Why references are not enough: the case for background checks

The reference call feels like due diligence. You dial the number, ask a few questions, hear warmth in the referee's voice, and hang up feeling satisfied. What you have actually done is listen to a selected, willing advocate describe someone who agreed to be described. That is not a verification. It is a testimony, and like all testimonies, it is shaped by the interests, the memory, and the relationship of the person giving it.

This is not a reason to abandon references. It is a reason to understand what they can and cannot tell you — and to build your process around both.

A reference has two embedded problems that no amount of careful questioning can fully overcome. The first is selection: the worker chooses who to offer as a referee. No rational person nominates someone likely to speak poorly of them. This is not dishonesty; it is the natural behaviour of anyone being evaluated. The result is that the population of people who give references is systematically skewed toward those who have positive views of the candidate, or who have been persuaded to give a positive account.

The second problem is interest. The referee has a relationship with the worker — often a personal one, sometimes an ongoing one. They may feel loyalty, affection, or obligation. They may be providing the reference as a favour. They may feel uncomfortable raising concerns they were not comfortable raising directly when the employment ended. In many cases, the most significant incidents in an employment relationship are also the least likely to be disclosed in a reference call, because they are the incidents that most damaged the relationship, and damaged relationships do not produce forthcoming referees.

▎ A reference is a story told by someone with an interest in the outcome. A background check is a record.

None of this means references are worthless. A thoughtful reference call with a candid, direct former employer can tell you something about a person's temperament, their reliability over time, their particular strengths, and how they handled difficult situations. If the referee is specific — if they can give you examples, describe how the worker responded when something went wrong, tell you about the relationship between the worker and other members of the household — you are getting real information. Vagueness is the tell: a referee who describes someone as "wonderful" and "very trustworthy" without any supporting detail is not giving you evidence; they are giving you an impression, and impressions are easily manufactured.

References also carry meaningful negative information, but it is often encoded. A referee who is conspicuously brief, who cannot remember specific details about a working relationship they apparently maintained for years, or who responds to probing questions by retreating to generalities is communicating something. The reluctant reference is frequently more informative than the enthusiastic one.

A reference cannot tell you whether the worker has a criminal record. This is the most consequential limitation, and it is an absolute one. A referee who knew their former employee well and found them entirely trustworthy has no insight into whether that person has a criminal history — in the jurisdiction where they worked for that referee, or anywhere else. A criminal record held by the Department of Justice is not visible to a former employer, and it is not something most workers will voluntarily disclose. It is available through the South African Police Service's criminal record verification process, and only through that process.

A reference cannot reliably confirm that the person standing in front of you is who they say they are. Identity fraud in the domestic employment sector is uncommon but not rare. A reference from a previous employer confirms that someone named by the worker worked for that employer; it does not confirm that the person you are hiring is the person to whom that name and that history actually belong. Identity verification requires matching the person against their identity document and confirming that document against the Department of Home Affairs' National Population Register.

A reference cannot confirm employment history with the rigour that matters. If a worker tells you they were employed by a household for three years and gives you a contact number, you have no way of knowing whether the number belongs to that household, whether the person who answers it is who they say they are, or whether the account they give you reflects what actually occurred. Independent employment verification — which contacts previous employers through channels the worker did not provide, and asks structured questions about the nature, duration, and end of the employment relationship — is a different activity from reference checking.

A properly conducted background check assembles information from independent sources that the candidate cannot curate. A criminal record check through an accredited bureau queries the SAPS criminal record database and returns a verified result. An identity verification check matches the candidate's biographic information against the Department of Home Affairs' records and returns a confirmation that the identity document is genuine and belongs to the person presenting it. An employment verification check contacts previous employers through independent channels to confirm dates of employment, position held, and reason for departure.

Each of these checks produces a record, not a testimony. The record does not have an interest in the outcome. It does not soften its findings out of loyalty or obligation. It reflects what the relevant authority's database contains, at the moment the query is made, with whatever limitations that database has.

Those limitations are real. No verification system is exhaustive. A criminal record check reflects convictions, not arrests or investigations. Employment verification reflects what a previous employer is willing to confirm, not everything that occurred. An identity verification confirms a document is genuine; it does not confirm every claim the person makes about themselves. These constraints are not reasons to dismiss background checks — they are reasons to understand them accurately, which is different from understanding them romantically.

References and background checks are not in competition. They answer different questions, and the most complete picture of a prospective employee comes from using both. A background check tells you what the records show. A reference call, conducted carefully with a candid referee, tells you what working with this person is actually like — and that qualitative dimension is genuinely valuable when hiring someone who will be in your home.

The error is to use a reference call as a substitute for verification rather than a complement to it. Employers who do this are not cutting corners consciously; they are simply not accounting for the structural limitations of a reference, because those limitations are rarely made explicit. The phone call goes well, the person seems trustworthy, and the hiring decision is made on the basis of a selected, interested testimony rather than a record.

The gap between those two things is not always consequential. Many employment relationships formed on the basis of references alone go perfectly well. But when the gap matters — when there was something in the record that the reference call could never have surfaced — it tends to matter significantly. The cost of a criminal record check, conducted before hiring rather than after a problem arises, is modest. The cost of not having conducted one, in the worst cases, is not.

The framing of references versus background checks sometimes implies a power dynamic that does not serve the broader conversation well. Verification is not inherently adversarial. Workers who have nothing concerning in their records typically benefit from being verifiable — it distinguishes them from those who are not, and it gives employers a basis for confidence that a reference call cannot fully provide.

A process that is transparent — that tells the worker what checks will be conducted, what they will look for, and how the results will be used — is both better practice under POPIA and more respectful of the worker's dignity. The domestic employment sector in South Africa has historically operated informally, with the risks of that informality falling disproportionately on workers. A verification culture that is fair, clear, and consistently applied is better for everyone than a system that uses references to create a false sense of security.