Survival is equal. Revenue is not. That rules a lot out.

Probably not because you run it worse. In 138,000 US firms founded in 2013, women-owned businesses survived at the same rate as men-owned ones — and earned a median of about $50,000 in year one against about $75,000. Equal survival with unequal revenue points away from execution and towards what is sold, to whom, at what price.

Those two findings are almost always reported apart — survival in stories about resilience, revenue in stories about ambition. Put them in one sentence and they become a single, much more precise fact: whatever produces the deficit, it is not the ability to run a company, because the thing that tests that came out level.

Two findings that are usually printed on different pages

The JPMorgan Chase Institute tracked 138,000 firms founded in 2013 through administrative banking data — not a survey, but the actual money moving through the accounts. Women-owned businesses had the same survival rates as men-owned businesses. Their revenues did not converge.

Median revenue of firms founded in 2013, by owner. JPMorgan Chase Institute, February 2019, 138,000 firms. Dollar figures rounded; percentages calculated on the unrounded data.

Year	Women-owned	Men-owned	Absolute gap
Year one	about $50,000	about $75,000	about $25,000 (34% lower)
Year four	about $68,000	about $104,000	about $36,000

Do the arithmetic on the last column, because it is the part that gets lost when the gap is quoted as a percentage. In relative terms the gap barely moves across four years. In money it goes from roughly $25,000 to roughly $36,000, an increase of about 44%. Four years of trading did not narrow it. Four years of trading compounded it.

The same pairing shows up in Britain at national scale. The Alison Rose Review of Female Entrepreneurship, published by HM Treasury in March 2019 on roughly 5,000 survey respondents, 200 or more interviews and GEM data from 2013 to 2018, found survival equal: 73% of both women-led and men-owned businesses still going beyond three and a half years. It also found men five times more likely to be running a business with turnover above £1m — 2.4% of working-age men against 0.5% of working-age women — and average gross value added of £56,000 a year for women-led businesses against £126,000 for men-led ones.

The difference is in the shape of the distribution, not the average

Averages hide this. The distributions do not. In the Rose Review, 13% of women entrepreneurs reach £1–50m turnover against 29% of men, and 81% of women-led businesses employ five or fewer people against 73% of men-owned ones.

A Frontier Economics survey across Denmark, Finland, France, Germany, Italy and Poland found the same shape: 42% of women-founded businesses report annual revenue under €100,000 against 22% of men-founded ones, 17% report over €1m against 35%, 31% are sole operators with no employees against 16%, and 29% have more than 20 employees against 45%. Two things about it need stating rather than footnoting: 608 founders and senior leaders, roughly 100 per country, so it is a survey and not a census — and it was commissioned by Amazon, which has a commercial interest in the population it describes.

What both datasets describe is not a set of firms performing slightly worse. It is a set of firms clustered at a smaller operating scale. Scale is not an outcome of effort. It is an outcome of capital, channel and price.

Started with roughly half the capital

The Rose Review found that women launch businesses with 53% less average start-up capital than men, and that women entrepreneurs estimate needing 40% less funding than men to launch in the first place.

53% less average start-up capital at launch, according to the Rose Review's analysis of pre-accelerator applications, April 2018.

Read those two together and they describe a plan, not a shortfall. A business capitalised at half the level is not the same business with less money in it. It is a different business: a smaller first order, a narrower launch, no salaried second person for two years, a channel chosen because it is cheap rather than because it is where the buyers are. Each of those choices is rational at the capital available, and each sets a ceiling that shows up as revenue three years later.

KfW's Gründungsmonitor 2026, with around 50,000 respondents in Germany, adds the follow-through: 40% of women realised their start-up capital as planned against 50% of men, and 37% invested more than planned against 26%. The plan was smaller, and the money was harder to assemble on schedule.

And more expensive credit afterwards

Alesina, Lotti and Mistrulli examined 1.2 million overdraft loans to approximately 150,000 Italian microfirms and self-employed individuals over 12 quarters from January 2004 to December 2006, using the Central Credit Register and the Bank of Italy's Loan Interest Rate Survey. Women paid about 29 basis points more in the basic specification, and about 20 basis points more after controlling for business characteristics and credit history.

The reason this study survives scrutiny is what it does next. The obvious explanation for a price difference in credit is risk. The risk went the other way: women-owned businesses in the data had lower bankruptcy rates, 1.9% against 2.2%, and slightly better credit histories. Cheaper borrowers, dearer credit.

The caveat is significant and I would rather state it than have it found. This is Italian data from 2004 to 2006. It is old, and it is one country. It also remains the largest and methodologically cleanest evidence on a gendered price of credit that exists, which tells you something uncomfortable about how little anyone has looked since.

You were asked how you would avoid losing

Kanze, Huang, Conley and Higgins coded video transcripts of investor question-and-answer sessions with 189 startups pitching at TechCrunch Disrupt New York between 2010 and 2016. Men founders were asked promotion-framed questions: customer acquisition, market size, revenue upside — how will you win. Women founders were asked prevention-framed questions: retention, break-even, defensibility, risk — how will you avoid losing.

Entrepreneurs raised $3.8 million less for every prevention-framed question they were asked.

$3.8m less raised for every prevention-framed question an entrepreneur was asked, across 189 startups pitching between 2010 and 2016.

In follow-up experiments with accredited angels and seed investors allocating a simulated $400,000, answering a prevention-framed question with a promotion frame increased the funding allocated. The question was not the whole problem. The answer to it was.

The founders who were asked how they would avoid losing answered the question they were asked. That is what it cost them.

