The Bureau of Labor Statistics tracks every private-sector establishment born in the country and follows each one, year by year, until it closes. For the cohort that opened in California in the year ending March 2020, 53.7% were still operating five years later, in March 2025. Nationally, the same cohort's five-year survival rate was 51.4%. California does a couple of points better than the country as a whole, which is a genuinely good sign, and it still means close to half of everything that opened that year is gone. For a founder who just closed a round earmarked for one specific product, that curve is the actual clock the business is running against, not an abstract statistic.
What the BLS actually tracks
The Business Employment Dynamics (BED) program is a joint federal-state effort built on unemployment insurance tax records, which means it counts nearly every private-sector establishment with a payroll rather than a survey sample. Its establishment age and survival tables follow each year's opening cohort forward: an establishment counts as having "died" once its employment drops to zero in the state's records, regardless of why, a sale, a relocation, a shutdown, or an owner's decision to stop filing under that entity. The Bureau publishes these tables as plain-text data files on bls.gov/bdm, covering cohorts opening as far back as March 1994, both nationally and for individual states including California specifically. Every figure below is read directly from those tables.
California's five-year curve, cohort born in the year to March 2020
| Survival point | California | National |
|---|---|---|
| At opening (March 2020) | 100.0% | 100.0% |
| Year 1 (March 2021) | 82.3% | 80.9% |
| Year 2 (March 2022) | 77.2% | 72.3% |
| Year 3 (March 2023) | 66.8% | 63.6% |
| Year 4 (March 2024) | 61.3% | 57.2% |
| Year 5 (March 2025) | 53.7% | 51.4% |
California's opening cohort for that year numbered 142,133 establishments. Five years later, 76,308 of them were still filing payroll. The gap to the national rate is real but small, roughly 2 to 3 points at every checkpoint after year one. The shape of the curve, not the state-versus-country gap, is the more useful thing to read off this table.
The steepest drop happens before the business turns one
Look at how much of each cohort's losses land in a single year. California lost 17.7 points of survival in year one alone, more than twice what it lost in year two (5.1 points) and more than it lost in any other single year in the five-year window. Nationally the pattern is the same, a 19.1-point year-one drop against single-digit losses in every year after. Whatever kills a new establishment, it is most likely to happen before the business's first birthday, not gradually worn down over a five-year stretch.
That's the opposite of the slow bleed a lot of funding conversations assume. A board deck built around "we have eighteen months of runway" implicitly treats risk as spread evenly across that window. The Bureau's own data says the risk is front-loaded hardest into roughly the first twelve of those months, the exact stretch in which a funded, non-technical founder is usually still turning the money into a product rather than running one.
What this means for a funded founder's actual timeline
None of this is destiny. It is a distribution, and plenty of the businesses that closed in year one closed for reasons no amount of faster shipping would have fixed, a bad market, a founder who moved on, a product nobody wanted regardless of how quickly it arrived. But for the subset of that first-year attrition that comes down to running out of time before there was anything live to test against real customers, the BLS curve says the clock is shorter and steeper than an eighteen-month runway number suggests on its own. A build that eats nine or ten months of that first year leaves only a couple of months of the steepest part of the curve to find out whether the product actually works, which is not much of a test.
A worked example
Picture a nine-person, seed-funded specialty foods company that closed its round in the spring, earmarked for an inventory and wholesale-ordering platform to replace the spreadsheets its two co-founders had been running the business on. The founders aren't technical and hadn't scoped a software build before. If the project takes a year to reach a usable version, the company spends its entire first-year survival window, the steepest part of the BLS curve, with no live product to show a single wholesale customer, running on the same spreadsheets the round was supposed to replace. If the same build is scoped tightly enough to reach a working version inside three or four months, the company gets most of that first year to actually operate the business on real software and find out whether the bet was right, which is the only part of this that was ever actually in the founders' control.
Where this data doesn't mean what it looks like it means
This table is not a verdict on any single company's product or team, and treating it that way would be dishonest. It counts every private-sector establishment, restaurants, retail, real estate, alongside funded tech builds, and "died" in this data includes a business that was acquired, quietly folded into a larger entity, or closed because its owner retired, not only outright failures. It also isn't a case for rushing a build past the point where it works. A fast build that ships something customers can't actually use burns the same runway a slow build does, just with worse evidence at the end of it. The lesson isn't "move faster than is responsible." It's that the window to find out whether a specific, funded idea works is shorter than an eighteen-month runway number implies, which is an argument for a tightly scoped first version over an ambitious one, not for skipping the scoping altogether.
If your round is earmarked for a build that hasn't shipped a usable version yet, what actually goes into scoping a project like that is a useful next read, and the same funded, pre-revenue population the SEC's own Form D filings describe is largely this table's first-year cohort. We scope every custom software build against what a round actually funds and the time it actually has to prove itself, with transparent pricing and an itemised estimate within 48 hours of a call, and we'll say plainly when the honest scope is smaller than what's been asked for.
Sources
- U.S. Bureau of Labor Statistics: Business Employment Dynamics, establishment age and survival, California (Table 7, cohorts opening March 1994 through March 2025; accessed September 2026)
- U.S. Bureau of Labor Statistics: Business Employment Dynamics home (national and state establishment age and survival tables, program methodology)
- U.S. Bureau of Labor Statistics: Quarterly Business Employment Dynamics news release