Ask a founder what share of tech startups survive five years and you will usually get a guess pulled from a conference slide, somewhere between "10%" and "one in three," rarely with a source attached. The real number, from the Office for National Statistics' Business Demography series, is more precise and less dramatic than either: 37.0% of the UK "information and communication" companies born in 2019 were still active five years later1. The all-industry rate for the same birth cohort was 38.4%. Tech is not the high-failure outlier its reputation suggests. It is close to the average, sitting very slightly below it.
That headline number hides the more useful part of the story for a founder actually planning a company: how the tech survival curve compares to the average at each stage, and why a sector can post strong early survival and a high churn rate at the same time. Both show up in the same dataset.
The 2019 cohort: the only complete 5-year read
ONS Business Demography tracks each year's new company "births" forward through successive anniversaries. Only the 2019 birth cohort has a full five-year survival window in the current edition; every cohort born since 2020 has fewer elapsed years, so their longer-horizon figures are not yet available1. For the 2019 cohort, 30,080 information-and-communication enterprises were born. Here is what happened to them at each anniversary, compared to the 363,825 all-industry births from the same year12.
| Years after birth | Tech (info & comms) survival | All-industry survival | Gap |
|---|---|---|---|
| 1 year | 95.9% | 94.6% | Tech +1.3pp |
| 2 years | 76.7% | 74.7% | Tech +2.0pp |
| 3 years | 56.0% | 55.9% | Tech +0.1pp |
| 4 years | 43.8% | 45.0% | Tech -1.2pp |
| 5 years | 37.0% | 38.4% | Tech -1.4pp |
The pattern is a crossover, not a straight-line disadvantage. In years one and two, the 2019 tech cohort outperformed the all-industry average. By year three the two curves are within a tenth of a point of each other. From year four onward, tech has slipped below the all-industry line and the gap widens slightly by year five. A founder reading only the headline five-year figure would conclude tech is a marginally worse bet than the average business. A founder reading only the one-year figure would conclude the opposite. Both readings are correct for their own horizon; neither is the whole picture.
The newest cohorts: early survival is holding up
The 2019 cohort is the only one with a full five-year reading, but ONS also tracks four more recent birth years through whatever anniversaries have elapsed so far1.
| Birth cohort | Tech births | 1-year survival | 2-year survival | 3-year survival | 4-year survival |
|---|---|---|---|---|---|
| 2019 | 30,080 | 95.9% | 76.7% | 56.0% | 43.8% |
| 2020 | 22,455 | 94.2% | 74.2% | 55.6% | 46.4% |
| 2021 | 22,350 | 95.0% | 75.3% | 60.2% | n/a (not yet elapsed) |
| 2022 | 22,455 | 94.7% | 76.3% | n/a (not yet elapsed) | n/a (not yet elapsed) |
| 2023 (latest) | 22,380 | 94.9% | n/a (not yet elapsed) | n/a (not yet elapsed) | n/a (not yet elapsed) |
The most recent fully tracked cohort, born in 2023, held 94.9% one-year survival, consistent with every other cohort back to 20191. One-year survival for the sector has not deteriorated. What is missing, deliberately, is any claim about how the 2020 to 2023 cohorts will look at five years: those windows have not elapsed yet, and ONS itself does not publish five-year figures for cohorts that have not reached that anniversary. Treat any five-year projection for a post-2019 birth year as a guess, not data.
The churn-and-growth paradox, and why it is not a contradiction
Founders and investors often describe tech as both a high-churn sector and a high-growth one, which sounds like it should be a contradiction. It is not, because the two claims describe different parts of the same distribution rather than the same companies.
