We Reject 9 Out of 10 Developer Candidates in Dubai: the 7 Steps That Stopped It Costing Us Rehires

A backend engineer we declined in March came back through a referral in September, took a different role on the same team and is now one of our stronger hires. That outcome was luck. I know it was luck because I went and read the message we sent him in March, and it was four lines of template that did not mention the role, the stage he reached or the work he had done. He came back in spite of us, not because of us.

Why Rejection Is Your Largest Hiring Channel

Run the arithmetic on a single Dubai engineering requisition. You screen somewhere between forty and a hundred applications, interview eight to fifteen people, run technical exercises with four or five, and hire one. Every person in that funnel except one receives a rejection, and most of them know other engineers in a market small enough that the good ones have overlapping networks. Your offer letter reaches one person a quarter. Your rejection reaches dozens.

The evidence on what poor handling costs is thinner than the consulting decks suggest, and worth reading with care. An older CareerArc study found that candidates left without an outcome were considerably less likely to apply to that employer again, and Starred’s benchmark work reports that candidates rejected after an assessment score far lower on candidate net promoter score than hired candidates. Both are vendor sources with a commercial interest in the finding, and the specific magnitudes move around between reports, so I would not put either number in front of a board. The direction is not controversial and matches what any recruiter sees in their own inbox: silence after an exercise is the thing candidates remember and repeat.

In the UAE two local factors sharpen it. Candidates are frequently managing a visa and a notice period simultaneously, so an unanswered process is not merely rude, it holds up a decision with immigration consequences. And a large share of senior engineering hiring here runs through referral and reputation rather than job boards, which means the opinion of a declined senior candidate propagates through exactly the network you will be sourcing from next quarter.

The 7 Steps, In Order

1. Set the rejection clock before the role opens

Agree a maximum number of working days between each stage and the decline, and write it into the hiring plan next to the salary band. We use two working days from the internal decision for anyone who was interviewed, and five working days from application for a screening rejection. The specific numbers matter less than the fact that they exist in writing before anyone is under pressure.

The reason to commit in advance is that delay is almost never a decision. The decision was made in the debrief; what followed was a queue, because writing a decline is unpleasant and nothing in anyone’s calendar forces it. A number in the plan converts an unpleasant task into an overdue one, which is the only reliable way it gets done.

2. Decide feedback depth by stage, not by mood

Fix how much detail each stage earns, and apply it uniformly. Ours runs: application screen gets a short, courteous close with no detail; first conversation gets one sentence naming the gap; technical exercise or panel gets three to five sentences of specific, evidence-based feedback and an offer to answer one follow up question.

The proportionality is the point. A candidate who spent six to eight hours on a take home and receives the same four-line template as an unscreened applicant has learned something true about how the company values other people’s time. Uniform rules also remove the failure mode where the candidate who was most pleasant gets the most feedback, which is how inconsistency enters a process that you otherwise work hard to keep structured.

3. Write the reason from the scorecard, never from impressions

Open the written assessment and quote it. Name the capability you were testing, state what you observed, and stop. For example: the role requires owning database schema changes in production, and in the exercise the migration plan had no rollback path, so we have gone with a candidate who had done that specific work before.

Two things happen when feedback is anchored to a scorecard. The message becomes genuinely useful, because the candidate can act on it. And your process becomes defensible, because the reason recorded internally and the reason sent externally are the same sentence. If you do not run structured scorecards, this step is impossible and the previous two are cosmetic; the groundwork is in our notes on building a technical interview scorecard and on structuring the interview process.

4. Strip the four phrases that create risk

Remove these from every template and every improvised message:

• Culture fit. It conveys no information and reads as a judgement about the person. If you mean the candidate argued with the interviewer about a design decision and would not engage with the counterargument, say that, because it is about observable behaviour at work.
• Overqualified. This almost always means the band was too low. Say that the level or the budget was not right for the scope, which is honest and leaves the door open for a role where it is.
• Anything touching nationality, age, gender, marital or family status. These have no place in a hiring decision and should not appear in your written record. The Ministry of Human Resources and Emiratisation publishes the UAE employment framework, and discrimination provisions are part of it. Treat your rejection archive as a document somebody may one day read.
• Any promise you have not diarised. Keeping someone in mind for future openings is a promise. If it is not written down with a date, it is a pleasantry that will be remembered as a commitment, and an unkept commitment is worse than a clean close.

