A line-by-line audit of Thai central bank and NITMX PromptPay statistics: two calibers, five figures that do not reconcile, and the one that does.
SeaPRwire – In May 2026, Thailand’s PromptPay handled 2.38 billion transactions worth 4.45 trillion baht. That is the number National ITMX published, and it travelled widely. In the same month, the Bank of Thailand’s own statistics portal recorded 2,527,436.20 thousand transactions worth 4,663.95 billion baht for the same system. Both figures are official. Both come from Thai public institutions. They differ by 6.2 per cent in count and 4.8 per cent in value, and neither body has said the other is wrong.
That gap is not a rounding error and it is not a typo. It is the visible edge of a larger problem: Thailand’s national payment rail is described by at least five sets of headline numbers, and several of them are routinely quoted as if they measured the same thing. A reporter writing “PromptPay processes 2.5 billion transactions a month” can be right on three different definitions at once, and wrong on all of them if the sentence is meant to describe a specific month under a specific publisher’s rules.
This piece does the arithmetic the headlines skip. It takes the published figures, splits each one into numerator, denominator and time frame, and then asks which pairs genuinely cannot both be true. Some reconcile cleanly once the definition is stated. Some do not, and the reason they do not is more interesting than the error.
The two official totals, quoted as published
Start with the primary sources, because the disagreement only appears when the words are exact.
The Bank of Thailand maintains report PS_PT_018, “Use of PromptPay”, on its statistics web service. The page carries a remark that matters more than the table: transactions counted are those “by using easy-to-remember IDs such as mobile phone numbers, national identity numbers, corporate tax numbers, or e-Wallet numbers as well as traditional banking account numbers.” The last clause is the whole story. BOT counts bank-account-number transfers that never touch a proxy ID.
National ITMX, the clearing operator, publishes monthly figures through its own channels and press coverage. NITMX counts activity inside the network it operates.
Table 1: What BOT and NITMX each published for the same months in 2026, in billions of transactions and trillions of baht
| Month | BOT volume (bn) | NITMX volume (bn) | Volume gap | BOT value (THB tn) | NITMX value (THB tn) | Value gap |
| February 2026 | 2.359 | 2.21 | +6.7% | 4.521 | 4.28 | +5.6% |
| May 2026 | 2.527 | 2.38 | +6.2% | 4.664 | 4.45 | +4.8% |
Source: Bank of Thailand 🔗, PS_PT_018, updated 30 September 2026; NITMX monthly releases, February and May 2026
The gap is stable. It runs 6.2 to 6.7 per cent on count and 4.8 to 5.6 per cent on value across the two months where both publishers can be lined up. A gap that stable is not noise. It is a scope difference: BOT’s series is wider, because it sweeps in account-number transfers that NITMX’s network figure does not.
So the first contradiction dissolves on inspection. These are not rival estimates of one quantity; they are two different quantities that share a name. What breaks is any sentence that mixes them. Annualising the BOT monthly average and calling it “the NITMX figure”, or quoting NITMX’s registered-user count next to BOT’s transaction value, produces a composite number that no institution published and none would defend.
The registration count that exceeds the population
The second figure set is where the denominators stop being recoverable at all.
NITMX reported 83.44 million registered PromptPay identifiers as of August 2026, split into 83.02 million individual registrations and 0.42 million business registrations. Thailand’s population stood at roughly 66 million in 2025. The identifier count is therefore about 1.26 per head, which sounds absurd until you read the BOT remark again: mobile numbers, national ID numbers, tax numbers and e-Wallet numbers are all separate registrations. One person with a phone number, a national ID and a wallet account holds three.
Two other registration figures circulate. BoT’s governor, speaking at the Group of Thirty in 2026, cited “over 83 million registered users”. A trade publication covering the February 2026 systemic designation put the figure at “more than 90 million registrations in a country of 72 million people”. Three publishers, three totals: 83.44 million, over 83 million, more than 90 million. The 83 million pair agree; the 90 million figure does not, and it also carries a population figure about 6 million higher than the one usually cited for Thailand.
None of this is necessarily a mistake. Registration totals move, and “registrations” and “users” are different units. But the practical consequence is that no one outside the two institutions can state how many people actually use PromptPay. Registration counts are an upper bound on users and a lower bound on identifiers, and the identifier-to-person ratio is never published.
