Checking references: the most underestimated part of an appointment

Anyone who rereads the file after an appointment that went wrong will often find that the signals were there all along. They were not written down as warnings. They were small things that were noticed at the time and never voiced, such as a departure that was never quite explained or a referee who was warm about the person and remarkably vague about the results.

The reference check is the part of the process where such signals could have surfaced. Yet in practice it is the part from which least is expected. It sits at the end of the process, when the interviews are done and the decision has effectively been made. Anyone who calls at that point is looking for confirmation. It is almost always available.

An instrument we know remarkably little about

The best-known overview of selection methods, from 1998, gave the reference check a predictive validity of 0.26. That is modest, but not negligible. When a group of selection psychologists revisited those estimates in the Journal of Applied Psychology in 2022, they could not do so for the reference check: the original studies did not contain enough data to recalculate the figure. The structured interview came out at 0.42 in the same revision.

That makes the reference check the least researched instrument in an average selection process, even though hardly any executive appointment is made without one. This says less about the value of references than about the way they are usually taken: unstructured and at a moment when little still depends on them.

Ask about behaviour rather than character

A referee who is asked what the candidate is like describes a character. That produces words such as loyal and committed, which are hard to test. A referee who is asked what the candidate did describes a situation.

The reversal that works in the behavioural interview therefore works in the reference conversation too. Ask about the moment things went wrong and what the candidate did first. Ask which decision he took that others disagreed with and how it turned out. Ask what remained standing of his work after he left. A warm answer is not enough for questions like these. A referee who hesitates or falls back on generalities is telling you something as well.

Timing matters just as much. We advise clients to hold the reference conversations before the choice is final. A conversation on which something still depends is conducted differently, at both ends of the line.

What recurs is what counts

A reference check we carried out last year for an operational management role shows where the information lies. We spoke to three referees from three stages of the candidate’s career: his direct manager in his most recent role, a colleague he had worked with years earlier and a manager from more than ten years before that. All three were positive and would gladly work with him again.

More interesting was what two of them said independently. The recent manager explained that the candidate sometimes took too much responsibility on himself rather than delegating. The manager from more than ten years earlier named delegation as the point on which he still had to grow at the time. Two people who worked with him more than a decade apart arrived at the same point. It was not an objection, but it does belong in the advice to the client and in the conversation about the first months.

One referee who names a development point gives you an opinion. Two referees who independently name the same thing give you a pattern. That is why we prefer to speak to people from different stages and from different positions relative to the candidate and why we put the same closing question to each of them: would you hire this person again and why or why not?

Whom you speak to

The referees a candidate puts forward have been chosen with care. That is understandable and legitimate, but it means you are speaking to people who wish the candidate well. How well became clear in American research published in early October: of more than a thousand workers surveyed, 40 percent said they had recommended someone despite their own reservations. It is a survey by a CV platform in a different labour market, so the figure is an indication rather than a measurement. The mechanism is universal, though: someone asked to act as a referee does not want to stand in the other person’s way.

Widen the circle, therefore, in consultation with the candidate. Ask whether you may also speak to someone who reported to the candidate and, for a board-level role, to the chair of the board to which he was accountable. In the Netherlands, consent is the starting point. The Dutch recruitment code of the NVP (July 2025 version) states that an organisation wishing to obtain a reference asks the applicant for permission beforehand.

In the same American survey, 37 percent said a prospective employer had contacted a former manager or colleague they had not listed. That kind of call behind the candidate’s back does not fit Dutch practice and it is not necessary either. A candidate who responds calmly to a request for an additional referee is already showing you something. So is a candidate who hesitates.

A board responsibility

In the financial sector, the regulator takes over part of this work. De Nederlandsche Bank carries out around two thousand fit and proper assessments of directors and supervisory board members each year, of which 3 to 5 percent have a negative outcome. References are weighed in the decision to assess a candidate in more depth. Outside that sector, no regulator is watching. There, the diligence rests with the supervisory board itself, which makes the reference check a governance matter rather than an administrative formality.

That diligence is becoming more important, not less. Research firm Gartner predicted last year that by 2028 one in four candidate profiles worldwide could be fake. That applies mainly to roles filled remotely and far less at board level. It does make clear that verification will weigh more heavily in every selection process.

The candidate checks references too

In most processes, a reference check runs in one direction. The client asks about the candidate and the candidate waits. For an executive appointment that is a curious asymmetry. The candidate commits for years to an organisation he has seen in three interviews, under a chief executive or board he barely knows.

We therefore advise candidates to do the same as the client. The contacts the organisation itself puts forward have been selected just as carefully. So ask whether you can speak to someone who previously reported to your future manager and no longer works there. It is an uncomfortable request. How it is received is information in itself.

For a client, that is an advantage rather than a threat. A candidate who asks such questions is a candidate who is considering staying. Appointments that still fail after a year more often start with a candidate who asked too little than with an organisation that told too little.

A small circle at the top

In a region where the people at the top know each other, the temptation of the informal phone call is greatest. Take the port region along the North Sea Canal, from IJmuiden to Westpoort. This month it was announced that the Amsterdam business association ORAM is looking for a new director, as Kees Noorman, who has held that post since 2013, becomes its chair. It is an ordinary change of leadership. It does show how closely the networks at the top of such a region overlap.

Anyone assessing a candidate in such a circle almost always knows someone who knows the candidate. That is precisely when it is wise to hold the conversation the regular way, with the candidate’s consent and with the same questions you would put to any other referee. An impression gained second hand is quickly formed and hard to refute, even when it is wrong.

What we tell clients

Treat the reference check as part of the decision rather than its conclusion. Hold the conversations before the choice is final and prepare them as you would an interview, with questions about concrete situations and decisions. Alongside the listed referees, speak to someone who reported to the candidate, always with the candidate’s consent. Record what was said, so that the board can read it back later. And expect a good candidate to put the same questions to you.

In closing

A reference conversation takes an hour. An appointment that fails after a year costs an organisation considerably more. That hour is therefore the best-spent hour of the entire process, provided it is spent before the decision is made.