I Hired 3 Delivery Managers Before One Stuck: the 7 Steps I Use Now in Dubai

The first two delivery managers I hired were, by any reasonable measure, good at their jobs. Our dates still moved. It took me an embarrassingly long time to work out that I had written the requisition for a narrator and then been disappointed that the story did not change. The third hire worked, and the only thing I did differently was decide, in one sentence, what she owned that nobody else did. Everything below is the method that came out of getting it wrong twice.

Step 1: Write the Single Accountability Sentence

Before any job advert, write one sentence in this shape: this person owns X, and if X goes wrong it is their name on it. Then show that sentence to your tech lead and your product owner and ask whether either of them thought they owned X.

The sentence that worked for us was: owns whether the dates we give the business are true, and owns telling the business early when they stop being true. Notice what is not in there. Not the roadmap, which belongs to product. Not technical design, which belongs to engineering. Not people management, which can sit anywhere but should be decided explicitly. The role is the truth of commitments and the timeliness of bad news.

If you cannot write a sentence that survives contact with your existing leads, you probably do not have a vacancy. You have an unresolved argument about who decides scope, and hiring someone into the middle of it is how you burn eighteen months. I have watched that happen to two companies in Dubai this year, both of whom believed they were solving a capacity problem.

Step 2: Choose the Shape Before the Seniority

There are three shapes that work in this market, and the decision is driven by how many parallel streams you run rather than by headcount.

The embedded option deserves a warning, because it is the one that looks most efficient on a budget line. If the person who tells you whether the date is realistic is paid by the company that set the date, you have not bought delivery management. You have bought a second account manager. Use it when you genuinely want to adopt the vendor’s process, and never when the relationship is already strained.

Step 3: Specify Three Week-One Artefacts, Not Years of Experience

Replace the experience paragraph in your job description with the three documents you expect in the first fortnight. Ours were:

1. A dependency map of everything currently blocking or about to block a stream, with a named owner against each item. Not a risk register with probabilities. Names and dates.
2. A forecast for the next two releases expressed as a range with a stated confidence, plus the two assumptions most likely to break it.
3. An escalation path that says who gets told what, within how many hours, when a commitment is at risk. Written down, circulated, and agreed by the people named in it.

This does two useful things. Strong candidates read it and get specific immediately, often pushing back on the second item before you have made an offer, which is exactly the behaviour you want. Weaker candidates read it and talk about frameworks. You will know inside one screening call.

It also quietly solves the seniority question. Someone who has produced those three artefacts under pressure has the experience, whatever their title history says. In this market that opens up candidates who ran delivery inside a full-stack product team without ever carrying the job title, and they are consistently better value than people whose CV is a list of certifications.

Step 4: Screen on a Real Slippage, Not a Methodology

One question does most of the work in the first interview: tell me about a date that was missed on your watch, when you first had a reliable signal it was going, and what you did between that signal and telling the stakeholder.

What you are listening for is the gap. Everyone has missed dates. The useful information is how many weeks sat between the first honest signal and the first honest conversation, and whether they volunteer that number without being pushed. The best answer I have heard was blunt: three weeks, because she had convinced herself a vendor would recover, and she now escalates on the signal rather than on the conviction.

Three answers that should end the process. A candidate who has never missed a date, which means they have only run work with soft commitments. A candidate who locates the entire cause outside their control, every time, with no decision of their own in the story. And a candidate who answers with a method rather than an incident, because when the pressure is real, methods are the first thing to go.

Step 5: Run the Forecast Exercise on Your Own Data

This is the step that replaced three rounds of interviews for us, and it is simple. Take three real sprints from your own team. Strip the names and the feature titles, keep the throughput, the carry-over, the interruptions and the one week where everything went sideways. Send it to the candidate two days before the interview with one question: when will the next twelve items be done, and how confident are you?

Then spend the interview on their reasoning. The signals that matter:

• Do they give a range, or a single date? A single date from three sprints of noisy data is either naivety or theatre.
• Do they notice the bad week and ask whether it was a one-off or a pattern? The question is more valuable than any answer they could give.
• Do they state assumptions you can check later, or assumptions phrased so they can never be wrong?
• Do they ask what the date is for? A forecast for a board meeting and a forecast for a customer contract carry different risk, and good candidates want to know which one they are signing.

Nobody passes this perfectly and that is the point. You are hiring the quality of the reasoning, not the accuracy of the number. If you want a reference point for what the underlying planning discipline looks like when it is working, our breakdown of project management in Dubai covers the artefacts side of it in more detail.

Step 6: Role-Play the Conversation Nobody Wants

Fifteen minutes, in the final stage, with whoever in your business is most likely to push back on a date. You play yourself. The brief to the candidate: the release will be six weeks late, you have known for two days, tell me.

Then push. Ask for the original date anyway. Offer to add two engineers. Suggest cutting testing. Imply that the last person in the role managed it. What you are measuring is whether they hold a position they can defend while still sounding like someone you would want in a room with a client.

The failure modes are informative. Some candidates collapse and re-commit to the date in the room, which tells you precisely what will happen in month three. Some go rigid and process-bound, quoting a change control procedure to a frustrated executive, which is technically correct and commercially useless. The ones worth hiring do something harder: they refuse the date, offer two concrete alternatives with their consequences, and leave the choice with you.

Step 7: Agree the 90 Day Scorecard Before They Sign

Write down what good looks like at 30, 60 and 90 days, and write down what would make this the wrong hire. Then give it to the candidate before the offer, not after.

Ours reads roughly like this. By day 30: the dependency map exists and the escalation path has been agreed by the people named in it. By day 60: two forecasts have been issued with ranges, and at least one has been revised in public with a reason. By day 90: a date has been successfully defended against pressure, or renegotiated early enough that nobody was surprised.

That last one is the real criterion and it is uncomfortable to write down, because it means a good delivery manager will, within three months, tell you something you do not want to hear and be right. If your leadership team cannot sign up to that in advance, the hire will fail regardless of who you pick, and it is better to find that out now than in February.

The exit criteria matter just as much. Mine: still producing status reports rather than decisions at day 60, or escalating everything, which is the same failure wearing the opposite costume. Share those too. Candidates who are right for the job find it reassuring, and the ones who find it aggressive were going to be the narrator hire.

Three Mistakes I Would Now Pay to Avoid

Hiring the role before resolving the scope argument. Covered in step 1 and worth repeating because it is the expensive one. A delivery manager cannot arbitrate between a product owner and a tech lead who disagree about what is in the release. They can only document the disagreement more clearly.

Placing them in the engineering time zone by default. It feels right, because that is where the work happens. But the commitments are made where the stakeholders are, and somebody who learns about a date change in their morning stand-up is already a day behind the conversation. If your engineering sits outside the UAE, the delivery manager belongs in Dubai and the overlap window belongs in their calendar as a standing obligation.

Measuring them on velocity. Velocity is a planning input, not an outcome, and the moment it becomes a target you have taught a competent person to inflate estimates. Measure predictability: of the commitments made in a quarter, how many held, and of the ones that did not, how early did you know. That number is harder to game and it is the only one your business actually cares about.

FAQ: Hiring a Delivery Manager in Dubai