Platform GMV of $157.6 billion, total trade of $185.5 billion, an e-Conomy estimate of $191 billion: one year, three denominators, no reconciliation
SeaPRwire – The number travels well. Southeast Asian ecommerce was worth $185.5 billion in 2025, according to Momentum Works, the Singapore-based venture builder whose annual report has become the region’s most quoted source on the subject. The figure appears in bank decks, in ministry slide packs, in the opening paragraph of at least a dozen funding announcements I read this month. It is repeated with the confidence of a customs return.
It is not a customs return. Nothing in this market is. The $185.5 billion is an estimate assembled from platform disclosures, interviews and modelling, published without a stated estimation method. That is not a scandal — private companies do not file GMV with a statistical agency — but it does mean the number carries a property that almost nobody quoting it acknowledges: it is one of three different totals for the same year, produced by two organisations using different definitions, and the gaps between them are large enough to change conclusions.
Three numbers, one year. Momentum Works puts 2025 platform GMV at $157.6 billion and non-platform trade at $27.8 billion, for $185.5 billion total. Google, Temasek and Bain, in the e-Conomy SEA 2026 report released on 21 October, put 2025 ecommerce GMV at $191 billion, rising to a forecast $226 billion in 2026. And in the previous edition of the same e-Conomy series, published eleven months earlier, the 2025 figure was $185 billion.
Same year. Same region. Three totals spanning $185 billion to $191 billion, a spread of $6 billion, or 3.2 percent. That spread is not the story by itself — 3 percent is within the normal error band of any estimate this size. The story is what happens when you multiply that error band across the growth rates, the segment splits and the seller-cost ratios that get built on top of it.
Three totals, and what each one is actually counting
The definitions diverge in ways that are rarely stated in the same sentence as the number.
| Source | 2025 figure | What it counts | What it excludes |
| Momentum Works, April 2026 | $157.6B platform GMV | Transactions completed on marketplaces | brand.com, multi-brand retail, social selling |
| Momentum Works, April 2026 | $185.5B total | Platforms plus non-platform channels | informal commerce, cross-border under-declaration |
| e-Conomy SEA 2025, Nov 2025 | $185B ecommerce GMV | Marketplaces, brand DTC, groceries, video commerce | informal commerce |
| e-Conomy SEA 2026, Oct 2026 | $191B ecommerce GMV (2025) | Same basket, restated | Same exclusions |
Read the last two rows together. The e-Conomy series restated its own 2025 estimate upward by $6 billion in twelve months, from $185 billion to $191 billion. That is a 3.2 percent revision to a base year, disclosed in a footnote rather than a correction notice, and it is entirely defensible — new platform disclosures arrive, methodology improves, base years get rebuilt. Every statistical series does this.
But it has a consequence that nobody in the coverage mentioned. When the 2026 report says ecommerce will grow 18 percent to $226 billion, that 18 percent is computed against the restated $191 billion, not the $185 billion the same organisation published a year earlier. An analyst who built a 2026 forecast last November — off $185 billion, plus the growth rate then on offer — is now working from a base that has moved 3.2 percent without any real-world event having occurred. If that analyst also models platform take rates as a share of GMV, the error propagates into revenue, into unit economics, into the valuation of every marketplace in the region.
The Momentum Works figures sit on the other side of the same problem. Its $157.6 billion is a platform-only number, and it is the one that gets compared, incorrectly, with e-Conomy’s $191 billion. The two organisations are counting different universes: $157.6 billion is money that moved through Shopee, TikTok Shop and Lazada; $191 billion includes brand.com and grocery and video commerce that never touches a marketplace checkout. Put side by side without labels, the $33 billion difference looks like an undercount. It is not. It is a definition.
The honest comparison is $185.5 billion against $191 billion — a 3.0 percent gap between two organisations trying to measure roughly the same thing. That is a respectable convergence for a market with this much informal trade. The dishonest comparison, which appears constantly, is $157.6 billion against $191 billion, which manufactures a 21 percent discrepancy out of a labelling error.
