiring quality will find that the bottleneck is almost never the assessment method — it is the willingness to define, clearly and publicly, what they are actually trying to select for.
To keep thinking about
If your organisation made its hiring criteria completely explicit — and committed to measuring whether hires actually met them — what would you discover about how you have been making decisions?