Programme manager interview questions: benefits, boards, and hard stops

Our street-lighting programme — 74,000 lanterns converting to LED across the county — had four workstreams, and I stopped the smart-controls one at month nine with £600,000 committed... let me say that more precisely, because the distinction mattered in the boardroom: £600,000 committed, of which £380,000 was already spent and £220,000 was recoverable by stopping.

The trigger was benefits arithmetic, not delivery failure — which is what made it hard to sell. The workstream was actually running to plan. But the central-management system's business case rested on dimming-schedule savings priced against 2023 energy contracts, and our new energy deal plus the LED conversion itself had eaten most of the saving the smart controls were supposed to harvest. I had the benefits team re-run the case quarterly — a discipline I'd instituted at programme start, and this is why — and by the third re-run the payback period had gone from six years to nineteen. Nobody disputed the arithmetic; the dispute was whether arithmetic should beat momentum.

The politics were the real workstream: the portfolio holder had announced smart streetlighting in the local press, and the supplier had a reference-site clause in the contract. I gave the board a one-page options paper — continue (nineteen-year payback, stated plainly), pause pending an energy-market review (my recommendation: £220,000 preserved, contract novated to a pilot corridor of 800 lanterns so the announcement stayed technically true and the supplier kept a reference site), or full stop with exit costs. The board took the pause-and-pilot inside twenty minutes, because the paper had already had the argument with itself.

The team was the part I refused to botch: the four engineers moved to the conversion workstream's acceleration — which is where the freed £220,000 went, buying eleven weeks off the main schedule — and I stood in front of them and said the stop was a success of our own quarterly discipline, not a failure of their delivery. Two of them later told me that framing was the difference between a redeployment and a demoralisation.

What I'd have you take from it: programmes don't get stopped by dashboards. They get stopped by someone deciding the benefits case outranks the announcement — and building the re-run discipline BEFORE anyone needs it, because you cannot install objectivity mid-argument.

Our new mobilisation system — the one that turns a 999 call into appliances moving — had a contractual go-live date, and six months out it became clear we could hit it. That sentence sounds like good news, and the programme board initially heard it that way. The problem: we could hit it only by going live before the fallback-procedures training and the station-end drills were complete — and the programme's benefits case wasn't 'new system installed', it was 'call-to-mobilisation time down by 40 seconds with NO degradation during transition'. An untrained go-live protected the date by spending the benefit — and in this service, degradation during transition is measured in something worse than seconds.

What I did first was make the collision undeniable rather than arguable: I had the training lead and the ops assurance lead jointly present the readiness picture to the board — not me, THEM, because benefits need their owners' voices, and a programme manager defending 'more training time' alone sounds like schedule cover. Their evidence: 14 of 31 stations drilled, fallback procedures tested at two of five control-room positions.

Then the options paper, in the board's own currency: go live on the date with graduated risk (stated plainly, including the worst case and who would carry it); slip eight weeks with the supplier's delay costs — £95,000 — and full drill coverage; or my recommendation, a phased cutover: go live on the contractual date for the three drilled station groups — which satisfied the contract's operational-commencement clause, our commercial team confirmed — with the remaining groups following in two waves as drills completed, full transition done at week seven.

The board took the phased path. The chief fire officer's question — 'why am I only hearing the training picture now?' — was fair, and my honest answer was that the programme's own reporting had been date-weighted until that quarter; I'd changed the board pack two months earlier precisely because benefits readiness had been arriving as an appendix. That change is the part I'd carry to this role: the front page of a programme report should be benefit readiness, not milestone RAG — because boards steer by whatever you put in front of them.

Full transition completed at week seven; call-to-mobilisation improvement measured at 43 seconds by month three; zero degradation incidents during cutover. The date, as contracted, was technically hit — but nobody in that boardroom believes anymore that the date was ever the point.

The merger of two further-education colleges — and the honest headline is that the two principals had signed the same merger for different reasons. One saw a rescue: consolidate estates, cut duplicated back-office, survive the funding climate. The other saw an expansion: a combined curriculum reaching employers neither college could serve alone. Same programme name, two incompatible programmes — and every workstream decision (which campus hosts engineering, which finance system survives, whose T-level model won?) became a proxy war between the two theories.

My read after the first month: I couldn't resolve the strategic question — that genuinely belonged to the principals and the incoming board — but I could stop every operational decision from re-litigating it. So I built the programme around that separation.

First, I forced the strategic fork into daylight: a two-page paper for the joint steering group that said, in effect, 'these are the eleven upcoming decisions that land differently under rescue versus growth — the programme needs the board to pick the frame, or every one of these becomes a six-week argument'. Uncomfortable meeting; the chair later called it the most useful paper of the merger. The board chose a sequenced frame — consolidation first year, growth investments gated on hitting the savings milestones — which had the enormous virtue of giving BOTH principals a version of events they could stand behind publicly.

Second, the decision rules got mechanised: every workstream paper thereafter carried a one-line header — 'consistent with the sequenced frame: yes/no/board-referral' — signed by the workstream lead. Eleven anticipated proxy wars became two genuine board referrals; the other nine resolved at workstream level because the frame did the arguing.

Third, the relationship maintenance nobody writes down: I briefed both principals separately before every steering group, same content, no surprises — because executives ambushed in front of each other stop being aligned regardless of what any framework says.

The merger completed with the year-one savings target hit — £2.1 million against £1.9 million planned — and the first gated growth investment (the employer-facing engineering hub) approved on schedule in year two. The lesson I'd bring here: senior alignment isn't agreement — it's a shared frame that lets people who want different futures make the same next decision.

A terminal refurbishment is a dependency lattice wearing a construction hat: our programme had five projects — structural works, security-lane replacement, retail fit-out, wayfinding, and systems — and the whole thing ran through a live terminal processing 11,000 passengers a day, which meant every dependency also had an operational guardian with veto power.

The mechanism I ran: a dependency register where every entry was a COMMITMENT, not a line — named giver, named receiver, date, and the giver's own stated confidence, re-declared monthly. That last field is the early-warning system: dates move suddenly, confidence erodes gradually, and a giver who drops from 'green' to 'green, probably' in their own words has told you something no milestone report will admit for another six weeks. Forty-plus dependencies, reviewed in a forty-five-minute monthly session where only movers and decliners got airtime.

The one that failed anyway: the security-lane supplier's equipment certification — a dependency owned outside my programme entirely, sitting with the regulator and the supplier's test house. Their confidence had been honest all along; the test-house queue was the thing nobody controlled. Certification slipped eleven weeks, and the security-lane project was the retail fit-out's predecessor: no new lanes, no hoarding removal, no retail handover — with retail's income commitments to the airport starting on a contractual date. A £700,000-class collision if taken passively.

The recovery, in order: I knew at slip-plus-two-days because the confidence field had flagged the test house a month earlier and I'd asked for weekly declarations — so we spent our response time redesigning rather than discovering. The redesign: resequence the retail fit-out to start in the two units NOT dependent on the lane hoarding — the fit-out contractor re-phased for £40,000 — and negotiate with the incoming retailers to shift the income-commitment trigger from 'access date' to 'trading-ready date', which their own delayed shopfitting schedules made painless in practice. Programme end date: held. The £700,000 exposure landed as £40,000 of re-phasing plus a fortnight of my commercial manager's diary.

What the failure taught the register: external dependencies with regulatory bodies in the chain now carry a mandatory independent verification — we ring the test house ourselves, politely, monthly. Trust the giver's honesty; never inherit their blind spots.