Referrals should open the door - not decide who gets through it

By Tom Stroud

AI is weakening the CV. But replacing it with “who you know” would be a mistake.

The Financial Times recently reported that employers are returning to personal recommendations and referrals as AI-generated applications become more polished, more numerous and increasingly difficult to distinguish.

It is an understandable response.

When almost any candidate can produce a well-written, carefully tailored CV, the quality of the document tells us less about the quality of the person behind it. A recommendation from someone we trust can feel like a stronger signal.

At Tazio, a number of our recent hires have come through referrals from our own network. Those introductions have helped us find excellent people, often more quickly than a conventional recruitment process would have done.

But they have also reinforced an important distinction:

A referral can be a good way to find a candidate. It is not evidence that they are the best candidate.

The risk is not the referral itself

Recommendations carry useful information.

The person making the introduction may know how someone works, how they respond under pressure or whether they can be relied upon. They are also putting some of their own credibility behind the recommendation.

That has value.

The problem comes when trust in the person making the referral quietly becomes assumed confidence in the person being referred. “Someone I trust knows this candidate” becomes “this candidate is probably stronger”. Those are not the same thing.

A warm introduction should earn someone consideration. It should not give them an advantage in the final decision.

This is also a social mobility issue

Professional networks are not evenly distributed.

Some people grow up surrounded by people who can make introductions, explain how industries work and put their name forward for opportunities. Others may have exactly the same ability but no access to those networks.

A career changer may have the capability but the wrong contacts. A returner may have lost touch with former colleagues. A young person from a lower-income background may have no one senior enough to recommend them.

Their lack of a referral tells us nothing about their ability to do the job. It tells us something about the opportunities and relationships they have had access to so far.

Networks also tend to reproduce themselves. People often know others who attended similar universities, worked for similar organisations or followed similar career paths.

That does not mean referred candidates are weak. Many will be excellent. The risk is that equally capable people never become candidates at all.

An organisation can run a reasonably fair selection process and still produce an inequitable result if access to that process depends too heavily on who already knows whom.

In that situation, bias has not disappeared. It has simply moved further upstream.

A better model: referral plus assessment

We do not need to choose between referrals and open recruitment. We need to be clearer about the role each should play.

Referrals should help us discover candidates. Assessment should determine who progresses.

In practice, that means doing four things.

• First, referred candidates should enter the same selection process as everyone else. A recommendation should not allow someone to bypass an assessment or replace a structured evaluation with an informal conversation.
  ‍
• Second, there must be a genuinely open route into the process. Candidates should not need an existing connection to be considered.
  ‍
• Third, people should be given an early opportunity to demonstrate their capability. That could involve a realistic work sample, a role-relevant simulation, a structured interview or a combination of methods. The point is to collect evidence that relates directly to the job rather than relying on how convincingly someone describes themselves.

• Finally, organisations should measure the results. How do referred and non-referred candidates progress? Who gets hired? What happens to performance, retention and progression afterwards? Do some groups consistently drop out at a particular stage?

Referral hiring should not be defended because it feels effective. Nor should it be rejected because it feels exclusive. It should be evaluated using data.

The CV is not the only weak signal

Candidates using AI to improve their applications are responding rationally to a system that has long rewarded presentation, keyword matching and polished career narratives.

The answer is not an escalating competition to detect who used which tool to write which sentence. It is to collect better evidence.

A CV can provide context. A referral can provide an introduction. Neither should provide the verdict.

Our own experience at Tazio has reinforced the value of recommendations. It has also reinforced our responsibility to ensure that being outside our network does not become a permanent disadvantage.

The future of hiring should not be about CVs versus referrals. It should be broader access combined with better assessment. Candidates should be able to arrive through different doors and still be judged against the same relevant evidence.

So perhaps the question for employers is this:

Should a referral earn someone an opportunity - or an advantage?

If you want to identify the best talent, based on their potential rather than who they know, talk to us about our objective assessments or connect with me on LinkedIn.