Can you predict candidate success?

Definition

What the research generally points to

A large part of the literature on personnel selection deals with how well different methods correlate with later job performance. Schmidt & Hunter's meta-analyses of selection methods, published in Psychological Bulletin (1998), are among the most cited sources in the field and suggest that structured methods - including structured interviews and work samples - generally correlate better with later performance than unstructured conversations and experience-based assessments alone.

The U.S. Office of Personnel Management has, on that basis, published practical guidance on structured interviews, describing how to achieve more comparable assessments across candidates. What these sources have in common is that they talk about relative reliability between methods - not absolute predictability for the individual. None of the sources claim that an individual outcome can be predicted with certainty.

Why prediction is uncertain

• Limited information base: a conversation or test only captures a slice of a person's behavior, not the whole picture.
• The role changes: requirements, responsibilities and expectations for a position often evolve during the course of employment.
• People develop: motivation, competencies and behavior aren't fixed traits that can be read once and for all.
• Measurement error: even structured methods carry a margin of error, and subsequent performance reviews can themselves be imprecise or biased.

The importance of context

How an employee performs depends heavily on factors outside the candidate themselves:

• Management: clear expectations and relevant feedback affect whether an employee finds their footing in the role.
• Onboarding: a structured introduction reduces uncertainty in the critical early months.
• Team: the collaborative climate and colleagues' support can make a significant difference to whether an otherwise strong candidate succeeds.
• Organizational conditions: restructuring, shifting priorities and workload can change the framework of the role after hiring.

A candidate who would have thrived in one context may therefore fall short in another - not because the pre-hire assessment was wrong, but because the context after hiring was decisive.

Why observation is stronger than prediction alone

Prediction is a judgment made at a point when the least amount of information is available. Observing actual outcomes - how onboarding went, how performance developed, whether the employee stayed in the role and progressed further - is instead based on what actually happened.

That doesn't make prediction redundant. Structured signals before hiring are still the best basis for the decision itself. But the real strength emerges when you systematically compare early signals against later outcomes and use that relationship to improve the next decision - rather than trying to make the prediction itself sharper and sharper. That's the core of Hiring Intelligence: combining pre-hire signals with actual employee outcomes afterwards.

A prerequisite for being able to compare signals and outcomes at all is that the signals were gathered systematically to begin with - see structured hiring. Without structure, there's no comparable basis to learn from, either about prediction or about outcomes.

What this doesn't mean

• It doesn't mean structured methods are unnecessary just because they aren't perfect.
• It doesn't mean you can predict an individual's performance with certainty.
• It doesn't mean Hiring Intelligence is a prediction model - it's a learning and decision layer.

Also read about what a bad hiring decision typically costs, and how the risk can be reduced, in bad hires.