Somebody senior asks how your transport costs compare, and the honest first answer is a question back: compare against what?
That sounds like evasion. It is not. Most benchmarking work in this industry fails at that exact point, because the comparison gets chosen before anybody decides what it is meant to prove. Six weeks later a deck appears with a national percentage on it, everyone nods, and nothing changes, because a national percentage was never going to end in an instruction anybody could carry out.
A benchmark earns its keep only if it can finish as a number of baht you could go and get. Everything below is aimed at that.
First decide which question the comparison is meant to settle
Three different questions get bundled into "how do we compare", and they need three different comparisons.
- Am I paying too much for the trucks I buy? A price question. It is settled against other prices for the same work, and only after the two prices have been made comparable.
- Am I running my own operation badly? A productivity question. It is settled against other operations doing similar work, and the closest ones are usually inside your own company.
- Is the shape of my distribution wrong? A design question. It is not settled by any benchmark at all. It is settled by modelling your own network, because the answer depends on where your customers are.
Getting this wrong is the most expensive mistake in the whole exercise. A team that suspects its network design is wrong, and answers the question by comparing rates, will find its rates are fine and stop looking.
The national number is not your number, and the arithmetic shows why
The figure everyone reaches for is Thailand's logistics cost as a share of GDP. NESDC puts it at 13.5 percent for 2024, or 2,509.4 billion baht, and projects 13.4 to 13.8 percent for 2025.
It is a real number, carefully built. It also cannot say anything about your company, and it is worth being precise about why.
Start by opening it. Transportation is 47.9 percent of it, inventory holding 44.7 percent, and administration 7.4 percent. So nearly half the number people wave at transport managers is the cost of holding stock, not of moving it.
Then open the transport half. Road freight is 42.8 percent of transport costs, water 23.8 percent, transport-related services such as freight forwarding and customs brokerage 16.7 percent, parcel 8.6 percent, pipeline 4.4 percent, air 3.5 percent and rail 0.2 percent.
Multiply that through. Road freight is 42.8 percent of 1,200.6 billion baht, or 513.8 billion. That is 20.5 percent of the national logistics bill. Against GDP, transport is 6.5 percent and road freight is 42.8 percent of it, so road freight is about 2.8 percent of GDP.
The headline is 13.5. The part of it that a truck touches is about 2.8.
There is a second reason not to treat this as a yardstick for your own overheads. The administration component is not measured at all. NESDC states that it is assumed at 8.04 percent of transportation and inventory holding costs combined, following its own 2019 calculation model. Check it: 8.04 percent of 1,200.6 plus 1,122.1 billion is 186.7 billion, which is exactly the published figure. Comparing your admin cost to a national admin cost is comparing yourself to a fixed ratio.
None of this is a criticism of the report. It is a national accounts estimate and it does that job well. It simply answers a question about the country, not about you.
Where the gap with the United States actually sits
The comparison that follows the 13.5 percent is always the United States at 8.7 percent, and the conclusion drawn from it is always that Thai transport is inefficient.
The two structures are not alike, so convert both into points of GDP before comparing anything.
Thailand: transport 6.5 points, inventory holding 6.0, administration 1.0. Total 13.5.
The United States, on the CSCMP figures NESDC reproduces, splits 2,580.8 billion USD into transportation 1,663.5 billion, inventory carrying 734.7 billion and other costs 182.6 billion. As shares that is 64.5, 28.5 and 7.1 percent, and applied to 8.7 points of GDP it gives transport 5.6 points, inventory 2.5 and other 0.6.
Now the gap is readable. Of the 4.8 point difference, 3.5 points are inventory holding, 0.9 points are transport and 0.4 points are administration. Roughly three quarters of the gap is stock sitting still. Under a fifth of it is goods moving.
That changes what you do with the number. If somebody hands you the 13.5 percent as a transport problem, the arithmetic says most of it is a stock problem, and the levers are order frequency, safety stock and how many places you hold inventory in. Delivery frequency moves that half of the bill directly, and it is a transport decision, which is why the two halves are worth keeping straight.
The index is a benchmark for movement, never for level
The Ministry of Commerce publishes a road freight index each quarter, on a 2021 base of 100. In the first quarter of 2026 it stood at 112.4 by commodity structure and 115.2 by vehicle structure.
Read its own description of what it is before using it. The commodity index is built from 51 priced items across 57 provinces, taken from freight rates per trip, per volume, per distance or per piece. The vehicle index is built from 16 items across the same 57 provinces, in the seven vehicle categories of the Land Transport Act.
Then read the weights, because they decide whose market it describes. Industrial products carry 85.24 percent of the commodity index, agricultural and fishery products 12.20 and mining 2.56. On the vehicle side, pickup trucks carry 57.59 percent and trailers 21.78, while liquid tankers are 1.93 and dangerous goods vehicles 0.96.
So the national road freight index is mostly manufactured goods on pickup trucks. If you move chilled produce on trailers, the headline is not your market and the relevant sub-index is. That matters most at budget time, and the machinery for using it properly is in budgeting when rates move.
For benchmarking, the limit is simpler and absolute. An index measures change, not level. It can tell you that your rate rose faster than the market. It can never tell you that your rate is high, because 100 was only ever the base year, not a fair price.
The peer benchmark almost nobody uses is a legal right
The comparison people actually want is against other companies, and they assume it is impossible because nobody publishes their costs. In Thailand, more is available than most buyers realise, and it is available by statute rather than by favour.
