# Reference Checks That Actually Mean Something

> Stop wasting time on nice references. Learn the specific questions that reveal a candidate's true performance and red flags.

- **Author:** Robin Keyen
- **Category:** rLUUKuIVRGK7Ne9YW3uA
- **Published:** 2026-04-01
- **Language:** en

## The Step Everyone Takes and Almost Nobody Does Well
Most organisations conduct reference checks. Almost all of them. It's on every hiring checklist, it's baked into every process, and it gives everyone involved a sense of due diligence.
The problem is that the vast majority of those checks produce almost nothing useful.
You call a number. Someone confirms dates and a title. You note it down. The hire goes ahead. Six months later, you're wondering why the person isn't working out — and somewhere in a previous employer's HR department, someone knew exactly why before you ever made an offer.
That's the gap this article is about.
## The Uncomfortable Numbers
Before looking at how to fix reference checks, it helps to understand how broken they currently are.
The issue isn't that reference checks don't work. It's that most of them are done in a way that makes it nearly impossible for them to work.
## Why Most Reference Checks Fail
There are three separate problems at play. They each require a different fix.
### Problem 1: The bias built into every open-ended question
This is the part that tends to surprise people, because it's counterintuitive. You'd assume that more information — longer, freer answers from referees — gives you a richer picture. The research says the opposite.
Academic studies consistently show that open-ended, unstructured reference questions produce biased results. The bias isn't always intentional. It's systematic.
- Referees use more "standout adjectives" for male candidates than female ones — directly affecting how hireable each group appears

- Female candidates receive more doubt-raising language: "she's unlikely to become a superstar, but she's solid" — Madera et al. (2019)

- Referees from similar backgrounds to candidates tend to describe them in warmer, more confident terms — a classic in-group favouritism effect

- Open-ended questions also create halo/horn effects — one strong impression colours every answer that follows


 ⚠️ Warning: If your reference checks use open-ended questions and no standardised scoring, you are almost certainly introducing bias into your hiring decisions — regardless of how carefully your interviews are designed. The reference stage undoes work done everywhere else.
### Problem 2: The "confirm dates and title only" culture
Most previous employers won't say much. And they have reasons.
Companies across Belgium and Europe have developed policies of extreme caution around references — triggered by defamation risk, discrimination claims, and a general fear of saying something that could become a legal issue. The result: referees confirm the bare minimum and say nothing that could be used against either party.
This isn't dishonesty. It's institutional self-protection. But the effect on your hiring decision is the same — you get nothing actionable.
What this looks like in practice:
- "Yes, [name] worked here from [date] to [date] in the role of [title]."

- "I can't comment beyond employment details."

- A warm-but-vague endorsement that's identical for every former employee they're asked about.


 💡 Quick tip: When a reference gives you nothing but dates and a title, that itself is data. It doesn't confirm the person is bad — but it tells you the previous employer won't vouch for them, which is worth noting. Press gently: "Is there anything you'd like to add?" Watch what they don't say.
### Problem 3: Coached references and outright fraud
The deliberate fraud problem is real and growing.
- References are coached on what to say and how to answer predictable questions

- Candidates list friends, former colleagues — anyone who'll speak positively — and label them as "direct managers"

- In some cases, fake references are entirely fabricated: numbers that go to associates rather than real previous employers


 ⚠️ Warning: The standard reference check process — calling the numbers a candidate provides — creates an obvious conflict of interest. You are asking people chosen by the candidate, who knew the questions were coming. Any signal you get has been filtered before you even start the conversation.
## What the Science Actually Says About Predictive Value
Here's where it gets useful — because reference checks, done right, are genuinely valuable.
The foundational research is the Schmidt & Hunter meta-analysis — a synthesis of 100 years of data on selection methods, covering over 85 years of published studies. It compares how well different hiring methods predict actual job performance.
The finding that matters: unstructured reference checks rank poorly as standalone predictors. But structured reference checks, particularly when combined with other tools, significantly outperform them — and climb sharply in predictive validity.
What "structured" means: the same set of job-relevant questions asked in the same order, using a consistent numerical scale, for every candidate. That's it. Simple, and dramatically more effective.
The US Office of Personnel Management — which manages hiring for the world's largest employer — classifies structured reference checking as a useful predictor of:
- Supervisory performance ratings

- Training success

- Promotion potential

- Employee turnover


There's also a more nuanced finding worth using: guilt-proneness — an individual's tendency to feel responsible when things go wrong — is one of the most predictive traits you can probe in a reference check. Research published on ResearchGate found that employees high in guilt-proneness show better job performance and even predict emerging leadership more reliably than extroversion. The implication: questions designed to surface accountability and responsibility are more valuable than questions about skills or accomplishments.
## What Good Actually Looks Like: The Practical Fix
### Ask the right people first
Not all references are equal. Direct supervisors are the most predictive sources. Colleagues and peers are significantly less so. Character references from outside the workplace are nearly worthless for predicting job performance.
When collecting references, specify: "We need at least two direct supervisors from roles held in the last five years." That single instruction already improves the quality of what you'll get.
 ✅ Best practice: If a candidate can't provide a single direct supervisor willing to give a reference, that's not a logistical problem — it's a signal. Note it and ask the candidate directly: "Why can't we speak with your previous manager?" Their answer tells you something.
### Standardise every question you ask
Build a short set of role-specific questions. Limit free-text answers. Use a numerical scale (1–5 or 1–10) for the questions that matter most. Ask every referee the same questions in the same order.
Questions that consistently produce useful signal:
- "On a scale of 1–10, how would you rate [candidate]'s ability to deliver without close supervision?"