5. Tag the silver medallists and set a date

At the end of every requisition, name the two or three candidates you would hire for a different role, a different level or a different quarter. Record three fields and no more: the name, the one-line reason they were wrong for this specific role, and a date in the calendar to make contact. Then actually diarise it, with an owner.

This is the step that pays for all the others, and it is the one that gets skipped because its benefit arrives one requisition later. A silver medallist is a candidate who is already screened, already understands your product, and has already decided they wanted to work with you. Compare that with starting a fresh search: new sourcing, new screening, new scheduling, and a stranger’s motivation to guess at. If you fill even one role a year from this list, the whole process has paid for itself several times over, which is roughly the economics of keeping a known bench available rather than rebuilding one each time.

6. Send it yourself if they met a human

Anyone who spoke to a person on your team gets the decline from a named person at the company, with a reply-to address a human reads. Screening rejections can be automated; interview rejections cannot, and the distinction is obvious to the candidate.

There is a practical reason beyond courtesy. A named sender invites a reply, and the replies are where you find out what your process actually looks like from the outside. Perhaps three in ten write back; some ask a question, a few argue, and occasionally one tells you something about your interview that you needed to hear. An automated address collects none of that, which makes it feel efficient while quietly removing your only feedback loop.

7. Measure the only two numbers that matter

Track two things per requisition. First, working days from the last interview to the decline, per candidate. Second, the share of rejected candidates who respond, reapply or accept a later approach. Review both in the hiring retrospective alongside time-to-hire.

Nearly every team measures the funnel down to the offer and measures nothing after the decline, which is why rejection quality drifts without anyone noticing. These two numbers are cheap to collect, they move when you change behaviour, and the second one is the closest thing you have to a measurement of whether the market thinks well of you. The same logic applies at the other end of the employment relationship, which our notes on engineering exit interviews cover.

What the Message Actually Looks Like

Templates in articles are usually useless because they are written to be inoffensive rather than specific. Here is the shape we use for a post-exercise decline, with the variable parts marked so it is clear where the work is.

Thank you for the time you put into the [exercise name] last week. We have decided to move forward with another candidate for the [role] position. The deciding factor was [specific capability the role requires]: in your submission, [observable, factual description of what was there or missing], and the person we have offered the role to had done that work in production before. Everything else in your process was strong, in particular [genuine specific]. If it is useful, I am happy to answer one question about the exercise. Written by [name], [title], and you can reply directly to this address.

The three load-bearing elements are the named deciding factor, the factual description with no adjectives about the person, and the named sender. Everything else is courtesy. Note what is absent: no culture fit, no overqualified, no hope about future roles, and no apology for the decision itself, which candidates find patronising rather than kind.

Where This Breaks in Practice

Three failure modes, all of which we have produced ourselves.

The hiring manager outsources the message and the specificity dies. When the decline is drafted by someone who was not in the room, step 3 degrades into a paraphrase and the candidate can tell. The fix is not more process, it is a one-line note from the interviewer in the debrief, written while the evidence is fresh, which whoever sends the message quotes verbatim.

Feedback turns into negotiation. Occasionally a candidate uses the reply to relitigate the decision. The boundary that works is to answer one question about the assessment and decline to reopen the outcome, politely and once. Setting that expectation in the message itself, by offering one question rather than an open conversation, prevents most of it.

The silver medallist list rots. A list without an owner and a date is a document nobody opens. We lost a strong data engineer this way: tagged in April, nobody diarised it, and by the time a suitable role opened in August she had signed elsewhere. The note existed. The calendar entry did not, and only one of those two things actually does anything.

FAQ: Declining Developer Candidates in the UAE