Somebody is told to take 10% out of transport. So they renegotiate the rate, re-plan the routes, and lean on the carrier.
Meanwhile three numbers that nobody in that meeting can change have already decided most of the bill. They were set by the commercial team, usually years ago, usually in one afternoon, and nobody in the room that day was thinking about trucks.
They are the order cut-off, the lead time and the minimum order. Between them they decide how many stops you buy, how full each vehicle leaves, and whether the planner has anything to work with. This is how to test yours.
The three numbers, and their second meaning
Each one has an obvious commercial reading, and a second reading that nobody prices.
| Rule | What it means to sales | What it means to transport |
|---|---|---|
| Cut-off | How late a customer can still order for tomorrow | How much of tomorrow you can see when you commit the vehicles |
| Lead time | How fast we promise to deliver | How long an order may wait for company |
| Minimum order | The smallest order worth taking | How many stops you are buying |
The second column is where the money is, and it is invisible from the first. That is the whole problem: the people who set these numbers are optimising the left-hand column and cannot see the right-hand one.
First, the thing that makes all three work: a stop has its own price
Most people carry a mental model of freight cost with one term in it, and that term is distance.
Look at how a delivery is actually costed. A Thai construction material store, delivering through a third-party trucking company from a single depot, had its daily planning modelled by researchers at Khon Kaen University. The cost function they had to build has three separate terms:
- a fixed cost per vehicle per trip
- a variable cost per kilometre
- a delivery cost per delivery point
That third term is the one missing from most people's heads, and it is the one that makes a small order expensive. A stop costs the same whether you unload fifty kilogrammes or three tonnes off the back.
You can see why in the time. Unloading at a customer splits into a fixed part, roughly ten to twenty-five minutes depending on access and what equipment is there, plus a variable part that depends on how much is on the truck. Only the variable part knows how big the order is. The fixed part is the stop itself: finding the gate, the security desk, the paperwork, the person who signs.
So the cost of an order is a share of the trip, plus its kilometres, plus one stop. Hold onto the third term. Every one of the three rules below is really a rule about it.
Rule 1: the cut-off decides how much you can see
Planning a day's deliveries is a matching problem. You can only combine orders you can see. The cut-off is the moment you stop seeing more of them and have to commit.
Here is what a cut-off actually buys, from a contract haulier's depot: orders close at midday. The afternoon goes on load scheduling and preparing picking notes, with time deliberately left to slot in urgent orders for the next day. The following afternoon the goods are picked and the returning driver loads them. Delivery is the day after that. One midday deadline, and everything downstream of it has room to work.
Now move that deadline to 17:00 and watch what happens. The planner is building loads in the evening or the next morning, against a shorter runway, with the pick and the load compressed behind it. Nothing is impossible. It is just that every decision gets made with less information and less time, and the way that shows up is extra vehicles.
How much is the pool worth? There is a Thai measurement of exactly this. A logistics company running collection rounds for four vehicle assembly plants planned every day, with real data, using people who knew the job. But it sorted every pickup into one of four time windows and never let a route cross one, and it kept routes separate by destination plant. Its vehicle utilisation averaged 58.07% and sat below its own 80% target in every month of a full year.
The researchers changed nothing about the planning. Same demand, same fleet, same algorithm. They merged the four windows into two and let one route serve two nearby plants. Utilisation went to 92.60% and cost fell 58.52%.
That study is about time buckets rather than cut-offs, but the mechanism is identical and it is worth saying plainly: the wider the pool of orders allowed to travel together, the fuller the truck. A cut-off is a rule about pool width, drawn in time. Every hour you move it earlier puts more of tomorrow into the same pool.
Test yours in an afternoon. Take four weeks of orders and, for each one, record the clock time it arrived. Then plot the cumulative share of a delivery day's volume against the hour it was ordered. You will usually find something like 70% of it is in by early afternoon and the last few hours bring a thin tail. If your cut-off is 17:00 and the curve is flat after 15:00, moving the cut-off to 15:00 costs almost no service and buys two hours of planning. If the curve is still climbing steeply at 16:30, your customers genuinely order late and the cut-off is doing real work. Either way you now know, instead of guessing.
Rule 2: lead time is a consolidation window, and you may already own one you are not using
Lead time is not really a promise about speed. From the transport side it is permission: how long an order may sit and wait for company before it has to go.
With one day of lead time, an order can only travel with the orders that arrived alongside it. With three days, it can wait for the day a truck is already going that way.
There is a real cost on the other side and it should be stated honestly. Someone holds stock to cover a longer lead time, usually the customer, and nationally that is not a small pile to be moving cost into: NESDC puts Thailand's 2024 transportation cost at 1,200.6 billion baht, or 47.9% of the logistics bill, against inventory holding at 1,122.1 billion, or 44.7%. Two piles of almost the same size. Lengthening a promised lead time moves money from the first into the second, and needs the arithmetic doing.
But there are two different levers here, and only one of them costs anybody stock.
- The promised lead time is how far ahead the customer must order. Lengthening it makes the customer carry more.
- The dispatch window is how long you may hold an accepted order before it must leave. Widening it costs nobody anything, as long as the order still arrives when you promised.
Almost every operation quietly gives the second one away. If you promise three days and you ship everything on the next morning's truck, you are holding two days of consolidation permission and using none of it. The order is already accepted, the goods are already in your warehouse, the customer's expectation is already set at day three. Waiting one more day for a second order on the same lane costs you nothing and can halve a trip.
The rule that follows is short. Ship to the promise, not to the day the order arrived. Then a small order in a quiet direction has somewhere to sit.
Two guards on it, or it turns into a service problem:
- Publish the shipping days per area, not a general promise. "Orders for the north-east ship Tuesday and Friday" is a schedule a customer can plan around. "Three days, usually" is a promise you will break the day someone counts.
