SEIS Success Rates: The Base-Rate Arithmetic Nobody Shows
The honest base rates, what the reliefs change, and the portfolio maths that follows.
In this article
What is the real SEIS success rate? Strip away the brochures and the base rates are stark: across UK early-stage investing generally, roughly half of the companies backed at seed return less than the capital invested, most of the rest return modest multiples, and something in the region of one in twenty drives nearly all the gains. SEIS does not repeal that arithmetic; it reprices it. This analysis lays out the honest base rates, what the reliefs change, and the portfolio maths that follows.
The SEIS success rate base numbers, stated plainly
Long-run studies of angel and seed investing in the UK and comparable markets keep finding the same shape: a majority of individual holdings lose money in nominal terms, with total write-offs a large minority of all positions; a middle band returns between one and five times over five to ten years; and a thin tail, a few percent of companies, returns ten times or more and accounts for most of the aggregate profit. Formal survival data points the same way: ONS business demography has around 40% of new UK businesses surviving five years, and venture-backed cohorts, though better selected, still see failure as the modal single-company outcome.
Two honest caveats. SEIS-specific outcome data is thin: HMRC publishes fundraising statistics, not portfolio returns, so the base rates borrow from the wider seed asset class. And selection matters: companies that clear advance assurance and attract organised angels are not a random sample of startups. Better than average is plausible; immune to the distribution is not.
Why averages mislead: the power law
Seed returns are not distributed like a bell curve; they are distributed like a lottery with better odds. The mean is dragged up by rare large wins the median investor never holds. Consequences: a portfolio’s result depends almost entirely on whether it contains a tail company; adding holdings raises the probability of catching the tail rather than smoothing toward a comfortable average; and any single-company “expected return” quote is close to meaningless. This is why every serious treatment of SEIS, including our fund performance analysis, judges portfolios by shape rather than average, and why a 60% write-off rate can coexist with excellent performance.
What the reliefs actually change
SEIS moves the arithmetic in three places, all mechanical. The entry price: 50% income tax relief means a pound of exposure costs 50p for an investor with the liability to absorb it. The failure case: loss relief on the net cost at the marginal rate takes the true worst case to 27.5p per pound for a 45% taxpayer (30p at 40%, 40p at 20%), arithmetic you can run for your own rates in the calculator.
The success case: no capital gains tax on qualifying disposals after three years, so tail wins are kept whole. Compress the distribution’s downside, leave its upside untaxed, and mediocre base rates can become a defensible expected value. That sentence, though, carries all the conditions: sufficient tax capacity, three-year qualifying status held, and the diversification to catch a tail.
The portfolio maths, worked
Take a deliberately unheroic 45%-taxpayer portfolio: ten holdings of £5,000. Suppose six fail completely, three return exactly their capital over the period, and one returns eight times.
Gross outcome: £15,000 from the middle band plus £40,000 from the winner equals £55,000 on £50,000 subscribed, a 1.1x that would be a disappointing decade unassisted. Now the SEIS layer: £25,000 of initial relief on the way in, £6,750 of loss relief on the six failures (45% of their £15,000 net cost, since £30,000 of failed subscriptions carried £15,000 of relief), and the £35,000 gain on the winner untaxed. Net cash position: £55,000 returned plus £31,750 in reliefs against £50,000 out, comfortably positive despite six failures in ten. Remove the winner and the reliefs soften the loss but do not rescue the portfolio. The lesson runs both ways.
What improves the odds, and what only feels like it does
Evidence-supported: diversification across enough holdings (the tail-catching argument suggests ten as a floor, more if accessible), diversification across sectors and vintages given the concentration in the scheme statistics, disciplined due diligence per our checklist, and following organised syndicates with real screening. Feels-like-it: backing only “sure things” (the sample does not contain any), concentrating into the one company you know well (idiosyncratic risk, not information advantage), and treating the reliefs as the investment case (they modify outcomes; they do not create them, and the suitability analysis shows what happens when that logic leads).
The honest close
The truthful SEIS pitch reads: most of your companies will probably fail; the reliefs mean each failure costs a fraction of its face value; your result will be decided by whether one or two holdings land in the tail; and diversification is the only lever that reliably improves your chance of holding one when it does. Investors comfortable with that sentence are the scheme’s natural constituency. Investors who need the sentence to be softer should read it again, because the data will not soften it for them.
Common questions
What percentage of SEIS companies fail?
No official SEIS-specific failure statistic exists; HMRC publishes fundraising data, not outcomes. The wider evidence base for UK seed investing puts total losses at a large minority of holdings and sub-capital returns at roughly half, with five-year survival for new businesses generally around 40% per ONS data.
Does SEIS make early-stage investing profitable?
It reprices it. Initial relief halves the entry cost, loss relief caps the worst case at 27.5p to 40p per pound depending on tax rate, and gains are CGT-free after three years. Whether a portfolio profits still depends on catching a power-law winner, which is a diversification question.
How many SEIS investments should a portfolio hold?
The power-law logic argues for at least ten holdings, spread across sectors and years, so the portfolio has a realistic chance of containing a tail outcome. Fewer holdings make the result a coin toss on one or two companies, whatever the reliefs do to the downside.
Sources
ONS business demography (five-year survival); British Business Bank equity research; UKBAA and academic angel-returns studies (Nesta/Intelligent Capital “Siding with the Angels” lineage); HMRC venture capital scheme statistics. Base rates are asset-class evidence, not SEIS-specific promises; correct as at 17 July 2026 and consistent with our rules and limits reference. Education, not investment advice.
