Operations manager interview questions: trade-offs, not tidiness

The one I'd pick cost us our biggest contract's confidence for a quarter: three consecutive missed morning deadlines on hospital linen — the one customer category where a late van isn't an apology, it's a ward problem.

Detection was the embarrassing part: I found out from the customer, not from us. Our despatch sheet showed the vans leaving on time; what it didn't show was that they were leaving on time HALF-LOADED because the finishing line was running behind, and the second run was arriving after the hospitals' receiving window. The sheet measured departure, not delivery — we'd built a metric that let us fail while showing green.

Containment first: I rang the trust's facilities manager myself, same day, took the timeline with me — no van-broke-down softening — and agreed a fortnight of daily 7am confirmations while we fixed the root cause. Costly in pride, cheap in trust.

The redesign: the finishing bottleneck traced to a folding machine running at two-thirds rate for six weeks — logged as 'awaiting parts' with no owner and no escalation. So three changes, all structural: maintenance items on customer-critical kit now carry an owner and an auto-escalation at 72 hours; the despatch metric changed from 'vans out' to 'orders receipted in window' — measuring the thing the customer experiences; and the morning huddle now starts with yesterday's receipted-late list, read aloud, every day.

Twelve months on: zero missed hospital windows, and the trust renewed. But the change I actually value is the metric one — we stopped being able to lie to ourselves by accident.

Peak week last summer, and our second bottling line started throwing intermittent fill-level variance — inside legal limits, but drifting toward the edge of our biggest customer's spec. Full stop for diagnosis meant losing roughly 40,000 units of a week we were contracted to deliver 310,000 in. Running on meant shipping product that MIGHT breach spec on a brand that wasn't ours to gamble with.

The decision needed making inside an hour, so I priced both paths in front of my production manager rather than in my head. Stopping: 40,000 units short, a contract penalty conversation, agency overtime to claw back what we could. Running: a possible spec breach that — worst case — meant a customer-side quality hold on the whole week's production, roughly seven times the exposure, plus a relationship we'd never fully rebuild.

I stopped the line — but the half I'd defend hardest is what happened next. I called the customer's supply planner BEFORE the diagnosis, not after: told them the variance, the stop, and a worst-case shortfall number by end of day. Then we found the fault — a worn filler valve seal, four hours to swap — and recovered 26,000 of the 40,000 with a weekend shift the team volunteered for once they'd seen me stop the line rather than lean on them to 'watch it closely'.

Shipped 296,000 against 310,000, paid a small pro-rated penalty, kept the spec record clean. The principle I'd state: I'll always trade a shortfall we can explain for a breach we'd have to confess. Shortfalls cost margin; breaches cost the relationship that generates the margin.

The classic one, in its purest form: our night pick operation and the daytime sales desk were at open war over late orders. Sales kept accepting orders past the 6pm cut-off — usually for their biggest accounts — and every one landed on the night shift as a resequenced pick, missed van slots, and a supervisor spending her shift firefighting instead of running the floor. Night blamed sales for selling what we couldn't pick; sales blamed night for 'inflexibility that loses customers'. Both were right, which is why it had run for two years before I picked it up.

What I did first was replace anecdote with a fortnight of data: every late order logged with its account, value, and what it displaced. The picture surprised both sides — 80-odd late orders, but two-thirds came from just four accounts, and the average late order was worth less than the picks it disrupted. The war was mostly a habit, not a commercial necessity.

Then the mechanism, agreed in one meeting with the sales manager because the data had done the arguing: a hard 6pm cut with two released exceptions — genuine same-day emergencies, capped at three per night, flagged by 7pm so the supervisor sequences them INTO the run rather than around it; and the four heavy accounts moved to a 4pm earlier cut with a small delivery-priority sweetener sales could sell as an upgrade.

My side's concession — and it mattered to the politics: night picked the three exceptions without grumbling, ever. Ninety days later: late orders down from forty a fortnight to under ten, all inside the exception lane, van slots hit, and the two managers who'd spent two years escalating about each other presented the new cut-off jointly at the sales conference.

Interfaces fail when they're personalities. They work when they're rules with a safety valve.

Second quarter last year: three of my five headline numbers went red in the same period — member attrition up, class occupancy down, and cost-per-visit rising as the fixed base spread over fewer visits. The tempting story was seasonal ('it's summer, it always dips') and my area director had half-accepted that framing before I had.

What I did first was refuse my own comfortable explanation. Seasonality was checkable: I pulled the same quarter across three prior years, and the dip was real but HALF the size of ours — so something else owned the other half. Ten days of digging with the centre managers found it, and it was self-inflicted: we'd 'optimised' the class timetable in April — my initiative, to cut instructor costs — and the cuts had landed disproportionately on the 6am and 7pm slots. The commuting members those slots served weren't switching classes; they were leaving. The saving on instructors was £9,000 a quarter; the attrition it drove was costing roughly £21,000 in lost membership revenue over the same period, and compounding.

What I told my director — in week two, not at quarter-end: the red was two-thirds mine, here's the mechanism, here's the reversal plan, and here's what the reversal costs. No seasonal cover story, though one was available and half-sold.

The fix: peak-shoulder slots restored at the eight centres with the worst attrition, instructor costs partly recovered through multi-site scheduling instead of slot cuts, and a rule I now run everywhere: any timetable change gets an attrition check at 60 days, owned by a named manager — because usage optimisations that ignore WHO uses the slot aren't optimisations, they're evictions.

The quarter after: attrition back inside two points of baseline, occupancy recovering, and — the number I actually watch now — 6am-slot retention above its pre-cut level.