Plant, cold-chain, 3PL, procurement, engineering — five UK roles, five questions of the same autonomous fleet this week. One capex sheet survives Friday.

It is Thursday morning in a Midlands distribution centre and the capex pack is due Friday. The Finance Director wants one sheet, three numbers, and a seven-year line through them. The Plant Director walked the floor at 06:00 and lost fourteen minutes to a missed shift; the Cold-Chain Warehouse Manager is on the radio because the 14:20 outbound trailer window is shrinking; the 3PL Operations Director has an HSE inspector booked for next Tuesday at the Coventry site and a second one the week after at Crick. The Procurement Lead is staring at a lease quote that runs to thirty-two pages. Five roles, one autonomous fleet on the floor, one capex sheet on Friday.

Thesis. When five operational roles read "good" through five different lenses, only a fleet that answers each lens on its own terms — without forcing a re-tender when the lens changes — survives the capex pack.

1. The Plant Director — fifteen minutes of starved line is a quarterly miss

Take a UK automotive plant running 24/7. The engineering memo from the line manager says a single missed forklift driver shift starves the line for fifteen to thirty minutes before the buffer absorbs it. On a site running at a £4.2m monthly margin, that is not a rounding error; it is a quarterly miss explained at the next operations review.

The Plant Director's reading of "good" is cadence protection. Any fleet that promises three trucks but only fields two for a Tuesday late shift has failed before the shift engineer has finished their tea. Mixed-traffic PUWER is the second axis — twenty operators, four pedestrian crossings, two engineered guarding zones — and the safety case has to be designed in, not retrofitted. The third axis is fault recovery: a stopped autonomous truck that takes twelve minutes to be diagnosed and four minutes to rejoin the cadence has cost the plant one missed shift before lunchtime.

The cadence arithmetic is brutal in its simplicity. Twenty-two missed-shift-equivalents across a quarter, at an average eighteen minutes of starved line each, is six and a half hours of unbuilt vehicles. The Plant Director carries that number to the operations review whether or not the truck supplier is in the room.

The question to ask in the next vendor meeting: what is your mean time to first remote diagnostic response from a UK-resident engineer on a Tuesday at 02:00, evidenced from your last ninety days of UK incident logs — not from the master service agreement?

2. The Cold-Chain Warehouse Manager — the twenty-minute trailer window does not flex

The summer peak in a UK cold store does not negotiate. The outbound trailer window is twenty minutes from the moment the doors part to the moment the seal goes on, and the freezer-driver churn rate that Logistics UK's last cold-chain survey put north of twenty per cent year-on-year is now a structural constraint, not a seasonal one. Glove changes, mandated rest breaks, seasonal re-training cycles — these are real, regulated, and they compress the throughput envelope precisely when it needs to expand.

The Warehouse Manager's reading of "good" is window protection at minus twenty-five. An autonomous lift truck that sustains pallet throughput at sub-zero without rest breaks, glove changes or seasonal re-training does not care that the previous freezer driver gave notice last Friday. It does care about battery thermal management, mast-seal integrity, and lidar performance through the mist halo at the strip curtain. Three engineering disciplines, one operational question: does the truck hold the twenty-minute window, or doesn't it?

The arithmetic in cold-chain is window-led. Miss the window once across the day and the outbound trailer slips into the next slot; miss it twice and the planner is moving the next morning's inbounds to make room. The Warehouse Manager is not optimising cost per pallet; they are protecting the slot.

The question to ask in the next vendor meeting: at what ambient minimum is your published throughput SLA still binding, and where in your UK reference base is it being run today between -22°C and -28°C with trailer-turn evidence to match?

3. The 3PL Operations Director — LOLER, PUWER, UKCA at inspector timescales

The HSE inspector arrives on Tuesday. The HSE inspector arrives at the second site the Tuesday after. Forklift trucks account for around thirty per cent of all reportable workplace-transport injuries in the UK, and that figure has not improved since autonomous fleets began entering UK 3PL contracts in 2024-25. An Operations Director running seven sites has to evidence LOLER thorough examinations, PUWER inspections and UKCA conformity for every truck on every floor, in the timescales the inspector will accept — which means at the truck side, in a binder, today.

The Ops Director's reading of "good" is portable evidence. A compliance pack that only lives in a vendor portal and only renders on the vendor's UK engineer's laptop is not a compliance pack — it is a leaflet. The pack has to leave with the inspector or be defensible without the vendor in the room. That is the line that the closed bundled stacks sit just below, and it is the line that an open-fleet design has to sit above on day one of go-live.

The multi-site dimension makes this worse, not easier. Seven sites means seven inspector calendars, seven LOLER thorough-examination dates, seven PUWER schedules, and one Operations Director who cannot be physically at all of them. The compliance pack has to scale to the site manager who has to face the inspector at 09:00 on the day the Ops Director is somewhere else.

