Where SME multiples actually come from
Every multiple you have read was published by someone selling something. Here is who collects the data, how, and why two datasets covering the same businesses disagree by a factor of two.
Search for what a business like yours is worth and you will have a number within about ten seconds. Somewhere between two and six times earnings, depending on which page you land on.
Take that range seriously for a moment. A business earning €500,000 a year is worth €1,305,000 at 2.61 times and €2,750,000 at 5.5 times. Both of those multiples were published in 2025, by organisations describing businesses of roughly that size. The gap between them is €1,445,000, which is more than the lower of the two valuations.
So it matters where a multiple came from. The answer is nearly always the same: someone who makes money from business sales collected it from other people who make money from business sales.
There is no official data. None.
Start here, because it eliminates a whole category of claim. No national statistics office anywhere publishes SME transaction multiples. Eurostat's structural business statistics and the OECD's equivalent cover how many businesses exist, what they turn over, how many people they employ and how much value they add. Not what they sell for.
If a page tells you that official figures show businesses in your sector selling at some multiple, it is either mistaken or lying. There is no such figure to cite.
Who actually publishes numbers
| Source | Who publishes it | Where the data comes from |
|---|---|---|
| DealStats | Business Valuation Resources, a data vendor | Broker self-reports plus regulatory filings |
| IBBA Market Pulse | Two trade bodies for business brokers | A quarterly survey of brokers' recollections |
| BizBuySell | A business-for-sale marketplace | Deals brokers voluntarily report to the platform |
| GF Data | The Association for Corporate Growth | Private equity buyers, reported confidentially |
| BDO PCPI | An accountancy and advisory firm | A third-party company register |
| Brookz Barometer | A Dutch business-for-sale marketplace | A survey of Dutch M&A advisors |
Every one of these has a commercial interest in the market it measures. That is not an accusation, it is the structure of the field: nobody else has any reason to collect the data.
The degree of conflict varies, though, and it is worth knowing which way. DealStats draws roughly 70% of its transactions from a contributor network of brokers and advisors, with the rest from regulatory filings, and those contributors are people whose marketing benefits from high reported multiples. The IBBA Market Pulse does not record transactions at all; it asks brokers what they remember from the quarter. GF Data is the interesting one, because its reporters are private equity buyers rather than sellers' agents, and buyers have no reason to overstate what they paid.
The disagreement is bigger than the effects being claimed
Here is the part that should change how you read every valuation article, including this one. Three published 2025 figures, all describing businesses an owner might plausibly sell. "Enterprise value" below means the price for the business itself, before the cash and the debt are settled (the difference is explained here).
- BizBuySell, 2025
- 2.61×
- median cash flow multiple, 9,586 closed US deals
- GF Data, H1 2025
- ≈5.5×
- EBITDA, deals with $1m–5m enterprise value
- Brookz, H1 2025
- 3.5×–4.1×
- EBITDA, Dutch SMEs at €200k–€1m EBITDA
These populations overlap substantially. A business with €500,000 of normalised earnings could plausibly appear in any of them. The published medians differ by roughly a factor of two.
Part of that gap is definitional, and it is worth being straight about it. EBITDA is earnings before interest, tax, depreciation and amortisation. BizBuySell's "cash flow" is closer to seller's discretionary earnings, which adds the owner's own salary back on top. A bigger earnings figure divided into the same price gives a smaller multiple, so some of the distance between 2.61× and 5.5× is arithmetic rather than disagreement. Not all of it, though, and no publisher tells you how much.
The rest is a mix of real differences and measurement. Some of it is certainly real: the deals private equity buys are not the deals a marketplace lists, and buyer type genuinely moves price. In the eurozone mid-market, the Argos Index put investment funds at 10.0× against 7.8× for strategic buyers in the first quarter of 2026. But some of it is measurement, and no published source tells you how much of which.
Now hold that next to the adjustments the same industry claims to measure. You will read that customer concentration costs "a full turn" on your multiple, or that owner dependence costs another. Those claimed effects are smaller than the disagreement between the datasets that would have to detect them. A one-turn adjustment inside a two-turn measurement gap is not a finding. It is noise with a decimal point.
What the better sources admit about themselves
The most useful thing in this whole literature is the disclaimers, because a source that tells you its own weakness is a source you can use.
