Plain explanations of how food logistics in London actually works — the cold chain, route economics, SLAs, compliance and switching provider — written for ops and supply-chain leaders whether or not they ever work with Fleetable. Every figure here is sourced or marked as an assumption; nothing is invented.
A cold chain is unbroken temperature control from production to service, judged on the temperature of the food rather than the air around it. In England, assimilated Regulation (EC) No 852/2004 requires that it is not interrupted, and the Food Safety and Hygiene (England) Regulations 2013 make it an offence to keep food that supports pathogen growth above 8°C. A break is a judgement on time and temperature, made by the food business receiving the load.
Food waste from failed deliveries is five lines, not one: the stock written off at landed cost, the recovery run or emergency purchase, the sales lost while the shelf was empty, the hours your team spent absorbing it, and the customer outcome. The delivery charge is the smallest of them. No published percentage describes your operation, so the number worth having is your own, measured over a quarter from a rejection log.
Own the vans when the delivery work is dense, stable and needs your own people on it. Outsource when volume swings, the site list moves, or the van would stand still. The number that decides it is utilisation — how much of a van's paid week is spent moving your goods. Restaurants and multi-site groups should cost a van fully loaded, divide by drops, then price the same round both ways before anyone argues about it.
Plan routes by density rather than by area. A vehicle-day costs much the same whether it makes 4 drops or 10, so the number of drops sharing that day sets the cost per drop. Group multi-site delivery rounds by the streets a vehicle actually passes, sequence back from the tightest window, and set frequency by what the receiving site can hold. Density comes from geography, window and frequency — software sequences a round, it cannot invent density.
How much a dedicated courier costs is whatever the miles, the vehicle class, the door time, the return leg and the route type add up to. There is no market average worth quoting, and a cost-per-drop figure means nothing without the drop density behind it. Build the benchmark from your own invoices: split fixed from variable, count drops and miles, then compare like rounds. Report cost per drop next to drops per vehicle-day, or it is not evidence.
OTIF is one score: a drop counts only if it arrived inside its window and complete. Scored separately, a provider can miss both tests and still report well. A delivery SLA should fix the window, what counts as a miss, who decides, what evidence is captured per drop, how a shortfall is remedied and how a dispute is settled. Anything looser is a description of good intentions.
Delivery windows for restaurants and commercial kitchens sit before prep and well clear of service: early enough that a shortfall can still be fixed, late enough that someone is there to check it. What narrows it is almost always the receiving site — its planning consent, its yard or service lift, its staffing. Set the window from the kitchen's timetable outwards, then build the round to fit it.
Vet on evidence rather than on adjectives. Establish the legal floor first: a carrier is a food business under Regulation (EC) No 178/2002, works to procedures based on HACCP principles, and in England must not keep food that supports pathogen growth above 8°C. Then ask for what is voluntary and evidenced: certification such as BRCGS Storage and Distribution, temperature records from a live route, the insurance basis, and who responds when the load is on a subcontractor's vehicle.
A logistics company that owns no vehicles buys the capacity. It wins the contract, then sources it from subcontracted operators, direct or through a courier exchange. How that work is bought decides reliability, and how a driver shortage lands: a round bought on the morning competes for whoever is free that morning; a contracted route was allocated before that competition started. A set driver, a named backup and one accountable contract turn a network into a route.
Plan a peak from your own delivery history rather than from a sales forecast. Count drops by weekday and by site for the last 2 years, find the week the volume actually changes shape, and contract that capacity before the market prices it. Christmas and the summer event run are the two peaks a London food business cannot flex through. Capacity held for a window whether or not you call it is priced as its own line.
Two things change the emissions of a delivery round: the vehicle, and the number of vehicle-miles spent per drop. The vehicle is a booking question — ask what class and what emissions standard is specified for your route, and whether an electric refrigerated van can be booked for it in London. The miles are a routing question, and route density decides them. Any figure you are handed should arrive with its method attached, or it is not a measurement.
Change provider one route at a time, in parallel. Read your exit clause first: notice, minimum terms, what leaving costs. Then run one representative route with the incoming provider alongside your current one, on your own traffic, for long enough to include a bad week. Measure on-time-in-full, door time and exceptions against definitions both sides were given in the same words. Move the rest only once that route holds.