- Keep an urgent route open and priced. The haulier's depot above kept three order categories, forward, normal and urgent, and brought in outside vehicles when an urgent one would not fit. That is the honest way to hold a consolidation window: it exists for the normal orders, and the genuine emergency has a door that is not the schedule.
Rule 3: the minimum order is a drop size wearing a disguise
Minimum order size looks like a commercial rule about which customers are worth having. It is actually the setting that decides how many stops you buy and what vehicle they need. The divide-by-payload test that decides whether frequency costs you anything is the same test applied to one order instead of a week of them.
This is worth modelling rather than arguing about, and at least one large soft drinks manufacturer does exactly that: it runs different minimum order and drop sizes through a routing package specifically to see what each would do, because a lower minimum lifts revenue and lifts delivery cost with it. The two move together and only the arithmetic separates them.
Here is the arithmetic, on one round.
Take a six-wheeler on a 180 km round trip. Using the Thai study's own figures for that vehicle, 2,505.25 baht of standing cost per round and 6.45 baht per kilometre:
2,505.25 + (180 × 6.45) = 3,666.25 baht for the round, before anybody thinks about stops.
Say it carries 24 cubic metres across eight drops, and one of those drops is a small order of 0.4 cubic metres.
- Allocate by volume, which is what most costing systems do: 0.4 ÷ 24 = 1.67% of the round, or 61 baht. That order looks almost free.
- Allocate by the day it consumed. Eight drops at about half an hour each is four hours, and 180 km at 60 km/h is three more, so this round is essentially the vehicle's whole usable day. The stop plus the detour to reach it takes about half an hour, which is one-sixteenth of an eight-hour day: 3,666.25 ÷ 16 = 229 baht.
Same order, same truck, same day. Costed by volume it is 61 baht. Costed by what it actually took, 229 baht, nearly four times as much.
Now turn that into the rule. If a stop is worth about 229 baht of vehicle time on your lanes, and your delivery margin is 12%, then an order has to be worth 229 ÷ 0.12 = about 1,900 baht just to pay for its own stop, before it contributes a single baht towards the trip itself. Run that on your own numbers and you have a defensible minimum order, arrived at rather than inherited.
And set it in the right unit. Most Thai minimum orders and free-delivery thresholds are written in baht, and baht does not fill a truck. Twenty thousand baht of light packaging and twenty thousand baht of floor tiles are completely different transport jobs, and one of them will cube out while the other weighs out. Set the minimum in whichever unit runs out first on your goods: pallets, cubic metres or kilogrammes. Keep a baht figure alongside it for the sales conversation if you must, but let the physical unit be the one that governs.
The Thai clock all three run into
The driving day. Under Section 103 bis of the Land Transport Act B.E. 2522, subject to the labour protection law, a licensed driver may not drive continuously for more than four hours from starting, and after a continuous break of at least half an hour may drive up to four more consecutive hours in the same twenty-four.
Worth being precise about what that section does and does not say, because it gets quoted loosely. The Act limits hours at the wheel, not distance. The Khon Kaen study capped each vehicle at 480 km per trip and described that as set by Thai transport law. There is no 480 km rule in the Act. What there is, is eight hours of driving, and 480 km is what eight hours looks like at 60 km/h. That makes 480 a perfectly sensible planning number for a Thai highway lane and a poor thing to call a legal limit, because the moment your average speed is 45 it is wrong by a third.
The Bangkok expressway windows, which are what actually punish a late cut-off. Six-wheelers are barred from the listed expressways between 06:00 and 09:00 and again between 16:00 and 20:00. Trucks of ten wheels and over, and trailers, are barred between 06:00 and 09:00 and between 15:00 and 21:00.
Read those next to your cut-off time. A cut-off that leaves picking finishing in the late afternoon has just posted your vehicle into a closed window. It will either sit until 20:00 or 21:00, or go the long way round, and neither of those appears on the invoice as "our cut-off is too late". In and around Bangkok the cut-off has to be early enough that the vehicle either clears before 06:00 or gets away in the open middle of the day. That single constraint is often the strongest argument available for moving a cut-off, and it is checkable rather than a matter of opinion.
What to change on Monday
- Write down your three numbers. If nobody can say what the cut-off, the promised lead time and the minimum order are without looking them up, that is the finding
- Plot four weeks of orders by the clock time they arrived, cumulative, against delivery-day volume. The shape of the tail after your current cut-off tells you what moving it would actually cost in service
- Compare your promised lead time with the day you actually ship. If you promise three days and ship next morning, you own two days of free consolidation and are giving them away
- Turn that window into published shipping days by area rather than a general promise, and keep a priced urgent route so the genuine emergency does not eat the schedule
- Work out what a stop is worth on your own lanes: standing cost plus kilometres for a typical round, divided by how many half-hours of the vehicle's usable day it holds. That number is the floor a minimum order has to clear
- Check what unit your minimum order and your free-delivery threshold are written in. If it is baht, convert them to pallets, cubic metres or kilogrammes and see which of your customers move sides
- Lay your cut-off against the expressway ban windows. If loading finishes after 15:00 or 16:00, the cut-off is buying your carrier a closed road, and that is a cost you are paying without seeing it
- Take the result to whoever owns the commercial rules, with the second column of that first table filled in. These are not transport decisions and transport cannot change them alone
None of this is a routing problem, which is why re-planning routes keeps returning less than expected. The planner is playing whatever hand the order rules deal, and how wide you let the pool of shareable jobs get decides the game before any algorithm runs. Cut-off, lead time and minimum order are the three dials that set the pool. They sit outside the transport department, and that is exactly why nobody has looked at them.