The ONS sector comparison for the 2022 birth cohort's first two years shows information and communication sitting in the upper-middle of the survival table, behind health, finance and education, but ahead of professional services, property, construction, retail and several other broad sectors1.
| Sector | 2022 births | 1-year survival | 2-year survival |
|---|---|---|---|
| Health | 10,890 | 96.6% | 83.3% |
| Finance and insurance | 2,775 | 93.2% | 79.5% |
| Education | 5,000 | 93.7% | 77.6% |
| Professional, scientific and technical | 47,445 | 94.6% | 77.0% |
| Information and communication (tech) | 22,455 | 94.7% | 76.3% |
| Property | 11,825 | 94.5% | 73.9% |
| Construction | 49,390 | 93.5% | 69.0% |
| Retail | 29,420 | 89.6% | 64.1% |
That places tech comfortably mid-pack on raw survival, not at either extreme. The "high churn" reputation comes from a different angle: within tech, the distribution of outcomes is unusually wide. A large share of new tech companies are single-product bets aimed at product-market fit, and when the product does not land, the company folds or pivots quickly rather than limping on as a stable small business, which is common in sectors like retail or professional services. At the same time, the minority of tech companies that do find traction tend to scale revenue and headcount far faster than the average small business in any other sector, because software has near-zero marginal distribution cost. The result is a sector where the bottom of the distribution clears out quickly and the top compounds quickly, both of which are consistent with a survival rate that lands close to the all-industry average rather than dramatically above or below it. High variance, not uniformly high or low survival, is the actual pattern the ONS data supports.
What this means for founder planning
None of these figures are a verdict on any specific company, and they should not be used as one. A cohort survival rate describes what happened to tens of thousands of companies grouped only by the year they were incorporated. It says nothing about your product, your team, your funding position or your specific market. What it is useful for is calibrating expectations and building a more honest financial plan.
Runway planning should assume attrition is normal, not a personal failure signal. If roughly six in ten tech companies from a given birth year do not make it past three years, then a founder whose company is still trading at that point is already outperforming the median outcome for the sector, even if growth has been slower than hoped. Runway and cash discipline matter more than sector-wide survival odds; see our guide on when a funded startup actually needs CFO-level financial oversight.
Reliefs and structuring decisions should be timed to your own trajectory, not a sector average. The SEIS and EIS schemes exist precisely because early-stage company failure is common and priced into how investors are compensated; understanding your own qualifying position matters more than a national failure rate. See our SEIS vs EIS comparison and our SEIS/EIS advance assurance service.
R&D-intensive companies that survive the early cliff often become the ones scaling fastest. The merged R&D scheme and the enhanced R&D-intensive support (ERIS) route exist to support exactly the loss-making, R&D-heavy companies most exposed to the early-years attrition shown in this data. See our ERIS: the R&D-intensive relief explained.
Grant and equity funding stack differently at different survival stages. A company past its third anniversary is in a different funding conversation than one at month six. Our UK startup grants landscape guide covers how grant funding sits alongside equity through this timeline.
A note on what this post is (and is not)
This post uses ONS Business Demography cohort data: fixed groups of companies born in the same calendar year, tracked forward through each anniversary. It is a different measurement from our separate Startup Formation & Survival Index, which uses a live Companies House snapshot of active-versus-dissolved company counts across eight software and IT SIC codes, updated quarterly. A snapshot tells you today's stock of active and dissolved companies across all birth years combined; a cohort curve tells you what happened, anniversary by anniversary, to one specific birth-year group. The two are not directly comparable and should not be quoted interchangeably. The full cohort tables, methodology and caveats behind this post are published on our Tech Startup Survival Curves research page, sourced from ONS Business Demography and released under the Open Government Licence v3.0.
If you are building a funded startup and want your financial plan, reliefs and structuring built around your company's actual position rather than a sector average, our funded startups hub and core compliance service are the right starting points.
Sources
- Founder Tax Partners: UK Tech Startup Survival Curves, derived from ONS Business Demography, UK: 2024 (Table 4.2, survival of newly born enterprises by broad industry group). Published under the Open Government Licence v3.0. Data pulled 2026-07-23.
- Office for National Statistics: Business Demography, UK: 2024. Open Government Licence v3.0.