The business line is worse. NITMX counted 0.41 million business registrations in May 2026 and 0.42 million in August. That is 10,000 net additions between May and August 2026. Read as a 2026 percentage it looks healthy, about 2.4 per cent against 0.6 per cent growth in individual registrations over the same May to August window. Read as an economy-wide penetration rate it cannot be evaluated, because the number of Thai businesses that could register is not published alongside it. The Bank of Thailand’s own payment direction paper reports that 48 per cent of businesses regularly use digital payments, but gives no business-population denominator for that either. Two percentages about businesses, no shared base.
The average ticket that is four times too small
Here is a contradiction that does not dissolve so easily.
Divide BOT’s May 2026 value by its May 2026 volume: 4,663.95 billion baht over 2.527 billion transactions gives 1,845 baht per transaction. Do the same for every month in the 2026 BOT series and the answer sits between 1,731 and 1,944 baht. NITMX’s own 2026 monthly pairs give 1,727 to 1,987 baht. Two independent publishers, six months, one answer: the arithmetic mean PromptPay transaction in 2026 is around 1,700 to 2,000 baht.
The Bank of Thailand’s payment direction paper for 2022 to 2024 states a PromptPay average transaction value of 480 baht.
Table 2: Implied mean transaction value from official monthly aggregates, 2026, in baht
| Publisher and month | Value (THB bn) | Volume (bn txns) | Implied mean (THB) |
| BOT, March 2026 | 4,945.50 | 2.544 | 1,944 |
| BOT, April 2026 | 4,459.48 | 2.353 | 1,895 |
| BOT, May 2026 | 4,663.95 | 2.527 | 1,845 |
| BOT, July 2026 | 4,606.72 | 2.661 | 1,731 |
| NITMX, January 2026 | 4,630 | 2.33 | 1,987 |
| NITMX, August 2026 | 4,370 | 2.53 | 1,727 |
Source: Bank of Thailand 🔗, PS_PT_018; NITMX monthly releases, 2026
A mean of 480 and the May 2026 mean of 1,845 cannot both describe the same population. The most likely reading is not that either number is fabricated. It is that they describe different things and only one of them is labelled accurately.
The direction paper’s horizon is 2022 to 2024. Its 480 baht most plausibly describes an earlier year, a point-of-sale or retail subset, or a median reported as a mean. The same paper also states that 95 per cent of PromptPay transactions fall below 5,000 baht, which is entirely compatible with a mean near 1,845 baht if the distribution is heavily right-skewed. Take the two constraints together and solve: if 95 per cent of transactions average 480 baht, the remaining 5 per cent must average roughly 27,800 baht to pull the overall mean to 1,845. That top slice would then carry about 75 per cent of total value.
That result is worth stating plainly, because it is the opposite of what “average transaction 480 baht” implies about the system. On this reading PromptPay is not primarily a small-ticket retail rail with a few large transfers attached. It is a rail where a small minority of large transfers carries three quarters of the money, and where the small-ticket majority matters because of what it does to habit and cash use, not because of the value it moves.
No 2026 official source verifies the 27,800 baht figure. It is a solved residual, not a measurement, and no publisher reports the value distribution. That is precisely the gap: the single most consequential fact about PromptPay’s economics, how value is distributed across transactions, is not published by anyone.
Cash: 87, 63, and 54
The fourth set is the cleanest example of a missing denominator.
The Bank of Thailand’s payment direction paper reports that in the 2021 payment diary, cash accounted for 87 per cent of point-of-sale payments. The same body’s later materials report cash at 63 per cent of POS transaction value for 2021 and 54 per cent for 2023.
Two numbers describe 2021. For that same 2021 diary one says 87, one says 63. Both are attributed to the same institution.
The reconciliation is almost certainly count versus value. Cash dominates small purchases by number and loses share by value, because a street-food transaction and a department-store transaction are not the same size. An 87 per cent share by count and a 63 per cent share by value in the same year is entirely ordinary. But neither figure is published with its denominator stated on the face of the number, and the 54 per cent for 2023 is a value share. Anyone who lines up 87, 63 and 54 as a time series is reading a count figure into a value series and manufacturing a decline that is faster than the one that happened.
