How do you plan delivery routes across multiple sites?
By Muhamed Selmani, Founder
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.
A vehicle-day costs what it costs
Whether a round makes 4 drops or 10, the vehicle, the driver, the hours and most of the fuel come within a few percent of each other. What changes is how many drops share the bill.
That is route density: drops per vehicle-day, and drops per mile. Nearly everything else in route planning moves one of those two numbers. It is also why a cost per drop quoted on its own tells you almost nothing — the denominator is doing most of the work.
For a restaurant group, or any multi-site food operation, it is the difference between a delivery operation that gets cheaper as you open sites and one that gets dearer. Sites clustered along a corridor add drops to rounds that already run; sites scattered across London add rounds.
The three numbers that describe a round
Before changing anything, measure the round you have. Three numbers describe it, and most operations hold the first two and never record the third.
- Drops per vehicle-day, counted daily rather than weekly. Look at the distribution rather than the average: one thin day a week is where the money leaks.
- Miles per drop: round mileage divided by drops, run out and empty return included.
- Minutes at the door: arrival to departure, including the walk, the wait at goods-in, the signature and the trays back.
Door time is the number nobody has and the one that binds
Two rounds with the same mileage and drop count can cost differently, and the difference is nearly always at the door. A site with a bay and a trolley is not the same drop as one with a service lift shared with a hotel kitchen.
Door time is a priced line rather than an opinion. On Fleetable's rate card, loading and waiting are charged at £0.45 a minute ex VAT, with 15 minutes included on loading and 15 on waiting before anything is billed. Whoever you buy from, find the equivalent: a rate that names no door-time allowance has one anyway, and it is inside the price.
If your delivery notes carry no arrival and departure times, stand at the doors for a week with a stopwatch. It changes more rounds than a software licence.
What density actually does to cost, and where it stops
Density is worth chasing, but not without limit. The return on each extra drop falls away quickly, and then the round hits a wall unrelated to cost.
Worked example
The density curve, and the window that ends it
Assumptions
- One vehicle and one delivery window of 4 hours, which is 240 minutes. An assumed window, not a standard.
- 30 minutes to run out to the first drop. Assumed, and the empty return leg is left out of the timing entirely.
- 15 minutes at the door per drop, the loading allowance published on Fleetable's rate card, plus an assumed 12 minutes of travel between drops.
- The vehicle-day costs the same whatever the round carries. Call it C: a figure here would be an invented benchmark, and the shape is the point.
Working
- Time consumed per drop: 15 at the door + 12 travelling = 27 minutes.
- Time available for drops: 240 - 30 = 210 minutes.
- Drops that fit: 210 / 27 = 7. Drop 8 would need 216 minutes against 210 available.
- Cost per drop: 4 drops = 0.250C. 5 drops = 0.200C. 6 drops = 0.167C. 7 drops = 0.143C.
- What each extra drop takes off the cost per drop: drop 5, 0.050C · drop 6, 0.033C · drop 7, 0.024C.
So: The curve is steep at the start, flat at the end, and then it stops rather than flattening: the binding limit is the window, not the vehicle. Adding drops is worth a great deal on a thin round, little on a full one, and less than nothing once the last site falls outside its window. Change any assumption above and the numbers move; the shape does not.
Where density comes from
Four things produce it, and only one of them is a routing decision.
- Geography, measured in driving time rather than borough names. Two sites in one postcode district can be further apart at 07:30 than two in different ones. Cluster by the streets a vehicle actually passes.
- The window. It is set by the receiving site far more often than by the carrier: yard hours, a shared service lift, the kitchen's prep timetable, a landlord's condition. A single immovable window can shape an entire round.
- Frequency: how many times a week each site is served, which is a commercial decision rather than a logistics one.
- Drop size, set by what the receiving site can hold. Fridge space at the far end is the real constraint on most food rounds.
Frequency is the lever nobody pulls
Most rounds are shaped by a decision nobody remembers making: 6 deliveries a week because the last supplier came 6 times, or 5 because the site list was different then.
Halving the frequency at a site roughly halves the vehicle-days it consumes and doubles its drop size. The constraint is almost never logistics — it is shelf life, storage and the prep timetable at the receiving end. So the question to put to each site manager is narrow and answerable: what would have to be true here to take 3 deliveries a week instead of 5? Sometimes the answer is a shelf.
Sometimes it genuinely cannot change: short-dated product is bought the day it is used, and forcing frequency down on those lines buys waste instead of saving mileage. This is a lever, not a doctrine.
How to run a route review
A route review is a fortnight of work rather than a software project, and it is worth doing yearly and after any site opening.
The method
- Pull 12 weeks of delivery data: date, site, arrival and departure time, order size, vehicle class, miles.
- Plot drops per vehicle-day across those weeks, and look at the thin days rather than the mean.
- List the hard windows and, beside each, who set it and why. Some are planning conditions, some are habits, and both look identical on a spreadsheet.
- Mark the sites that break the sequence: the drop a round detours for, and the one served last because it is easiest to be late for.
- Measure door time, and where it is not recorded, go and record it. Then model 2 or 3 shapes: tighter geography, lower frequency where sites can take it, a different vehicle class on the peak day.
- Test one route in parallel for a fortnight before changing anything. A plan that has never met a Monday morning is a hypothesis.
Four patterns worth looking for
Nothing is claimed here about what a review usually finds — that would be a statistic, and there is not one. Four things are worth looking for.
- A round shaped by history rather than geography: built for a different site list, never rebuilt.
- A site whose window costs more than its order is worth. Price it first: the conversation changes once the extra vehicle-hours are a number.
- A vehicle class set on the average day, so the peak day needs a second vehicle it would not need one class up.
- A return leg nobody prices. An out-and-back round to one site earns nothing on half its miles, and is the strongest candidate for merging into another.
What route optimisation software can and cannot do
Sequencing software solves the order of the drops you give it, under the constraints you give it. It cannot change frequency, negotiate a window or persuade a site to take a larger delivery, and those are the three levers that move density. All three are commercial conversations rather than algorithms.
It earns its keep on a drop list that changes daily. On a stable multi-site round of about 12 drops, a fixed sequence the driver knows is usually quicker than one recomputed overnight, because door time falls when the driver knows the door, the buzzer and where the trolley lives.
That is the honest limit of delivery route optimisation for a restaurant group. Software will tidy the sequence you already have. The three levers that move the cost — how often each restaurant is served, what window it will accept, and how much it can hold in one delivery — are settled in conversations with your own site managers, and they are worth having before anybody prices a licence.
One pricing note for anyone modelling a change: a multi-drop round is a different product from a single dedicated run. On Fleetable's rate card multi-drop carries a 10% uplift on the mileage rate, and a route invoices at a floor of £95 ex VAT whatever it carries. Adding drops to a round that already runs is still the cheapest way to move more food.
Related
Continue reading
FAQ
Questions, answered.
How many drops should one route carry?
Is a fixed sequence better than re-planning the round each day?
What data do I need for a route review?
Is it cheaper to deliver less often?
Start