The 15 percent that grows three times faster than the 85 percent
Here is the number inside the Momentum Works data that deserves more attention than the headline.
Platform channels were 89 percent of total ecommerce in 2024 and 85 percent in 2025. Work backwards from that. If platform GMV was $128.3 billion in 2024 at an 89 percent share, total trade was about $144.2 billion, which puts non-platform channels at roughly $15.9 billion. In 2025, non-platform channels were $27.8 billion.
That is growth of about 75 percent in a single year, against 22.8 percent for the platforms. The segment that nobody measures properly, that has no standard definition, that includes brand.com stores, Facebook Live sellers, WhatsApp catalogs and multi-brand retailers, grew more than three times as fast as the segment with audited, quarterly-reported, investor-scrutinised numbers.
I am not suggesting the $27.8 billion is wrong. I am pointing out that it is the fastest-growing part of the market and the least verifiable part of the market, and that the share shift from 89 to 85 percent is the single most consequential line in the report. If that trend continues at the same rate, non-platform channels pass 30 percent of the market before 2029, and every platform-take-rate model in the region quietly breaks.
There is one more thing about the 85 percent figure. It fell four points in a year. Momentum Works attributes the shift to brand.com and social selling. But 2025 was also the year Bukalapak exited physical ecommerce and TikTok completed its absorption of Tokopedia — two events that moved volume between categories of channel without adding any. Some of that four-point shift is genuine diversification. Some of it is a marketplace becoming a different kind of marketplace. The report does not separate the two, and neither does anyone quoting it.
Two 32 percent claims that cannot both be true
Content commerce is the other number doing heavy lifting in this year’s coverage, and it is where the denominators come apart entirely.
Momentum Works estimates content commerce at $49.7 billion in 2025, or 32 percent of platform GMV, up from 20 percent in 2024. That is internally consistent: $49.7 billion divided by $157.6 billion is 31.5 percent, which rounds to 32.
A separate widely-cited analysis puts live commerce at 32 percent of Southeast Asian ecommerce GMV. Same percentage, different base. If the base is $185.5 billion, 32 percent is $59.4 billion. If the base is $191 billion, it is $61.1 billion.
| Claim | Stated share | Implied base | Implied value |
| Content commerce, Momentum Works | 32% | $157.6B platform GMV | $49.7B |
| Live commerce, trade analysis | 32% | $185.5B total ecommerce | $59.4B |
| Live commerce, trade analysis | 32% | $191B e-Conomy total | $61.1B |
Live commerce is a subset of content commerce — a livestream is content, but not all content is live. So content commerce should be the larger number. In the table above, applied to their stated bases, it is the smaller one by $10 billion to $11 billion. Both figures can be sourced. Both are probably close to right on their own terms. They cannot both be describing the same market in the same year with the same word.
What has happened, almost certainly, is that one 32 percent was computed on platform GMV and the other on total ecommerce GMV, and both were then written up as “32 percent of the market” because 32 is a memorable number and the adjective got dropped in editing. This is how a definitional difference becomes a fact. It will now be cited for three years.
What a seller actually pays, and why the 30 percent is not a take rate
The number that matters most to the 57 percent of Southeast Asian ecommerce transacted by small and medium enterprises is not GMV at all. It is cost.
Momentum Works estimates total seller operating cost at about 30 percent of GMV. That figure gets repeated as though it were the platform commission, and it is not. Pull the platform’s own disclosures apart and a different picture emerges.
| Line item | Amount, FY2025 | As share of Shopee GMV ($83.2B) |
| Shopee cost of services | $9.5B | 11.4% |
| Shopee sales and marketing | $3.55B | 4.3% |
| Combined platform-side cost | $13.05B | 15.7% |
| Seller total operating cost (all-in) | not disclosed | ~30% |
The gap between 15.7 percent and 30 percent — roughly 14 points of GMV — is logistics, packaging, returns, ad spend bought outside the platform, affiliate commissions, and the seller’s own labour and warehousing. Lazada alone spends about $100 million a year on affiliate marketing, money that flows to creators and not to the platform’s P&L. None of that is a platform fee. All of it is a seller cost.
