Two orders, one production run, two customers about ninety minutes apart. Somebody in the room says put them both on one truck. Somebody else says we will miss the second window. Nobody has any numbers, so the loudest person wins, and the same argument happens again next month with a different pair of orders.
It is one of the few freight questions with a real answer. Not a rule of thumb, an answer, arrived at in about ten minutes with a map and a calculator. This is how to get it.
The saving has a shape, and the shape is a triangle
Start with the geometry, because it is where most people's intuition fails.
Your site is O. The two drops are A and B. Call the distance from you to A a, from you to B b, and between A and B c.
Send two vehicles and each does an out-and-back, so the fleet covers 2a + 2b. Send one vehicle through both drops and it covers a + b + c. Subtract, and the distance you save by linking the two drops is:
a + b β c
That is the whole of it. Everything else in routing software is this sum, run across every possible pair, sorted by which pair saves the most.
Put numbers on it. Say A is 60 km out, B is 75 km out, and the two drops are 25 km apart.
| Distance covered | |
|---|---|
| Two vehicles, one drop each | 2(60) + 2(75) = 270 km |
| One vehicle, both drops | 60 + 75 + 25 = 160 km |
| Saved | 110 km |
Now move B to the other side of you. Same 60 and 75, but the drops are now 130 km apart.
| Distance covered | |
|---|---|
| Two vehicles, one drop each | 270 km |
| One vehicle, both drops | 60 + 75 + 130 = 265 km |
| Saved | 5 km |
Two jobs that look identical on a spreadsheet, and one of them saves nothing at all. The eyeball version of the test:
Combining pays when the two drops are closer to each other than they are to you. Once the leg between the drops is longer than the legs out to them, the triangle has flattened, and you are not building a route. You are building a detour.
One practical note on measuring. Use road distance, not the straight line, because a river or a one-way industrial estate gate can double a leg that looks short. If you are sketching quickly, the usual planning shortcut is to measure the straight line and multiply by about 1.2 to approximate the road.
The kilometres are the small part of the prize
Here is where the naive version of this calculation goes wrong. It treats the saving as fuel. Two Thai studies applied exactly the arithmetic above and came back with very different shapes of answer, and the difference between them is the useful part.
A community enterprise in Kaset Wisai district, Roi Et, delivering jasmine rice to 43 buyers across eight districts. It was running eight separate routes covering 675 km, burning 44.82 litres and costing 954 baht per delivery round. Routed properly under distance, cost and load weight constraints, the same 43 buyers were served by three routes covering 413 km, on 27.53 litres and 586 baht.
Distance fell 38.8%. Cost fell 38.6%. The two move together almost exactly, and there is a reason: divide either cost by either fuel figure and you get 21.29 baht a litre both times. That study costed fuel and nothing else. Of course the saving tracked the distance.
A transport operator in Chonburi, moving parts, documents and people out of client factories, three months of its own job records: 76 jobs a month for four customers across six vehicles, every job run as its own direct trip. Baseline, 12,351.75 km and 125,998.00 baht a month. After the same savings arithmetic was applied, with the weight of each load also used to pick the vehicle type: 10,671.40 km and 82,334.36 baht a month.
Distance fell 13.60%. Cost fell 34.65%.
Distance came down a little. Money came down by more than a third. The route tables show why: the exercise did not only decide which jobs travelled together, it changed which vehicle went out on a lot of them, moving small jobs off a pickup onto a van. What actually paid was not the kilometres. It was the number of vehicle-days spent and the size of the vehicle spending them.
That is the thing to carry into the decision. Freight cost is mostly a daily cost, not a per-kilometre cost, so the prize for combining two drops is roughly one whole vehicle-day, and the diesel is the rounding error. It is also, at national scale, what NESDC saw happening in 2024: Thai road tonnage rose 11.4% while the cost of moving it fell, which it read as carriers consolidating loads from several customers along similar routes to fill trucks. The same fixed costs, spread over more freight.
Which flips the question. Given that one truck is almost always cheaper on paper, the decision is not really about cost. It is about fit.
Four things the load has to fit inside
Combining is cheaper whenever it works, so the whole exercise is checking whether it works. Four ceilings, in the order they usually bite.
1. The day
This is the one that kills most multi-drop plans, and it is not distance that kills them.
A land transport employee's working day in Thailand is eight hours, plus up to two hours of overtime with written consent, and at least ten hours must pass before the next day can start. Ten hours of working time is the ceiling you are packing into, and the driver's day, not the truck, is what runs out.
The trap is that each drop costs a fixed block of that day regardless of how small the drop is. Arrival, gate, paperwork, dock wait, unload, signature, release. Delivering four pallets instead of twenty saves you almost none of it.
So the arithmetic that matters is not kilometres, it is drops. Take a lane with four hours of driving in total and drop handling that averages forty-five minutes:
| Drops | Driving | At the drops | Loading, before you start | Total |
|---|---|---|---|---|
| 2 | 4h 00 | 1h 30 | 1h 30 | 7h 00 |
| 3 | 4h 15 | 2h 15 | 1h 30 | 8h 00 |
| 4 | 4h 30 | 3h 00 | 1h 30 | 9h 00 |
| 5 | 4h 45 | 3h 45 | 1h 30 | 10h 00 |
| 6 | 5h 00 | 4h 30 | 1h 30 | 11h 00 |
Read the two middle columns against each other. Adding drops moves driving time by fifteen minutes a time, because by definition you only combine drops that are near one another. It moves drop time by forty-five. The day is consumed by the number of stops, not by the map. That is why the honest answer to "can we add one more" changes so suddenly: nothing at all goes wrong until the day runs out, and then everything does at once.
