A reference check that only confirms dates of employment is not a reference check. It is a formality that happens to feel like due diligence. Most recruiters run exactly this kind of call: three closed questions, a polite thank-you, a tick in the box. The candidate was never at risk of failing it, and the client learns nothing they could not have read on the CV.
The version worth running is different. It treats the referee as a source of specific, checkable detail about how someone actually works, not a character witness reciting adjectives. That shift, from confirmation to investigation, is what separates a reference check that catches a bad hire from one that just adds paperwork to a placement that was always going to happen anyway.
A referee who cannot give you one concrete story about the candidate has probably not worked closely with them, whatever the job title on record says.
What is reference checking supposed to achieve?
Reference checking exists to verify a candidate’s account of their own work history and to surface how they actually perform day to day, catching gaps between the story on the CV and the reality a former manager describes. Done well, it is the last independent check before a client commits.
It sits at the end of the process for a reason. By the time you are checking references, the client has already interviewed, the candidate has already impressed, and everyone in the room wants the answer to be yes. That momentum is exactly why the check needs structure: without it, confirmation bias does the interviewing for you and the referee call becomes theatre.
Bodies like the CIPD treat referencing as part of a wider selection and assessment discipline, not a bolt-on compliance step. Recruiters who treat it the same way get more out of the fifteen minutes on the phone.
What legal limits apply to reference checks in the UK and EU?
In the UK and EU, reference checking is constrained mainly by data protection law and, in some sectors, by employment agency regulation, both of which require consent, purpose limitation, and a documented lawful basis before you contact a third party about a candidate. Get this wrong and the check itself becomes the compliance risk.
Under GDPR, a referee’s comments about a candidate are personal data about that candidate, processed via a third party. You need the candidate’s informed consent to seek references, you should tell them who you are contacting and roughly what you will ask, and you should keep the record only as long as it serves the placement decision. In the UK, employment agencies also fall under specific conduct rules set out via gov.uk guidance for employment agencies and businesses, which covers how candidate information can be shared with clients.
A practical habit: ask for consent in writing, at the point the candidate submits referee details, not verbally after the fact. That single step removes most of the compliance risk from the rest of the process.
| Practice | Compliant approach | Common mistake |
|---|---|---|
| Consent | Written, before contacting any referee | Verbal, assumed from a CV listing referees |
| Scope of questions | Job-relevant conduct and performance | Health, family status, unrelated personal opinion |
| Record keeping | Logged in the ATS with referee identity and date | Notes left in a recruiter’s personal inbox |
| Sharing with client | Summarised findings, referee consent noted | Forwarding raw call notes verbatim |
What questions predict on-the-job performance?
Behavioural, incident-specific questions predict performance far better than closed or character-based ones, because they force the referee to describe something that actually happened rather than offer a general impression. “Was Anna reliable?” invites a reflexive yes; “walk me through a time Anna missed a deadline” does not.
A short, reusable set works better than a long generic form. Ask for one example of the candidate handling a difficult stakeholder, one example of a mistake and how they recovered from it, and one example of unprompted initiative. Then ask the hardest question of the call directly: would you rehire this person, and why or why not. Listeners who go quiet here are telling you something a scripted answer never would.
This is also where SHRM’s talent acquisition coverage and LinkedIn’s talent blog keep converging on the same point: structured, comparable questioning across candidates beats an unstructured chat every time, for the same reason structured interviews beat unstructured ones.
The best reference question is not on any template. It is the follow-up you ask because the first answer was too smooth.
When do references actively mislead recruiters?
References mislead most often when the referee has an incentive to get rid of the candidate quietly, when the relationship was too brief to be meaningful, or when the candidate has hand-picked a friendly colleague instead of a direct manager. All three produce glowing calls that tell you nothing.
A manager who wants an underperformer off their team has every reason to give a warm, brief, non-committal reference rather than raise a formal issue, particularly in jurisdictions where defamation exposure makes candour risky. That is why the specificity test matters more than the tone: a vague positive is worth less than a detailed negative, because detail is harder to fabricate than enthusiasm.
Cross-checking against a second referee, ideally someone the candidate did not nominate directly but who worked alongside them, catches most of this. It is slower. It is also the difference between a reference check that protects a placement and one that just protects the recruiter’s conscience.
What are the red flags recruiters should escalate?
The clearest red flags are inconsistency between the candidate’s own account and the referee’s, reluctance to confirm basic facts like dates or reason for leaving, and a referee who cannot describe any specific piece of work. Any one of these on its own is worth a follow-up call before the candidate goes forward.
None of these are automatic disqualifiers. People leave jobs for messy, human reasons that do not map cleanly onto a reference form, and a nervous referee is not evidence of anything by itself. The point of a red flag is to trigger one more question, not one automatic rejection. Recruiters who over-index on a single hesitant reference lose good candidates as often as they catch bad ones.
What does matter is documenting the flag and the follow-up, so the client sees the full picture rather than a sanitised summary. A shortlist that survives scrutiny is worth more to a client relationship than a fast one that does not, which is the same principle behind good talent pipeline management generally: fewer, better-vetted candidates beat volume.
How should recruiters document reference checks?
Every reference check should be logged in the candidate’s record with the referee’s name, role, relationship to the candidate, date of the call, and a summary of what was asked and answered, not just a pass or fail flag. That record is what protects the agency if a placement is later disputed.
This is unglamorous work, and it is exactly the kind of step that gets skipped when recruiters are juggling twelve live roles. A structured candidate record, built once and updated consistently, turns a fifteen-minute call into a durable asset instead of a memory that fades by the next placement. This is one of the areas where a proper recruitment agency software setup earns its cost: a searchable, timestamped reference log beats a folder of notes every time a client asks what due diligence looked like.
This is also where sourcing and vetting workflow start to matter as much as the reference call itself. If a candidate never should have reached the shortlist, no reference check saves the placement. Yena is built around that earlier stage, helping recruiters build a cleaner shortlist before references are ever picked up, though the reference call itself still needs a human on the line asking the second question.
FAQ
How many references should a recruiter collect?
Two references minimum, ideally one direct line manager and one peer or cross-functional stakeholder. A single reference gives you one point of view; two lets you cross-check consistency, especially on collaboration and reliability, before you put a candidate in front of a client.
Can a recruiter contact a referee without the candidate’s consent in the EU?
No. Under GDPR, contacting a named referee is processing personal data about a third party, so recruiters need clear candidate consent before the call, and ideally the referee’s awareness that the check is happening, recorded in the candidate file.
What questions actually predict performance?
Behavioural questions tied to real incidents predict performance far better than yes/no character questions. Ask referees to describe a specific project, a specific conflict, and a specific deadline the candidate handled, then listen for concrete detail versus vague praise.
What is a red flag in a reference call?
The clearest red flag is a referee who answers everything in generalities and avoids specifics, especially around why the candidate left. Hesitation before confirming dates, titles, or reason for leaving is worth a direct follow-up question, not a shrug.
Should reference notes go in the ATS record?
Yes, always, with the referee’s name, role, relationship to the candidate, and date of the call. An undocumented reference is a liability if a placement goes wrong and the client asks what due diligence was actually done.
If your shortlists are already tight, a better reference process is a cheap, high-leverage upgrade. If the bigger problem is what reaches the shortlist in the first place, that is the part Yena is built to fix — start a free account and see what your next shortlist looks like before you pick up the phone to a single referee.