The distinction matters because the policy conversation in Jakarta, Kuala Lumpur and Bangkok has converged on the idea that platform commissions are squeezing small sellers. That may be true. But the evidence being used for it — a 30 percent all-in cost ratio — does not measure commission, and the platform-side number that does measure it is roughly half as large. A government that caps commission at, say, 12 percent would leave the 30 percent almost untouched, because two-thirds of it was never commission.
The counter-argument is that platforms capture the logistics and ads revenue too, through affiliated delivery arms and in-app advertising, so the 15.7 percent understates their real take. That is fair, and it is exactly the point: nobody has published a consolidated figure that nets out intra-group flows. The number that would settle this is total platform revenue from Southeast Asian sellers, all lines, divided by Southeast Asian GMV. It does not exist in public.
Methodology: what each choice buys, and who it favours
Six decisions sit underneath every one of these figures. Naming them is the only way to compare across sources.
- GMV is the value of goods sold, before returns, discounts, subsidies and fraud. It is not revenue and not profit. A market can grow GMV while every participant loses money, and in 2025 several did.
- Base year. e-Conomy restated 2025 from $185 billion to $191 billion. Momentum Works did not restate. Any cross-source growth rate therefore mixes two base years.
- Channel boundary. Platform-only versus platform-plus-non-platform is a $28 billion difference in 2025. Most coverage does not say which it is using.
- Informal trade. Both sources exclude it, for the same reason: no reliable data. In Indonesia and the Philippines that exclusion is large, and it is systematically biased toward the formal, urban, brand-heavy end of the market.
- Everything is reported in US dollars. The rupiah moved enough in 2025 that a flat dollar figure can conceal local-currency growth or decline. Indonesia’s 2.2 percent dollar growth is not necessarily 2.2 percent in rupiah.
- Who benefits from the choice. Platforms benefit from platform-only GMV, because it is the largest number that is entirely theirs. Governments benefit from total GMV, because it is the largest number that is entirely their jurisdiction. Both benefit from excluding informal trade, because it cannot be taxed.
That last point is not cynicism. It is the reason the definitions differ, and the reason they will not converge.
Where the informal economy hides
Consider Indonesia, which at 37 percent of platform GMV is the largest single market in the region and grew 2.2 percent in 2025 — the slowest of the six.
That 2.2 percent is doing a lot of work. It is a dollar figure, on a platform-only base, in a year when Bukalapak left physical ecommerce and Tokopedia’s GMV was rationalised after the TikTok takeover. Two structural removals, one currency effect, and a residual growth rate that is presented as market maturity. It might be maturity. It might also be that the market moved to channels the report does not count — social commerce, WhatsApp groups, TikTok Live outside the Shopee ecosystem — and the 2.2 percent is measuring the part that stayed put.
The Philippines gives the counter-example. It grew 32 percent and overtook Vietnam as the region’s third-largest ecommerce market. Nothing about that is ambiguous. When a market is adding platform-sold volume fast enough, the definitional questions stop mattering, because the direction is the same under every definition. Definitions matter most in slow markets, and the slowest market in the region is also the biggest one.
The AI numbers are survey answers wearing a market size
There is a fourth set of numbers in this year’s coverage, and it has a different problem from the others. These are not estimates of trade. They are survey responses, and they are being quoted as though they measured behaviour.
The e-Conomy SEA 2026 report says 69 percent of consumers in six countries use AI to compare products, that sellers see a 10 to 15 percent conversion uplift from AI tools, and that 75 percent of users would be willing to let an AI agent find a price and complete a purchase. All three have travelled into bank research and vendor decks this month with the word “consumers” attached.