Measure your own drop time before you use anyone's. Take gate-in to gate-out from your last twenty deliveries at each receiving site, and use the slow ones, not the average. A single site that habitually takes two hours is not a drop, it is half a day, and it should be on its own truck.
2. The vehicle
Two ceilings here, and they are not the same one.
Volume usually binds first for anything manufactured or packaged. If your goods fill the deck before they get anywhere near the weight limit, the weight rules below are not what is deciding your answer, and the size of the vehicle is the live question instead.
Weight binds for anything dense, and here it is not a commercial choice. Under Section 61 of the Highway Act B.E. 2535, highway directors may publish in the Royal Gazette a prohibition on vehicles over a set weight, and the table currently in force comes from an announcement published on 28 December 2005 and amended several times since. Two figures worth knowing:
| Vehicle | Maximum gross weight |
|---|---|
| Three-axle ten-wheel rigid | 25 tonnes |
| Six-axle twenty-two-wheel semi-trailer combination | 50.5 tonnes |
Three things people get wrong about those numbers. They are gross, so the vehicle's own weight is inside them and your payload is what is left. They apply to motorways, national highways and concession highways, because that is what the announcement covers, and rural and local roads are announced separately by their own authorities, so the ceiling can change under the same truck partway through a trip. And breaching one is not a commercial matter: Section 73/2 makes it imprisonment of up to six months, a fine of up to ten thousand baht, or both.
If the combined load is over the limit, the debate was never real. You were always sending two vehicles.
3. The windows
Two customers who both insist on 09:00 to 11:00 are two trucks, whatever the triangle says. There is no arithmetic that fits a single vehicle into two places at once.
This is worth attacking rather than accepting, because delivery windows are frequently inherited rather than needed. Before you pay for a second vehicle-day, find out which of the two windows is a real constraint at the receiving end and which one is just what somebody wrote on a form in 2019. Widening one window by two hours is often the cheapest freight saving available to you, and it costs nothing but a phone call.
4. The order
The last one to be delivered has to be the first one loaded. That sounds obvious and it is routinely forgotten until the truck is at the first gate.
Two ways it goes wrong. The delivery order is fixed by the windows, and the loading order is fixed by what came off the line first, and the two do not match, so somebody restacks the trailer at 05:00. Or nobody notices, and the driver has to unload half of drop two on the kerb to reach drop one, in the rain, on a shoulder, with the clock running. That is drop time paid twice and damage risk taken for free.
Check it before the load is built, not after. If the order that the windows demand is not an order you can load in, that is a genuine reason to split, and it is the one people discover latest.
What one truck concentrates, two trucks spread
A last consideration that never appears in the cost comparison, and should.
One truck chains the risk together. A two-hour wait at drop one is not a two-hour problem, it is a missed window at drop four and a driver who runs out of legal day on the way home. Two trucks contain the same delay to one customer. So when you are deciding which jobs to combine, look at which receiving sites actually release a vehicle on time and put the unreliable one last in the sequence, or on its own.
One truck also blurs a damage claim. The load is disturbed at every stop, and the acceptance of the goods happens separately with each consignee, so damage found at the third drop may well have been done while digging out the first. That does not change what the carrier is liable for, but it changes how easily you can prove where it happened. The fix is cheap: a condition note signed at each drop, not just at the last one.
The test, in five lines
Run these in order and stop at the first failure.
- Measure the triangle. Is the distance between the drops less than the distance out to each of them? If not, combining saves nothing. Send two.
- Add the drops to the day. Driving time, plus your own measured gate-in to gate-out at every stop, plus loading. Against ten hours of working time. If it does not fit, no rate negotiation will make it fit.
- Check both ceilings on the vehicle. Volume first, then gross weight against the legal limit for the class of road you are on.
- Check the windows are compatible in the sequence the route requires, and check whether the tighter one is real.
- Check the delivery order can also be the loading order.
Pass all five and one truck is cheaper by about a vehicle-day, which is a much bigger number than the fuel difference you were arguing about. Fail any one, and forcing it costs more than the second truck: overtime, a missed window, a redelivery, or a damaged pallet nobody can assign.
What to change on Monday
- Measure the gate-in to gate-out time at each of your regular receiving sites and write the slow number, not the average, next to its name. That single list decides more of these arguments than the map does
- Sketch the triangle before proposing a combined run. If the drops are further from each other than from you, do not propose it
- Count drops rather than kilometres when you are asking whether a day is full. Distance grows slowly as you add stops, drop time does not
- Find out the gross weight of your own densest full load and check it against 25 tonnes and against the class of road it travels on. Gross, not payload
- Ask which of your delivery windows are actually required at the receiving end. The ones that are not are free capacity
- Confirm that the delivery order the windows demand is an order the load can be built in, before the load is built
- Put your least reliable receiving site last in any combined sequence, or leave it on its own vehicle. It is the one that turns one late delivery into four
- Take a signed condition note at every drop on a multi-drop run, not only the final one
None of this is a preference for one truck or two. It is a way to stop paying for the wrong one out of habit, in both directions: the shipper who splits everything is buying vehicle-days it does not need, and the shipper who combines everything is paying for it in overtime, failed windows and claims nobody can settle.