Reference Check

What a reference check is

A reference check (التحقق من المراجع) is the step in which an organisation asks people who have worked with a candidate about the candidate’s performance, conduct and way of working, so that their views add to what the interviews showed. The candidate can name these people, or the organisation can reach them by another route, and they are known as references or referees. At this step the organisation is not confirming that a document was issued. It is gathering an assessment from people who saw the candidate at work.

The concept concerns the recruiter, the manager the candidate would report to and the owner of the organisation. An interview shows what candidates say about themselves in a limited time, whereas a referee can describe what they saw over a longer period. The definition describes a management practice in recruitment, not a legal rule.

How a reference check differs from a background check

The two differ in the nature of what is gathered. Confirming facts that can be proven from what the candidate submitted, such as the qualification, the job titles and the periods of employment, is the work of a Background Check. The result of a reference check is one person’s assessment. It can vary from one referee to another, and no document proves or disproves it.

The handling of each result can differ accordingly. A qualification with no basis is a fact placed before whoever makes the decision. A referee’s opinion of a candidate is taken as one opinion to be weighed with others, and no firm judgement is built on it alone. This is our view of how the two results are handled, not an established rule. A reference check does not replace the interview or the background check either. Its result is one input, considered together with the others.

Choosing the referee

The right referee depends on what the organisation wants to learn about the candidate, and we suggest choosing among these categories according to the requirements of the job:

• The previous direct manager: can speak to what was assigned to the candidate, how the work was carried out and how the candidate responded to direction.
• A peer at the same level: can describe how the candidate cooperated and shared the work.
• Someone who reported to the candidate: can be useful for supervisory roles, having seen the candidate’s way of managing from a subordinate’s position.
• A client or business partner: can be useful for roles that deal with people outside the organisation.

Candidates who name their own referees can pick people they expect to speak well of them. That is a possibility we assume, not a measured finding. For that reason a candidate can be asked for a referee from a particular category, such as their most recent direct manager, rather than being left the whole choice.

Where a reference check sits among nearby recruitment concepts

Reference checking can be confused with several other recruitment concepts, and each of them answers a different question:

• Employee referral: an employee of the organisation recommends someone they know for a vacant job under an announced programme. A referral is a route for bringing a candidate in before selection, whereas a reference check concerns a candidate who has reached an advanced stage of selection. Both can apply to the same candidate, but each has a separate purpose.
• Cybervetting: its source is what the candidate has published openly online. In a reference check, by contrast, someone who worked with the candidate speaks in answer to a question put to them. Cybervetting gathers material the candidate was never asked about, while a reference check gathers what a referee was asked to say.
• Exit interview: a meeting with an employee who is leaving, in which that employee describes their experience of the organisation. The referee in a reference check is an outside party describing a candidate who has not yet joined, so the description runs in the opposite direction.

These distinctions show where each concept sits in the recruitment process. They do not mean that an organisation never combines them.

Questions to put to referees

The aim is to gather evidence on what the job requires, not general impressions. We recommend deriving the referee questions from the same job requirements that underpin a Structured Interview for the role, and putting the same question to every referee asked about candidates for the same job. Possible wordings include:

• What tasks were assigned to the candidate while working with you, and how were they carried out?
• In what situation did the candidate’s way of handling pressure, or a disagreement within the team, become clear?
• Where did the candidate need support or monitoring?
• If you had the chance to work with the candidate again, would you, and why?

We favour questions that ask for an example over questions answered with yes or no, because an example can be checked against what another referee said. These question forms are our suggestion, not a list taken from an authoritative source.

Timing within the recruitment process

The definition sets no single point in time. The check can take place after the final interview and before an offer is made, or after an offer that is conditional on its result. When an offer depends on the outcome of a reference check, our glossary account of the conditional Job Offer states that the condition is written into the offer in express terms, together with what follows if it is not met, so that the candidate knows what is expected.

We also recommend telling the candidate that the step exists and roughly when it will happen, because a long silence between two steps can be understood by the candidate differently from what the organisation intends. Telling candidates where they stand in the process is a matter of Candidate Experience, meaning the candidate’s impression of the organisation over the whole recruitment journey.

A hypothetical example of combining referee scores

Suppose two finalists, A and B, each with 3 referees, and suppose each referee gave a score out of 5 on two criteria: punctuality and cooperation. The scale in this example is hypothetical and does not represent any recognised scale.