BDO says of its own index that disclosed profits "tend to be suppressed by various expenses that may be non-recurring under a new owner," so the multiples it calculates may be overstated, and that the index is "an average measure and a guide, not an absolute measure of value."
Business Valuation Resources notes that each point in its sector chart rests on a minimum of ten transactions, and disclaims accuracy outright: the information "may be condensed or incomplete."
BizBuySell states that its deals are "reported to BizBuySell.com on a voluntary basis."
None of those sources publishes a confidence interval. None reports how many transactions sit behind an individual band. The IBBA Market Pulse spreads roughly 250 to 300 reported transactions across five size bands each quarter, which puts perhaps twenty to sixty deals behind each figure, and its published series has moved implausibly far between editions. Read those band multiples as noise rather than signal, whichever direction they happen to move.
How to use any of this
Use the direction, not the decimal. Larger businesses sell for more than smaller ones. That replicates across every dataset above, collected in different ways on three continents. It is the most robust finding in the field. The specific gap between two adjacent size bands is not.
Match the population before you use a figure. A multiple drawn from private-equity-sponsored deals above ten million in enterprise value tells you almost nothing about an owner selling to another owner. If a comparable does not state its size band, buyer type and country, it is decoration.
Ask who paid for the number. Not to dismiss it, but to know which way to discount it. A figure published by brokers who are paid a percentage of the sale price and a figure reported confidentially by the buyers who wrote the cheque are not the same kind of evidence.
Expect a range, and expect it to be wide. Anyone handing you a single number for what your business is worth is either hiding the range or has not calculated it.
What this does not tell you
This article is about the quality of the evidence, not about your business. Knowing that the published multiples are shaky does not tell you what yours should be. It tells you how much weight to put on a comparable when someone quotes one at you across a table.
It also does not mean the numbers are useless. A voluntary, broker-reported dataset of nearly ten thousand transactions still contains real information about the shape of the market. It just cannot support the precision it is routinely quoted with.
And there is a survivorship problem nobody in this field can solve: every dataset here contains only businesses that sold. The ones that went to market and found no buyer are in none of them, and there is no register of those either.
Sources
- BizBuySell, 2025 Year in Review: 9,586 closed transactions, median sale price $350,000, median revenue $703,000, median cash flow $158,950, median cash flow multiple 2.61×. https://www.bizbuysell.com/blog/2025-year-in-review/
- GF Data (Association for Corporate Growth), Small-Deal Resilience, H1 2025: $1–5m enterprise value at approximately 5.5×, from 118 transactions in the $1–25m tier. https://gfdata.com/small-deal-resilience-h1-2025/
- Brookz Research, Overname Barometer H1-2025: average EBITDA multiples by normalised EBITDA for Dutch SMEs; 126 completed questionnaires from 289 advisory firms invited, a 44% response rate. https://www.accountancyvanmorgen.nl/wp-content/uploads/sites/2/2025/08/Brookz-Overname-Barometer-H1-2025.pdf
- Argos Wityu and Epsilon Research, Argos Index, Q1 2026: eurozone mid-market at 8.6× EBITDA overall, 10.0× for investment funds and 7.8× for strategic buyers. https://argos.fund/mid-market-argos-index-for-the-first-quarter-of-2026/
- BDO LLP, Private Company Price Index 2024: methodology, the Experian MarketIQ source, and BDO's own overstatement warning. https://www.bdo.co.uk/getmedia/5ace9586-7549-4705-99f1-fdedb1d700b1/BDO-Report-PCPI-2024.pdf
- Business Valuation Resources, DealStats FAQs: approximately 30% of transactions sourced from SEC or SEDAR filings as of July 2018, the remainder from the contributor network. https://www.bvresources.com/products/faqs/dealstats
- International Business Brokers Association and M&A Source, Market Pulse Q3 2025: 300 respondents reporting 247 completed transactions. https://www.prnewswire.com/news-releases/the-ibba-and-ma-source-announce-the-results-of-the-market-pulse-q3-2025-survey-302617915.html
- Eurostat, Structural business statistics metadata, and the OECD Structural and Demographic Business Statistics: both cover counts, turnover, employment and value added, not transaction prices. https://ec.europa.eu/eurostat/cache/metadata/en/sbs_esms.htm