The direction paper also reports counter withdrawals rising at 4.1 per cent a year on a three-year average, while another summary of the same programme reports withdrawal value falling 10.4 per cent a year over five years. Again: count versus value, different windows, opposite signs. A reader who wants one sentence about whether Thais are taking cash out more or less can find official support for both.
The per-capita figures, and the one that cannot be rescued
Thailand’s per-capita PromptPay figures have three published values.
The IMF’s SIP survey puts fast payment transactions per capita at 40 in 2019 and close to 350 in 2024. The Bank of Thailand’s direction paper for 2022 to 2024 puts digital payments per capita at 651. NITMX’s August 2026 monthly figure, annualised and divided across a population of about 66 million, gives roughly 460 PromptPay transactions per person per year.
The IMF’s 2024 figure of 350 and the 2026 figure of 460 are reconcilable. They are three years apart, and PromptPay volume grew about 15 per cent year on year between 2025 and 2026. Roll 2024’s implied total back from a 2026 count at that growth rate and the two land within a few per cent of each other. Both describe PromptPay specifically.
651 does not reconcile, and this time the problem is real. The same 2022 to 2024 direction paper states that mobile and internet banking transactions run at 38 billion a year annualised. Across 66 million people that is 576 per capita from mobile and internet banking alone. Add PromptPay’s 460 and the total already exceeds 1,000 digital payments per person per year. The paper’s own published total of 651 is below what its 38 billion component implies for 2026.
The only reading that works is time frame. “Digital payments per capita 651” and “480 baht average ticket” belong to a paper whose horizon is 2022 to 2024, when PromptPay monthly volumes were lower and mobile banking had not reached 38 billion annualised. Set them against 2026 aggregates and you get a contradiction. Set them against their own year and they are plausible. The error is the comparison, not the numbers.
That is the general finding of this audit. Four of the five apparent contradictions are scope, denominator or time-frame mismatches rather than arithmetic failures. One, the average ticket, is a genuine labelling problem where a published mean is roughly a quarter of the mean implied by the same institution’s own aggregates, and no published denominator explains it.
The figure that does hold: 54 trillion baht
One headline survives contact with the arithmetic, and it is worth showing because an audit that finds only errors is not an audit.
The BoT governor stated that PromptPay moved 54 trillion baht in system transfers during 2025, more than twice GDP, with over 83 million registered users. Take BOT’s own monthly value series for 2026, which averages about 4.67 trillion baht a month. Twelve months at that 2026 rate is 56 trillion. For 2025, with monthly values a little lower, 54 trillion is exactly the right order of magnitude. The governor’s figure and the central bank’s statistical portal agree.
They should, since both come from the same institution. The point is that the number is internally consistent, which makes it the safest single figure to quote about PromptPay’s scale, and also the one that shows how much the small-ticket story depends on which slice you look at.
Cross-border: the number that shrinks when you give it a denominator
Cross-border QR is the part of PromptPay most often written about as a breakthrough, and it is the clearest case of a figure that collapses once you state what it is a share of.
NITMX reported inbound cross-border QR payments of 1.516 billion baht in May 2026, up 382 per cent year on year. China accounted for 911.66 million baht, more than 60 per cent of the total, followed by Malaysia at 318.46 million and Laos at 112.98 million. By August 2026 the three largest corridors had reached 1,278.63 million, 373.06 million and 138.32 million baht respectively.
Table 3: Inbound cross-border QR against total PromptPay value, 2026, in billions of baht
| Month | Inbound cross-border QR (THB bn) | Total PromptPay value (THB bn) | Share of system value |
| May 2026 | 1.52 | 4,450 | 0.034% |
| August 2026 | 1.79 | 4,370 | 0.041% |
Source: NITMX monthly releases via Nation Thailand 🔗, May 2026, and NITMX August 2026 release
Growth of 382 per cent in May 2026 sounds like a transformation. Measured against the system it sits inside, inbound cross-border QR was about three hundredths of one per cent of PromptPay value in May 2026 and four hundredths in August. Both statements are true, and they are not in tension. The 382 per cent is a May 2026 growth rate on a small base; the 0.04 per cent is a level. Any account of Thai cross-border payments that quotes one without the other is choosing the reader’s conclusion in advance.