- "If you had the opportunity to hire [candidate] again, would you? Why or why not?"

- "Can you describe a situation where [candidate] made a mistake? How did they handle it?"

- "What does [candidate] need from a manager or environment to do their best work?"

- "Is there anything about [candidate]'s working style that future managers should know about?"


The last two are often the most revealing. They invite honest answers without requiring the referee to say anything explicitly negative.
 💡 Quick tip: "Would you rehire?" is one of the most debated reference questions in hiring research. LinkedIn and Yardstick Management call it the single most actionable question. Critics argue it's too easy to give a technically honest yes while omitting critical context. Use it — but never as your only question. It's a starting point, not a conclusion.
 ✅ Best practice: Score answers immediately after each call before discussing with the team. Group scores create anchoring — the first person who speaks influences everyone else. Independent scoring first, discussion second.
### Separate the signal from the silence
Learn to read what isn't said. A referee who answers every question about skills confidently but hesitates when asked about collaboration is telling you something. A referee who gives three-word answers throughout but lights up when asked about one specific project is telling you something.
 ⚠️ Warning: Enthusiasm is contagious and misleading. A referee who is warm, funny, and clearly likes the candidate will unconsciously improve your impression of them regardless of what is actually said. If you're susceptible to this — and most people are — use a written score sheet anchored to specific answers, not overall impression.
### Consider structured digital reference checking
Automated, structured platforms like Checkster (now part of Harver) and Xref change the economics and the quality of reference checking simultaneously.
The numbers from Checkster's platform data are worth knowing:
- 24-hour turnaround on average (vs. days of phone tag)

- Average of 6 references per candidate (vs. the industry average of 2)

- Stops an estimated 10–12% of potentially bad hires before they start

- Turns 52% of references into passive candidate leads — a sourcing benefit most employers don't consider


Platforms like these — and the structured thinking behind them — are part of a broader shift in how the Belgian market is approaching candidate evaluation. Tools like Joobs, which surface verified professional context alongside candidate profiles, reflect the same direction: moving away from trusting what candidates present, and toward building a fuller, verified picture before any decision is made.
## The Belgian & European Legal Reality
Reference checks in Belgium are not just a best practice question. They're a compliance question.
Belgian background screening is among the most strictly regulated in Europe, governed by GDPR and enforced by the Belgian Data Protection Authority (Gegevensbeschermingsautoriteit). Getting this wrong has real financial consequences.
What you must do:
- Obtain written consent from the candidate before conducting any reference or background verification

- Apply data minimisation — collect only what is strictly necessary for the specific hiring decision

- Document your legitimate interest for each type of check

- Appoint a Data Protection Officer if you conduct systematic screening as part of your hiring process

- Store and handle reference data in compliance with GDPR retention limits


What you cannot do:
- Contact references not provided by the candidate without their consent

- Check social media without explicit written consent

- Ask about or investigate: political affiliation, religious beliefs, health data, genetic data, sexual orientation, or union membership

- Keep reference notes indefinitely — they must be deleted once the hiring process concludes


 ✅ Best practice: Keep a simple one-page reference check consent form signed by every candidate before you contact anyone on their list. Document what you asked, what was said, and when the data was deleted. This takes five minutes to set up and removes almost all of your legal exposure.
 ⚠️ Warning: "Off the record" reference calls — phoning a contact at a former employer without the candidate's knowledge or consent — are not legally grey in Belgium. They are a GDPR violation. The informal market intelligence they produce is not worth the risk.
## Frequently Asked Questions
If most references are coached or vague, is there any point in calling them?Yes — but shift what you're listening for. You're not just gathering endorsements. You're assessing the enthusiasm level, the specificity of answers, the hesitations, and what gets avoided. A coached reference who answers every question smoothly is still telling you something. An unusually brief call from a direct manager who "can't really say much" is too.
How many references should we ask for?Aim for three — ideally two direct supervisors and one peer or cross-functional collaborator. Three is achievable, gives you enough to cross-reference, and avoids the "confirm dates only" dead end you hit with just one or two.
Should we check references before or after making an offer?Both timings have logic. Checking before an offer gives you the chance to adjust or walk away without complications. Checking after a conditional offer is more time-efficient and signals commitment to the candidate. In Belgium, a conditional offer with reference checking as a stated condition is a clean approach — the candidate knows the process and has consented to it.
What if the candidate says their previous employer has a "no reference" policy?Accept that for the employment dates confirmation. But push for a direct manager or colleague willing to speak personally, outside the company's formal policy. Most people are willing to speak as an individual even if the company won't. If the candidate can't produce a single person willing to give a substantive reference for any previous role, that's a pattern worth exploring.
How do we handle it if a reference says something negative?Treat it as a data point, not a verdict. One negative reference from one person for one specific situation doesn't disqualify someone. Three referees all hedging on the same dimension — reliability, people management, handling pressure — does. Cross-reference before concluding.
What records do we need to keep, and for how long?Under GDPR, reference notes are personal data. You need consent to hold them, a clear retention period tied to your hiring process, and a deletion process once the role is filled or the candidate is no longer being considered. In practice: keep your score sheets until the probation period ends for the hired candidate, then delete. Consult your DPA-registered process documentation for the precise timeline your organisation has committed to.

---
*Source: [Joobs.be](https://joobs.be/en/blog/reference-checks-that-actually-mean-something)*