Venture capital is a tiny channel. The asymmetry it exposed is not, and the transfer is the part almost nobody makes. The same shape appears in a Swiss bank relationship manager's credit questions — what happens if your largest client leaves, how do you cover the repayment in a bad quarter. It appears in a retail buyer's category review — what is your returns rate, what happens if this does not sell through. It appears in a distributor negotiation — what are your minimum volumes, what is your exposure. Each of those is a prevention-framed question asked in good faith by someone who has no idea they are asking it more often of one person than another.

The mechanism is worth understanding rather than scripting. A prevention question restricts the frame of the answer to downside. Answer it completely and accurately and you have spent the entire exchange discussing your own risk, in a meeting whose purpose was to establish upside. The counterparty leaves with a full picture of what could go wrong and no picture at all of what could go right, and they will price accordingly — not out of prejudice, but because that is the only information they were given. The correction is not confidence and not deflection. It is answering the question that was asked and then supplying the frame that was not: the risk, quantified and handled, followed by the growth case the meeting never got to.

Full-time founding is falling, and it is a capacity structure

KfW's 2026 figures contain a shift worth naming. Women are 35% of all German founders, down slightly from 36% in 2024. But their share of full-time founding — Vollerwerb — fell from 33% to 27% in a single year, while their share of side-business founding held at 38%.

That is not a change in ambition over twelve months. Ambition does not move six points in a year. Capacity does.

Set it against the Swiss labour market. In BFS SAKE in Kürze 2024, 58.7% of employed women work under a 90% workload against 20.5% of men, and women are 71.8% of all part-time workers. Meanwhile the Swiss Confederation's SME portal, citing the Federal Statistical Office's Labour Force Survey for the second quarter of 2024, reports that women are 41% of all self-employed people in Switzerland, up from 28.2% in 1991, with 7.2% of women entrepreneurs against 10.5% of men. Business demography statistics for 2022 record 17,724 new enterprises founded by women alone — 37.7% of the total.

More firms, founded with less capital, disproportionately alongside something else. That combination produces exactly the revenue distribution the JPMorgan and Rose data describe, without anyone being less capable.

Two numbers that would help this argument, which I will not use

The first is the claim that women reinvest 90% of their income in their families against 30 to 40% for men. It is quoted constantly. Every citation trail ends at an advocacy page citing another advocacy page. No study, no sample, no country, no year. I would use it if it existed.

The second is McKinsey's Diversity Matters, Diversity Delivers and Diversity Wins finding that companies with more women in leadership are more likely to outperform financially. Green and Hand attempted a quasi-replication in Econ Journal Watch in March 2024, using S&P 500 firms as of 31 December 2019 — 497 firms and 7,246 executives, with performance measured from 2015 to 2019 — and found no statistically significant relationship, with 37 of 40 test statistics insignificant.

Be precise about what that shows. Green and Hand tested racial and ethnic executive diversity, not gender specifically. The correct conclusion is that McKinsey's studies are correlational, built on non-public data, and have not replicated. It is not that diversity does not work. A statistic that cannot be checked is not evidence, whichever direction it points, and a case built on one collapses the moment somebody checks.

What this means in practice

If survival is equal and revenue is not, the diagnostic question is not "what am I doing wrong" — the data has already answered that — but "which of the four things that set scale is binding here".

Capital is the first, and it is usually historic rather than current. A business launched on half the capital carries that launch for years, as a channel it could not afford, a hire it could not make, a product range narrower than the market it sells into. The constraint may have lifted long ago without anyone revisiting the decisions it caused.

Price is the second, and it is the one most visible in the accounts of the firms this describes: realised price by client and by project, and how far it sits below the rate card once discounts, unbilled hours and scope creep are counted. I have set out the evidence on that gap separately.

Channel is the third, and it is where the sole-operator clustering in the European survey actually lives. A business that sells only through referral has a growth rate capped by the size of the network, regardless of how good the work is.

The fourth is the one the Kanze study exposes, and it is not a soft factor. Every credit line, every listing, every distribution agreement is decided in a conversation. If those conversations are systematically framed around your downside, the terms you get will be systematically worse, and the record will show it as a business that did not grow rather than as a business that was consistently priced for risk it did not carry. That is worth auditing in your own last five funding or supplier conversations — not for confidence, for framing. What were you asked, and what did you spend the meeting talking about? The evidence on negotiation points the same way: what changes the outcome is information and structure, not disposition.

What this cannot tell you from the outside

The honest position on all of the above is that it locates the question and cannot answer it for one company. JPMorgan is US banking data on firms founded in a single year. The Rose Review is British, and I have deliberately used only its structural figures, not its self-reported attitude findings, because attitudes measured by survey are the weakest thing in it. Alesina is Italian and two decades old. Kanze is 189 startups in one venture market, and the transfer I have made to bank and buyer conversations is an argument by analogy, not a replication — I think it is right, and it has not been tested. Frontier Economics is roughly 100 respondents per country, commissioned by a company with an interest in the answer. KfW and BFS are national aggregates that know nothing about any individual firm.

What would settle it for one business is four weeks in its own records. Contribution margin by product and by customer, so the profitable core is visible rather than assumed. Revenue split by channel with the acquisition cost and the realised price of each. Five years of growth decomposed into volume and price. The financing history, including the terms offered and the terms taken. And a straight read on which of those four constraints is actually binding, because they are not all binding at once and treating the wrong one is how three years disappear.

Those facts exist already, in the ledger, the order history, the invoices and the loan file. Nobody has put them together. The Treuhänder reports the past accurately, which is a different task. The bank models the ability to repay, which is a different task again. The tax adviser minimises. None of them has been engaged to explain why a business that survives as well as its competitors sells a third less than they do, and none of them has been asked.

It is a question with an answer. It is simply nobody's job.