• Candidate A: punctuality scores of 4, 4 and 5, a total of 13 and an average of 13 ÷ 3 = 4.33. Cooperation scores of 3, 4 and 3, a total of 10 and an average of 10 ÷ 3 = 3.33.
• Candidate B: punctuality scores of 5, 3 and 3, a total of 11 and an average of 11 ÷ 3 = 3.67. Cooperation scores of 5, 4 and 4, a total of 13 and an average of 13 ÷ 3 = 4.33.

Across both criteria, A averages 23 ÷ 6 = 3.83 and B averages 24 ÷ 6 = 4.00. The difference, 4.00 minus 3.83, is about 0.17 of a point. A gap of that size on a scale out of 5 may not be enough on its own to choose between the two, so the examples the referees gave can be reviewed and the spread of the scores examined. B’s punctuality scores are split between 5, 3 and 3, which could justify a question about where each score came from.

The effort can be estimated as well. If a round has 5 finalists with 3 referees each, the number of calls is 5 × 3 = 15. If each call takes 20 minutes, the time is 15 × 20 = 300 minutes, or 5 hours. If the check is limited to 2 finalists, the time is 2 × 3 × 20 = 120 minutes, or 2 hours. The figures are hypothetical, and no benchmark for the number of referees or minutes follows from them.

The definition offers no real figures either: none for how many organisations check references, none for the share of CVs that are exaggerated, and none for the right number of referees for a particular job. In the sources we reviewed, we found no such figures that we could confidently attribute to a known source.

How to weigh what a referee says

What a referee says is that person’s opinion, not an established fact, so we suggest considering points such as these when weighing it:

• How close the referee was to the work: did the referee see the candidate at first hand, or is the account passed on from someone else?
• How recent the shared work was: what happened years ago can differ from what happens today.
• Agreement among referees: a point repeated by several referees can deserve attention, although a single remark can also deserve a question about the example behind it.
• Examples over descriptions: a word such as “excellent” needs a question about what supports it.

These are criteria we propose for weighing the answers, not a recognised scale. In the sources we reviewed, we found no measured predictive power for reference checking, expressed as a percentage, available to cite. Any effect of a reference check might show up months later in Quality of Hire, the measure that answers whether the hire was a good one, so the day of the check cannot reveal whether the decision was right.

Whether one referee’s opinion is enough to reject or accept a candidate is not settled by the definition. That decision rests with the organisation, according to the job and everything it has gathered. What is said about a candidate at this step can be coloured by the referee’s relationship with that candidate, which is why the decision remains the organisation’s and not the referee’s.

How reference checking is misused

In practice, an activity can be called a reference check when it is something else:

• A check in name only: the referee is asked only to confirm the period of employment, which turns the step into a background check carried out as a phone call.
• Leading questions: they are worded to suggest the answer, for example asking whether the candidate was always committed, so the referee can agree without giving an example.
• Different questions for each candidate: they make the candidates hard to compare.
• Nothing recorded: what was said goes unwritten, and memory can differ from what happened when the decision is later reviewed.
• Contact without the candidate’s knowledge: it can unsettle the candidate or put them in a difficult position with their current employer.

These patterns come from thinking through the practice, not from the results of a measured study.

Candidate consent and personal data

We suggest informing the candidate before any contact is made, and asking the candidate whom to contact and when, particularly where the referee is a current employer. That is a suggested administrative arrangement for dealing with the candidate, not a statement of what any regulation requires or permits.

Opinions gathered about a candidate can count as data relating to that candidate. For the general principle, an organisation can consult our explanation of Employee Data Privacy and our entry on Data Subject Rights. The definition of a reference check settles nothing about whether it is permissible to collect or keep particular information, or to contact a particular person, and anyone facing a question of that kind should take it to whoever holds the legal opinion in the organisation.

Review questions for the recruiter

• What do we want to know about the candidate that the interviews did not reveal?
• Who is on the list of referees, and who named them?
• Are the same questions put to every referee for the same job?
• Where is what was said recorded, and who reads it?
• Has the candidate been told about the step and its timing?

The questions produce no score. What they show is where the practice may need tightening.

How an organisation can make use of reference checks

We propose that the organisation write, for each job, a short list of the questions to put to referees, and decide who makes the calls and who records what was said. The step then becomes part of the recruitment process, reviewed in the same way as its other steps, rather than a call made whenever time allows.

The recommendation is ours and is not part of the definition. A reference check is defined as gathering the views of people who worked with the candidate, and it is judged by what it added to the decision after the hire.

This is an explanation of the concept, not legal advice.