A warehouse lease gets signed for reasons that fit on one page. Rent per square metre. Incentives. How the site looked on the day somebody visited. Distance to a port, usually measured in kilometres on a map.
Then the building runs for ten years, and every one of those years a fleet of trucks drives a route that the lease decided. Nobody signs that part. It arrives monthly, as freight invoices, and by then the decision is a building.
The useful thing to know before you sign is that the number on the lease is the smallest of the numbers you are choosing.
A location fixes four costs, and rent is the least of them
Four costs move when a site moves. Inbound transport, outbound transport, the building itself, and the money tied up in the stock inside it.
Nationally, these are measured. Thailand's logistics costs in 2024 came to 2,509.4 billion baht, or 13.5% of GDP, down from 14.2% in 2023. Within that:
| Component | 2024, billion baht | Share of the logistics bill |
|---|---|---|
| Transportation | 1,200.6 | 47.9% |
| Inventory holding | 1,122.1 | 44.7% |
| of which inventory carrying | 913.7 | 36.4% |
| of which warehousing | 208.4 | 8.3% |
| Administration | 186.7 | 7.4% |
Warehousing is the smallest real line on the page. The stock inside the building costs the country roughly four and a half times what the buildings do, and moving goods costs nearly six times.
That is the national picture rather than your company's, and your own ratios will differ. But the shape holds for most businesses, and the shape is what matters here: a site chosen on rent per square metre is a decision optimised on the smallest of the four numbers it controls.
There is a second reason to care about that middle line. The cost of holding stock is mostly the cost of the money in it, and money was not cheap in the year those figures cover: the minimum lending rate averaged 7.1% in 2024, up from 6.8% in 2023. Anything a location decision does to your stock levels gets multiplied by that.
The two transport legs move in opposite directions
Inbound and outbound are different products, and a location change does different things to each.
Inbound, sometimes called primary transport, is bulk. Full vehicles, from a port, a factory or a supplier, arriving on a small number of runs.
Outbound is delivery. Part loads, many drops, small vehicles more often than large ones.
Move the warehouse towards your customers and outbound distance falls while inbound distance rises. Move it towards the port and the opposite happens. The obvious conclusion is that there is a point in between where the two balance out, and the obvious conclusion is roughly right but for the wrong reason, because the two legs do not respond equally.
The reason is worth holding on to. Delivery distance splits into two parts:
- Stem distance, the run out to the delivery area and back
- Drop distance, the distance travelled once you are inside the delivery area, going from customer to customer
Only the stem changes when you move the building. The drop distance is fixed by where your customers are relative to each other, and no lease will change it.
So the first sanity check on any location proposal is to ask what share of your delivery kilometres are stem and what share are drop. A business delivering into a dense industrial cluster is mostly drop, and moving 40 kilometres closer will save less than the map suggests. A business delivering thinly across several provinces is mostly stem, and the same 40 kilometres is worth real money.
The rule that resolves the tie: sit near the leg you cannot consolidate
When inbound and outbound are close to balanced on distance, the tiebreaker is not distance at all. It is which leg can be filled.
A full vehicle is the cheapest freight there is per tonne, because the large fixed share of a truck's cost is spread across a full load. A multi-drop delivery run is the most expensive, because the day gets consumed by drops rather than by distance.
That gives an asymmetry:
- If inbound is full loads and outbound is many small drops, lean towards the customers. Your inbound leg tolerates extra distance well because it is already consolidated. Your outbound leg does not
- If inbound is a steady container flow and outbound is a handful of large deliveries, lean towards the port. Now the consolidated leg is the outbound one
The second case is more common among manufacturers than people expect, and it is the case where a site far from the customer base can still be correct.
The general form of the rule: put the building near the end of the chain that cannot be consolidated, because that is the end where extra kilometres turn into extra vehicle-days rather than extra fuel.
Thailand narrows the map before you start
The generic method assumes demand is spread over a country. Here it is not.
As at September 2025 the Industrial Estate Authority of Thailand had 79 industrial estates and one industrial port, across 17 provinces. Fifteen of the estates it runs itself and 59 are joint operations, with five not yet open. Together they hold 5,615 factories and 1,083,180 jobs on 203,714 rai. The Authority's own regional read is that the East has the highest growth potential while the Central region remains the country's manufacturing and transport centre, and that the North and Northeast have so far attracted comparatively little interest.
If your customers or your suppliers are industrial, that is where they are. Two consequences follow.
A centre-of-gravity calculation over a national map will mislead you. It will place you at the weighted middle of your demand points, and in a country where the demand is bunched into a few provinces plus a metropolitan area, the weighted middle can be a place with no motorway, no labour pool and no counter-flow of freight.
Your realistic candidate list is short. That is good news. It means this decision can be costed properly rather than modelled abstractly, because there are perhaps five to ten places worth pricing, not five hundred.