Read the methodology and the population changes. The AI survey was conducted by Google and YouGov in July 2026 with 9,006 respondents across Indonesia, Malaysia, the Philippines, Singapore, Thailand and Vietnam. It covers existing AI users, not the general population of those six countries, and not the other four markets the report measures. Seventy-five percent of that group said they would be comfortable delegating a purchase, conditional on the assistant being generally reliable. That is a statement about stated willingness among people who already use AI, in six of ten markets, under an assumption the respondent was asked to accept.
The gap between that and “three quarters of Southeast Asian shoppers are ready to hand over their wallets” is not a nuance. It is the difference between a market forecast and a survey question. The same report’s companion analysis says plainly that transactions remain largely human-led, and that the survey measures intention rather than completed orders.
| Claim as reported | What the sample actually was | What it can support |
| 69% of consumers use AI to compare products | 9,006 existing AI users, 6 of 10 markets, July 2026 | adoption among AI users, not population share |
| 75% willing to let agents purchase | conditional willingness, same sample | pricing research, not order volume |
| 10–15% seller conversion uplift | no sample size disclosed | a range to test, not a benchmark |
| 243 million GenAI adopters | modelled from population and survey data | scale indicator only |
The 10 to 15 percent conversion uplift is the one I would treat most carefully. It has no disclosed sample, no control group and no stated definition of conversion. It is a useful range for a seller deciding whether to test AI product video. It is not a number to underwrite a logistics investment against.
None of this makes the survey worthless. Nine thousand respondents across six countries is a serious instrument, better than most trade research in the region. The failure is in the transmission: the population qualifier drops out in the second paragraph of every write-up, and what lands in the deck is a market-size claim wearing a survey’s clothes.
The number that would settle this
If I could ask the two organisations one question each, it would be the same question.
To Momentum Works: what is the total GMV of the non-platform channels, by country, and what method produced the $27.8 billion? A 75 percent year-on-year jump in an unmeasured segment is either the most important fact in Southeast Asian retail or a modelling artefact, and the report does not let a reader tell which.
To Google, Temasek and Bain: what changed between the 2025 and 2026 editions that added $6 billion to the same year? A base-year revision of that size is normal, but it needs a note, because every growth rate in the region was computed off the old base.
Neither answer would be commercially sensitive. Both would let the market stop treating three numbers as one.
Until then, the useful discipline is simple. When you see a Southeast Asian ecommerce figure, check three things before you use it: whether it is platform-only or total, which edition of which report it came from, and whether the growth rate was computed on a restated base. The first check catches most errors. The second catches the rest.
The 85 percent of this market that runs through three platforms is measured about as well as any emerging-market retail sector can be. The 15 percent that is growing three times faster is not measured at all. That is the actual state of knowledge about a $185 billion market, and no headline number should be allowed to obscure it.
Sources
- Momentum Works 🔗 — “Southeast Asia’s Ecommerce Market Reaches US$157.6B in 2025, Top platforms expand share to 98.8%”, press release for Ecommerce in Southeast Asia 2026, 14 April 2026
- The Straits Times 🔗 — “South-east Asia’s digital economy to hit US$368b in 2026, up 17% as AI uplifts growth”, 21 October 2026
- Yicai Global 🔗 — “SE Asia’s E-Commerce Sites See GMV Jump 23% to Over USD157.6 Billion in 2025”, 15 April 2026
- Lianhe Zaobao 🔗 — “Southeast Asian ecommerce reaches US$157.6 billion as the market enters a three-platform monopoly,” 13 May 2026（Chinese-language original）
- Databoks Katadata 🔗 — “E-Commerce Bolsters Southeast Asia’s Digital Economy in 2026”, 27 October 2026
- Sea Limited 🔗 — FY2025 results, cost of services and sales and marketing disclosures（site home）
- Bain & Company 🔗 — e-Conomy SEA series, report library（site home）
- Momentum Works, Ecommerce in Southeast Asia 2026, full report, April 2026（no public link）