There is a reason to keep watching the small number, though. A corridor that grows from nothing at 382 per cent a year compounds into something material within three years, and the composition told you where it starts: Chinese visitors, concentrated around holiday periods, spending in a narrow set of merchant categories. The cross-border story is a tourism story with a payments interface, which means it inherits tourism’s volatility rather than payments’ stability.
What to ask before quoting any PromptPay number
The practical output of this audit is a short checklist rather than a verdict. Before using a headline figure, establish four things.
- Which publisher issued it, BOT or NITMX, and never divide one by the other.
- Whether the count is transactions or value, and whether a share is a share of count or a share of value.
- Which month and which year, then check whether any second figure you are pairing it with comes from the same window.
- Whether an “average” is a mean of the whole system, a mean of a retail subset, or a median wearing the wrong label.
Three of those four are answerable from public material. The fourth is not, and until someone publishes the value distribution of PromptPay transactions, the most quoted average in Thai payments will stay unverifiable.
Where the reconciliation stops paying: the rail underneath
The most consequential unpublished number in Thai payments is not a PromptPay number at all.
The Bank of Thailand has stated in its own policy material that BAHTNET, the high-value real-time gross settlement system, is a legacy system in service for more than 30 years. As described in 2026 central bank material, it operates roughly nine hours a day, about 32,400 seconds, and five days a week. PromptPay settles through it. Outside those hours, transfers queue, and the central bank’s consultation paper puts PromptPay’s overnight and holiday settlement exposure at 7.8 billion baht.
On 21 February 2026 BOT designated PromptPay a systemically important retail payment system, giving NITMX 90 days to comply, to roughly 22 May 2026. The framework requires a two-hour recovery time objective, a board at least one third independent, and annual emergency testing. In 2026 trade coverage of the February designation, PromptPay was described as processing about 74 million transactions a day against more than 90 million registrations.
Set that against the direction paper’s own admission that cash still holds 54 per cent of POS transaction value and that in 2026 small businesses avoid digital payments partly because of tax scrutiny, and the picture stops being a success story with footnotes. Thailand built one of the world’s most-used retail instant payment rails, then left the wholesale layer underneath it running a nine-hour day. The retail system is now formally systemically important. The layer it settles on is formally described by its owner as legacy.
Methodology
Data as of September 2026. The primary metric is monthly PromptPay transaction count and value, as published by the Bank of Thailand in report PS_PT_018 and by National ITMX in monthly releases. BOT figures are taken in reported units (thousand transactions, billions of baht) and converted; NITMX figures are taken as published, which are rounded to two or three significant figures, so gap percentages carry roughly plus or minus 0.3 points of imprecision. Implied mean ticket size is value divided by volume for the same month and publisher, with no seasonal or calendar adjustment. The 5 per cent tail estimate is a solved residual from two published constraints (mean of 1,845 baht; 95 per cent of transactions below 5,000 baht) and is not a measurement. Population is taken as approximately 66 million and is an assumption, not a published figure in any payment source. No cross-caliber splicing was performed: BOT counts were only ever divided by BOT counts.
Limitations
- The two monthly gaps rest on two months where both publishers can be lined up; NITMX figures are rounded, so the true gap may sit between 5.9 and 6.9 per cent.
- The 27,800 baht tail estimate assumes the 480 baht figure is the mean of the bottom 95 per cent. If it is a median or a point-of-sale subset, the residual is not valid.
- Registration-to-user ratios cannot be verified because no publisher discloses identifiers per person.
Sources
- Bank of Thailand 🔗, PS_PT_018 Use of PromptPay, updated 30 September 2026
- National ITMX 🔗, PromptPay monthly statistics releases, January to August 2026 (site home)
- Nation Thailand 🔗, PromptPay May 2026 volume, registrations and inbound cross-border QR, June 2026
- ClearingPost 🔗, Thailand applies central bank-grade oversight to PromptPay under SIRPS, 2026
- International Monetary Fund, SIP survey figures for Thailand fast payment transactions per capita, 2019 and 2024 (no public link)