A Rejection Is Often a Timestamp, Not a Verdict

Within the space of a few days, I found myself discussing several candidates who had already been rejected, including some by the very companies now considering them again. After the third or fourth example, I caught myself wondering what, exactly, we had rejected the first time.

One candidate had previously applied while living abroad and preparing to relocate. By the time we spoke again, they were already here and legally able to work. Another had been considered by a highly specialized company only a few months earlier, but the company now wanted us to approach them again because its understanding of the role had evolved. A third had already received a rejection from me, but further conversations with the client changed my understanding of which backgrounds might transfer successfully into the position.

In each case, the person was largely the same. What had changed was the information around them, or the circumstances against which that information was being interpreted.

It made me think about something strange we do in recruitment. We take decisions that are highly specific to a particular moment and record them using language that makes them sound far more permanent than they really are.

Open almost any applicant tracking system and you'll find familiar labels: rejected, not selected, archived, disqualified. I understand why those labels exist. Recruitment processes need structure, and an ATS cannot contain lengthy explanations for every decision ever made.

Still, "rejected" sounds as though something conclusive has been established about a person, when the reality is usually much narrower. At one particular moment, somebody evaluated one candidate for one role, using the information available at the time, and decided not to continue.

Most of that context disappears once the decision is stored. What remains is the outcome.

The more I think about it, the more a rejection starts to look less like a verdict and more like a timestamp.

Recruitment processes assume the vacancy is the stable object and candidates are the variables. We define the requirements, people enter the process, and we assess how closely each profile matches what the company needs.

That structure is necessary, but it isn't always an accurate description of what happens once a search begins.

Jobs change. Budgets change. Hiring managers learn from the candidates they meet. A requirement described as essential becomes negotiable because the market has shown that the original profile barely exists. An unconventional candidate makes somebody realize that a background they hadn't considered might actually solve the problem.

Sometimes the company simply understands the role better after trying to recruit for it.

I've become increasingly cautious about how confidently we use the word fit. We describe somebody as a good or bad fit as though it's an attribute that belongs to the candidate. In reality, fit is a relationship between two things that are both moving.

The candidate changes, certainly, but so does the organisation. Priorities shift, requirements evolve, and sometimes my own interpretation as a recruiter changes too.

I've looked at the same CV twice and reached different conclusions. That used to make me uncomfortable because it felt inconsistent. Now I think the more important question is whether I can explain why my interpretation changed. Perhaps the first decision was reasonable given what I knew then, and the second is equally reasonable because I now understand the role differently.

The problem isn't that we change our minds. It's that we preserve the first conclusion far better than the reasoning behind it.

This question has become less theoretical because of the work we're doing at skeeled.

Like any ATS, skeeled tracks applications, including who applied, how they progressed and whether they were hired or rejected. But we're working on something I find more interesting: making accumulated candidate information useful beyond the vacancy that originally created it.

One important shift is moving from thinking primarily about applications to thinking about candidates.

If somebody applied to three positions over two years, those are three applications, but they're still one person. When another opportunity opens, I want to understand who that person is, what we already know about them and whether previous decisions should still influence how we assess them.

Semantic search helps us find relevant candidates even when their CV uses different terminology. But finding someone is only the beginning. The harder question is whether a system can understand why an experienced recruiter would consider that person relevant.

Our recruitment team has been evaluating candidates across real vacancies to test AI matching against recruiter judgment. We started with dimensions such as experience, education and languages, but quickly encountered the complexity behind apparently straightforward criteria.

Does experience in one industry transfer into another? Does somebody who lacks the requested number of years already operate at the necessary level? Is the real obstacle their professional background, or something else entirely, such as location, salary expectations or work authorization?

Recruiters make these judgments constantly, often almost automatically. The challenge isn't simply teaching a system to reproduce a final score. It's preserving more of the thinking underneath it.

Because if we can do that, a candidate database starts becoming something different from a warehouse of CVs and historical statuses.

It starts becoming a form of organizational memory.

This has influenced something much more mundane in how I ask my team to work.

When we disqualify a candidate, I want the actual reason recorded rather than whichever generic option makes it quickest to remove them from the pipeline.

"We reviewed 100 candidates and rejected 80" tells me almost nothing.

But if those 80 include people missing a particular language, candidates whose salary expectations exceeded the range and others whose experience wasn't transferable enough, the process suddenly tells us something useful.

Maybe our sourcing needs to change. Maybe the salary is wrong. Maybe the requirements are too narrow. Or perhaps the candidates exist, but the opportunity isn't attractive enough.

That reasoning helps us understand the current search, but it becomes even more valuable when somebody appears again later.

Some rejection reasons have a long shelf life. A missing legally required qualification matters, as does evidence of behaviour that raises legitimate concerns for similar positions.

Other reasons decay surprisingly quickly.

Location changes. Work authorization changes. Salary expectations move. Experience accumulates. A company discovers that something it originally described as essential was actually a preference.

A candidate who was too junior two years ago may now be exactly right. Someone who needed relocation support may already live nearby. A finalist who narrowly lost out might be the first person worth contacting when another position opens.

It makes me wonder whether some recruitment decisions should almost come with an expiry date. Not because a system should automatically reverse them, but because we shouldn't assume the circumstances that made them valid will remain unchanged forever.

The obvious AI question is, "Who matches this job?"

I'm increasingly interested in a different one: who have we already met whose previous rejection no longer tells us very much, and what did we learn about them that's still worth remembering?

There's another reason this matters to me, and it has little to do with technology.

A recruiter can click "reject", move on to the next profile and perhaps review another fifty applications before the end of the afternoon. The candidate receives one email, and their experience of that decision is understandably much more personal.

The natural interpretation is simple: they don't want me.

Sometimes that's broadly accurate. But often the decision was narrower. Another applicant might have been slightly stronger. The budget may have changed. The company may have misunderstood part of the profile, or relocation support wasn't possible at that moment.

The rejection is real, but the meaning somebody attaches to it can become much bigger than the decision itself.

I was reminded of this when a candidate replied to one of our rejection emails asking what, specifically, had prevented them from progressing. They weren't challenging the decision. They wanted to understand it.

I've spent years arguing that the minimum standard in recruitment is to close the loop with candidates. A significant part of what eventually became The Humane Recruiter grew out of seeing candidates thank me simply for giving them an answer.

I still believe that matters. But I'm starting to question whether I've sometimes treated the answer as the end of our responsibility when the reasoning might be the more useful part.

A rejection email tells somebody the process is finished. An explanation might help them understand whether there's something worth improving, whether they were genuinely far from the requirements, or whether another candidate was simply better matched.

Of course, meaningful individual feedback isn't always realistic. Recruiters manage large volumes, and some information cannot appropriately be shared. But those limitations shouldn't stop us from being more thoughtful about what we record internally and what we communicate when we can.

The same explanation that helps a candidate may also help us. Six months later, I could be the recruiter looking at that profile again, wondering what my previous decision was based on.

We will always need a rejection button. I just don't want the button to become the most important thing we remember.

A rejection tells us something about a candidate, a role and a particular moment in time. Sometimes that decision remains valid. Sometimes the circumstances change surprisingly quickly.

As our technology becomes better at remembering everyone we've considered, I hope we also become better at remembering enough about why we said no to recognize when that no no longer means what it once did.

If you've made it this far, thank you. Sincerely. I'm grateful you're here.

If anything made you think, react or question, let me know by responding directly to this email.

Until next time 🍁