The question to ask in the next vendor meeting: show me the PUWER inspection sheet, the LOLER thorough-examination certificate, and the UKCA conformity declaration for one named truck on one named site, dated within the last twelve weeks — at the truck side, not in the deck.

4. The Engineering Warehouse Manager — aftercare is where capex becomes stranded

In a UK engineering plant, an autonomous forklift that waits eleven days for a mast-cylinder seal from an overseas warehouse is not a capex asset; it is stranded production risk. PUWER 1998 puts the duty to keep the truck in efficient working order on the operator, not the vendor — and the Warehouse Manager is the operator's named representative on the shop floor.

The Warehouse Manager's reading of "good" is aftercare with teeth. Engineer-attendance SLA in hours, not best-effort. Remote diagnostics that resolve the easier sixty per cent of faults before an engineer is dispatched. A spares depth that is held in the UK, not on a manifest. The arithmetic is direct: at a UK engineering plant turning over £180k a shift, one stranded truck across three shifts is a five-figure margin call before the parts have arrived from anywhere.

The hidden cost is in the SLA breach mechanics, not the SLA promise. A four-hour engineer-attendance target with no contractual penalty is a marketing line. A four-hour target with a per-hour service credit and a thirty-day escalation to a regional engineering manager is a commitment. The Warehouse Manager reads the small print on the second page of the schedule and quietly knows which of the two is on the table.

The question to ask in the next vendor meeting: what is your UK spares-in-country list by part number for the FW1400Pro-class trucks, what is your contractual first-time-fix rate on those parts, and what is the published penalty if you breach engineer-attendance on a Tuesday at 02:00?

5. The E-commerce Procurement Lead — lease against a five-year throughput plan

The procurement pack at a UK e-commerce fulfilment centre already assumes an eighteen-per-cent peak headcount risk on the 2026-27 horizon. Against that, the lease-versus-purchase decision on an autonomous lift truck fleet has to clear three hurdles: five-year throughput sensitivity, year-seven residual at exit, and the lift-out cost if the 3PL contract is re-tendered at year three.

The Procurement Lead's reading of "good" is the arithmetic on one page. Automated lift truck AGVs cut UK e-commerce fulfilment pallet-move costs by thirty-eight to fifty-five per cent across two shifts compared with operator-driven counterbalance trucks. Break-even on a five-year operating lease falls inside eighteen months at over sixty-five per cent night-shift utilisation. But the page has a fourth column the closed-stack vendors prefer the reader does not see: lift-out cost. Re-tendering at year three with a closed fleet means buying the integration twice; with an open fleet running VDA 5050, the integration is portable.

The peak-headcount assumption is what makes the page move. If the eighteen-per-cent risk lands as twenty-two, the throughput sensitivity collapses any fleet whose night-shift utilisation cannot scale linearly. The open-fleet design holds the line because the next truck on order is the same VDA 5050 endpoint as the last one; the closed-fleet design holds the line only if the original vendor agrees the new pricing.

The question to ask in the next vendor meeting: model the year-seven residual against a year-three re-tender and show me the lift-out cost both ways — one number per truck, on one page.

The arithmetic

  • Forklift trucks account for ~30% of UK reportable workplace-transport injuries (HSE), and the figure has not improved since autonomous fleets began entering UK 3PL contracts in 2024-25.
  • Automated lift truck fleets cut UK e-commerce pallet-move costs by 38-55% across two shifts versus operator-driven counterbalance trucks.
  • Break-even on a five-year operating lease falls inside 18 months at over 65% night-shift utilisation.
  • One missed forklift driver shift on a UK automotive 24/7 line = 15-30 minutes of starved line before the buffer absorbs it.
  • A 20-minute outbound trailer window at -25°C, summer peak, does not flex around freezer-driver churn now running >20% year-on-year (Logistics UK).
  • Year-7 residual on a UK 1.4-tonne autonomous lift truck on an open fleet typically clears £15-25k per truck above a closed-stack equivalent, because the integration is portable to the next operator.

What to do on Monday morning

  • Walk the floor with a printed PUWER schedule and a printed LOLER thorough-examination calendar. Mark in red any truck an HSE inspector could not clear today. Hand the list to the 09:00 Operations meeting, not to month-end.
  • Pull the last ninety days of shift-attendance data. Cost out one row per cell — per missed driver shift, the cadence loss in minutes and the margin loss in pounds. Take the average to the capex pack on Friday and put it next to the lease line.
  • Add one column to the year-7 residual model headed "lift-out cost at year-3 re-tender". Price the open-fleet case and the closed-stack case against it. If the second number is more than twice the first, the capex sheet has a fourth line on it that nobody wrote down yet.

Five roles, five readings of "good", one capex sheet on Friday. If you would like a quiet read of an open-fleet design that answers all five on the same page — without the deck — reply to this edition or comment below. We will send a single-page model written for your operation, no slot required.