Under the Accounting Act, a limited company established under Thai law files its financial statements within one month of the day the general meeting approved them. A registered partnership, a foreign juristic person or a joint venture files within five months of the closing date, and accounts must be closed every twelve months. Those statements must be audited and opined on by a certified public accountant, with one exception worth remembering: a registered partnership under Thai law whose capital, assets or revenue fall under the ministerial thresholds does not need an audit.
Then Section 18 does the useful part. Financial statements filed under Section 11 may be inspected, or a copy obtained, by an interested person or by any member of the general public, on payment of the fee the Director-General sets. That is a right of access, not a service somebody can decline to give you. The Department of Business Development also runs an online service showing the key items of the income statement and the statement of financial position, with a search by TSIC business code.
Be clear about what this can and cannot do.
It cannot give you a cost per kilometre. Nobody files that, and no filed statement will tell you what a carrier's cost to serve your lane is.
It can tell you three things that change decisions. Whether a carrier that has just quoted you an aggressive rate is making money at all. What gross margin the segment actually runs at, which sets the realistic ceiling on how much a negotiation can win. And whether the company you are about to hand a year of volume to is growing, shrinking or being kept alive by its shareholders. The last of those is worth more than the rate.
Two cautions. Statements are old by the time you read them, so treat them as a picture of a finished year rather than of today. And an unaudited partnership filing is not evidence in the way an audited company filing is.
Normalise before you compare
Whatever you end up comparing, the comparison is worthless until both sides mean the same thing. The textbook list of pitfalls in benchmarking data is availability, sampling, consistency of accounting treatment, category definitions, time periods and units of measure. In Thai road freight those turn into a short and very concrete checklist.
- Scope. Does the figure include waiting time, loading labour, tolls, a second crew member, pallet returns? A rate with unloading in it is not comparable to one without, and neither is wrong.
- Tax treatment. Domestic road freight is exempt from VAT and withheld at 1 percent. The same work invoiced as a bundled service carries 7 percent VAT and 3 percent withholding. Two figures can differ by the whole of that before a wheel turns, so check what the invoice says the service is before comparing.
- Period. Twelve calendar months against thirteen four-week periods is the classic trap, and Thailand adds its own: a period containing Songkran, or a month-end delivery spike you created yourself, is not a normal period.
- Denominator. Cost per kilometre on loaded kilometres and cost per kilometre on all kilometres are different numbers from the same operation. Which unit hides what is the whole subject of cost per unit.
- Own fleet against bought-in. An internal cost that leaves out driver overtime, depreciation, the yard and the vehicle that stood idle is not comparable to an invoice, which includes all four.
- Controllable against not. Diesel, tolls and the minimum wage are not yours. Strip them out before you rank anybody, or you will rank the market and call it performance.
That last one is the one people skip. A depot can look expensive because its building is expensive, and no amount of management effort will change the rent.
The league table that actually pays is your own
External benchmarking has a structural weakness that is rarely said out loud: a figure collected from a spread of contributors is an average of the best and the worst of them, and the stated aim of the exercise was to match the best. Matching an average is not an ambition.
The internal comparison has none of that problem. The definitions are already identical, the data already exists, nobody is hiding anything, and the gap is provably reachable, because somebody in your own operation has already reached it.
Here is the shape of it, with numbers.
Take one delivery operation: six vehicles, six delivery days a week, over eight weeks. That is 48 delivery days each and 288 vehicle-days in total. At 14 drops a vehicle-day, each vehicle makes 672 drops in the window and the operation makes 4,032.
Rank the six by cost per drop over the whole window: 282, 301, 327, 355, 398 and 505 baht. The median is 341.
Three vehicles sit above it, by 14, 57 and 164 baht a drop. Bring those three to the median and the saving is 672 multiplied by 235, or 157,920 baht over eight weeks. Annualised at 6.5 windows a year, that is 1,026,480 baht.
That is the number that goes in the first draft of the deck, and it is wrong.
What the number is worth after you group like with like
Before claiming a million baht, ask what is different about the expensive vehicles, and use the same list any benchmarking study uses: the operating environment, not the operator. Route type, drop density, customer profile, vehicle class, the receiving sites' opening hours.
Say two of the six run upcountry lanes with 40 kilometres between drops, and four run urban rounds with three. A league table that mixes them is ranking geography.
Regroup. The four urban vehicles are 282, 301, 355 and 398, with a median of 328. Two sit above it, by 27 and 70 baht. That is 672 multiplied by 97, or 65,184 baht over eight weeks, and 423,696 baht a year.
The upcountry pair, 327 and 505, is not a league table. Two data points have no middle. The honest output there is not a saving, it is an instruction to go and look at why one costs 178 baht a drop more than the other.
So the defensible prize is 423,696 baht rather than 1,026,480, which is a haircut of nearly three fifths. That haircut is the entire value of the exercise. The first number would have been challenged in the meeting and lost. The second one survives, because every objection to it was already answered before it was written down.
Where to start this month
- Write down the question first, in one sentence, and check it can end in an action. If it cannot, do not run the comparison.
- Stop using the national ratio as a verdict on your operation. Use it to argue about the country, and remember that most of the gap with other countries is stock, not trucks.
- Use the road freight index for direction only, and use the sub-index that matches your commodity and vehicle rather than the headline.
- Pull the filed accounts of your three largest carriers. Not for their cost, which is not there, but for whether they can survive the rate they just gave you.
- Build one internal league table on one measure. Cost per drop is the usual best first choice, because it is the unit a vehicle-day actually produces.
- Group like with like before you publish a saving, and put the group definitions on the same page as the number.
- Keep the whole exercise to one process at a time. Studies of this kind generate far more data than insight, and the discipline that makes them work is refusing to look at everything at once.
A benchmark is not a score. It is a way of finding a gap that somebody has already proved is closeable, and then closing it.