The Candidate the System Couldn't See

Over the last few months, I’ve been helping a former colleague find his next role. He’d returned from six months of travelling to a job market that was proving much harder to navigate than either of us had expected. At one point, he was submitting thirty or forty applications a week, often receiving little more than a generic rejection. The repetition was dispiriting, particularly because I knew from working alongside him that he had considerably more to offer than his previous job title suggested.

I had seen him tackle unfamiliar problems, learn quickly and take responsibility beyond the formal boundaries of his role. That kind of capability is difficult to describe convincingly on a CV, and harder still to recognise through a process designed to reduce hundreds of applications to a manageable shortlist. It made me think about the difference between what recruitment systems can readily observe and what employers are actually trying to find.

What experience can and cannot tell us

Most recruitment processes understandably begin with evidence of previous experience. Job titles, years in a particular discipline, qualifications and familiar employers all offer useful signals, especially when recruiters have limited time. But experience and capability aren’t interchangeable. A person may have spent years in a role without developing the judgement needed to move beyond it, while someone with a less conventional background may already be operating well above the level their CV implies.

My former colleague was applying for positions that looked like a considerable step up from his previous role as an Operational Performance Analyst. I understood why that might raise questions on paper. What concerned me was whether he would ever get the opportunity to answer those questions. An application form can record the work someone has done, but it rarely captures how they think through uncertainty, respond to challenge or approach something they haven’t encountered before.

It’s tempting to blame automated recruitment for this, particularly when the rejection emails all look the same. But we didn’t know how each employer had assessed his application, or whether an automated system had made the decision at all. Human recruiters can be just as influenced by familiar job titles, credentials and assumptions about career progression. The problem isn’t exclusive to AI. It’s the risk of treating convenient indicators as though they represent the whole person.

Giving someone the chance to be understood

Eventually, I suggested changing the approach. Rather than concentrating almost entirely on applications, he began making contact with recruiters and having conversations. The intention wasn’t to sidestep a fair recruitment process. It was to create an opportunity for someone to understand what he could contribute beyond the evidence captured in a standard application.

That change led to an interview for a Senior Data Consultant position. On paper, the move from Operational Performance Analyst looked ambitious. Having worked with him, I didn’t see it as an unreasonable leap. I saw someone whose ability had developed faster than the labels used to describe his experience.

We worked together on his CV, interview preparation and a case study presentation. The case study was particularly revealing because it invited the kind of response that technical people can be quick to produce: an architecture, a platform recommendation and a set of decisions based on very limited information. We discussed why a better consulting response would begin with curiosity. What was the organisation actually trying to achieve? What assumptions were being made? Who would need to be involved before a responsible recommendation could be offered?

The presentation became less about demonstrating that he knew which technologies to choose and more about showing how he would approach an unfamiliar business problem. Shortly afterwards, he received a job offer.

The limits of what we make visible

I don’t take that outcome as proof that every unsuccessful applicant has been overlooked, or that speaking to a recruiter will solve the problems of a difficult labour market. What it did reinforce was something I encounter repeatedly in data and technology: the information we choose to capture shapes the conclusions we’re able to reach.

Recruitment systems need structure, and organisations need practical ways to assess candidates consistently. But consistency is not the same as completeness. When we rely too heavily on what can be classified and compared, we risk overlooking qualities that emerge through discussion, problem solving and experience of working with someone.

This is also a useful challenge for those of us building analytical and AI systems. A system may be perfectly accurate in processing the evidence available to it while remaining unable to recognise something important that the evidence doesn’t describe. That doesn’t make the system useless. It means we need to understand its limits and design opportunities for those limits to be challenged.

My colleague didn’t become capable when he was offered the job. He had already demonstrated that capability in his previous work. What changed was that somebody finally had the opportunity to see it.