skip to content

Hire a candidate, not a score

Hire a candidate, not a score
contents

One candidate gets a 92% match with the job description. Another tops out at 68%. The first is called in for interview, the second is turned away unread. Nobody stops to ask what those two numbers actually measure.

Assessment tools have multiplied: matching scores, online tests, interview scorecards, cross-ratings between interviewers. They have made hiring more rigorous, more comparable and often fairer. That is real progress.

But each of these tools produces a representation of the candidate, not the candidate. And the more precise those representations appear, the greater the temptation to confuse them with the person they describe.

For a hiring manager, the question is therefore not whether to choose measurement or intuition. It is to know what each measure says and, above all, what it does not.

Structure the assessment, yes. Reduce it to a single number, no.

We showed this in an earlier article: hiring interviews often measure the wrong thing (opens a new window). An unstructured interview gives too much weight to general impressions and to each assessor's biases. Structuring the assessment, with common questions and a shared scorecard, markedly improves the quality of decisions.

This article does not challenge that conclusion. It takes the logic further.

A scorecard is a tool for organising judgement. It does not replace it. When it becomes a machine for producing a final rating, added up and then compared, it recreates in another form the flaw it was meant to correct: a decision made on an impression, merely expressed in numbers.

What scores cannot see

A matching score compares a profile with a description. It measures how closely keywords, job titles, years of experience and stated skills line up. It is useful for sorting a large volume of applications. It is far less reliable at spotting what makes the difference.

It cannot see a career path. A candidate who has changed sector three times may be penalised by an algorithm, when that mobility is precisely a sign of adaptability.

It cannot see context either. The same experience as a sales director does not carry the same weight in a fast-growing SME and in a large group with a settled organisation. To a matching tool, the two lines on the CV look the same.

Tests have their own blind spots. They measure what they were designed to measure, in the conditions in which they were taken. They say little about how a person works with a team, reacts to a situation they have not met before or learns in a new line of work.

Finally, the job description itself is often flawed. Written from the predecessor's profile or an outdated template, it sometimes describes yesterday's need more than tomorrow's. A high score then mostly shows that a candidate resembles what already existed.

When the measure becomes the target

Economists know the phenomenon well. It is usually summed up as Goodhart's law: when a measure becomes a target, it ceases to be a good measure.

Hiring is no exception. When candidates know that a tool screens CVs by keyword, they adapt their CV to the keywords. When interview questions become predictable, answers are rehearsed and standardised. The score goes up, but the information it carries does not.

The same drift can occur on the company's side. A process judged on its ability to produce comparable ratings ends up favouring candidates who are easy to rate. Unusual profiles, those that do not sit neatly in the boxes of the scorecard, are set aside before anyone has understood them.

Take two candidates for an operations manager role. The first ticks every box: same sector, same job title, same tools. Their score is excellent. The second comes from a neighbouring sector and has led a smaller team, but has carried out two complete reorganisations in difficult circumstances. Their score is average.

If the role is to keep an already stable organisation running, the first is probably the right choice. If it is to transform an organisation, it may be the second. The score cannot tell the difference. Only knowledge of the role and a genuine conversation with the candidate can.

Use the tools without deferring to them

The aim is not to give up the tools, but to put them back in their place: that of an aid to decision-making.

It starts with a simple question, to ask before every hire: what must the person hired actually achieve in their first twelve months? The evaluation criteria, and the tools that measure them, are worth something only if they follow from that answer.

Faced with a score or a rating, a few questions then help you keep control of the decision:

  • What exactly does this indicator measure, and against which criteria?
  • Do those criteria match what the role demands today?
  • Which candidates did the tool screen out, and why?
  • What does this score not tell us about this candidate?
  • Would our decision hold up if we had to explain it without citing the score?

It is also worth rereading, from time to time, a few applications that the tool rejected automatically. This simple exercise checks that the tool is not quietly filtering out the very profiles the company is looking for.

Finally, an interview scorecard should leave room for what cannot be rated. A free comment, an open question about the candidate's career path or time to discuss their motivations can capture what the boxes cannot.

Conclusion

Assessment tools have made hiring more rigorous. They have reduced some biases, made decisions more comparable and made it possible to handle volumes of applications that were once unmanageable.

But a score remains an abstraction. It summarises part of the information, against criteria defined in advance, at a given moment.

The best hires rest on that balance: a structured assessment to limit bias, and human judgement to understand the person behind the measure. The hiring manager does not hire a score. They hire someone who will have to succeed in a real role.

FAQ

Are matching scores reliable?

They are useful for quickly sorting a large number of applications against explicit criteria. They are far less so for judging a career path, potential or suitability for a specific context.

Should you abandon assessment scorecards?

No. Shared scorecards improve the quality of decisions by limiting individual bias. The risk appears when they are used only to produce a final rating, with no analysis of the reasoning behind it.

What is Goodhart's law applied to hiring?

The idea that when a measure becomes a target, it loses its value. If candidates optimise their CV for a screening tool, the score they get reflects how well they have adapted to the tool more than their real skills.

What is the hiring manager's role when using assessment tools?

They must make sure the criteria measured match the needs of the role, interpret the results rather than apply them mechanically, and remain able to justify their decision beyond the score.

take action

turn reading into hiring

describe your role, our specialised headhunters take it on. 26 days on average.