Centre of gravity is still worth running. Just use it for what it is: a way of generating candidates. It weights by tonnes, but cost is driven by drops and days. It measures straight lines, and trucks drive roads. Take its answer, draw a circle around it, and cost the real sites inside that circle.
The number of sites is really a question about stock
Sooner or later somebody proposes two warehouses instead of one, usually to improve delivery times. This is where the fourth cost bites, and it is the one nobody puts in the business case.
As the number of sites goes up:
| Cost | Direction | Why |
|---|---|---|
| Delivery transport | Down | Shorter stem distance to each area |
| Inbound transport | Up, slightly | More destinations to feed, in smaller consignments |
| The buildings | Up | Two medium sites cost more than one large one, though not double |
| Inventory | Up | Each site needs its own safety stock |
| Systems and management | Up | More to run, more to reconcile |
The inventory line is the one that surprises people, and there is a rule of thumb for it. The square root law holds that total safety stock across a network is proportional to the square root of the number of locations. Run it forwards:
| Sites | Safety stock, indexed |
|---|---|
| 1 | 100 |
| 2 | 141 |
| 3 | 173 |
| 4 | 200 |
Splitting one warehouse into two raises safety stock by about 41%, not by nothing and not by 100%. Run the same law backwards and a consolidation from ten sites to five cuts safety stock by roughly 29%.
Now put that next to the interest rate. If the money in your stock costs around 7%, and stock is already the second largest block in the national logistics bill, a 41% increase in safety stock is a permanent annual cost that has to be paid for by the delivery saving before the second site is worth anything.
None of which says two sites is wrong. It says the case for two sites has to be made against inventory, not against delivery time alone, and that is the calculation most proposals skip.
What the site does to your freight rates, permanently
One more thing a location decides, and it is invisible until you are already there: whether trucks can find work in both directions.
The research on road distribution networks funded by the National Research Council of Thailand and the Thailand Research Fund, reported in 2018, looked at trucks serving the Free Zone at Suvarnabhumi Airport. Of the trips it sampled, 94.95% did one thing only, arriving for an import or for an export but not both. Measuring distance run without goods against total distance, four-wheel vehicles ran empty for 46.26% of their distance, six-wheelers for 38.45% and ten-wheelers for 23.27%.
The pattern is the point. Small vehicles run emptiest, because nobody bothers to arrange a return for them, and larger vehicles run fuller because somebody planned the pairing.
A location in an area with a healthy two-way flow of goods gives every carrier serving you a chance at a return load. A location where goods only ever flow one way does not, and the cost of that empty leg is inside every rate you are ever quoted there. It never appears as a line item. It is simply the reason a lane out of one estate prices differently from a visually identical lane out of another.
Worth asking before you sign, and easy to ask: which carriers already run into this area, and what do they carry out of it. A carrier that can see a return leg prices differently from one that cannot.
How to test a location before you commit
Do this on a spreadsheet, with your own last twelve months of orders, for each of your shortlisted sites. It takes a day and it outlives the lease.
- Take real order history, not forecasts. Twelve months of actual delivery lines: destination, weight, cube, and how many separate drops
- Group the destinations into delivery areas, the way a route planner would. Then measure two things separately: the stem distance from each candidate site to each area, and the drop distance inside each area. Only the first changes between candidates
- Convert to vehicle-days, not kilometres. Count how many drops fit in a working day from each site, remembering that the drop count usually fills the day before the distance does. This is the step that turns a distance comparison into a cost comparison
- Price the inbound leg separately, in full loads per month, not in tonne-kilometres
- Add the stock. If the candidate changes how many sites you run, apply the square root law to safety stock and cost the difference at your actual borrowing rate
- Then add rent, last, where it belongs in the order of magnitude
- Re-run the whole thing with next year's volumes, and again with a plausible bad year. A site that is only correct at one volume is a site you will regret
What to work out before signing a lease
- Split your delivery distance into stem and drop. If most of it is drop distance, moving the building will not save what the map implies, and the money is in routing rather than in location
- Decide which leg you cannot consolidate, and lean the site towards that end. Full loads absorb distance cheaply; multi-drop days do not
- Price the candidates in vehicle-days. Kilometres are the input. Days are the cost
- If more than one site is on the table, cost the safety stock first. One site to two is about a 41% increase, carried every year at your cost of money
- Ask what already moves in and out of the area, and in which direction. A location with two-way freight flow is buying you cheaper rates for the length of the lease
- Check the site as a work site, not as an address. Access for the vehicle types you actually use, room to turn, and somewhere for a truck to wait without blocking a public road
- Put the transport cost model in the lease decision paper. If the only number in the paper is rent per square metre, the paper is about the smallest of the four costs the decision fixes
The building is the visible part and the cheapest part. What you are really choosing is a set of routes that will be driven several thousand times, and a stock position that gets financed every day at the prevailing rate. Cost those two first, and the right building is usually obvious by the time you get to the rent.