Half of all businesses die by year five. The survivors share a trait.

The Bureau of Labor Statistics has tracked business survival for decades, and the curve barely moves no matter what the economy is doing: about 20% of new businesses fail in their first year. Roughly half are gone within five years. By year ten, about 65% have closed. Those numbers have held remarkably steady since the 1990s, through booms and recessions alike — which tells you the cause is structural, not circumstantial.

The figures shift by industry, and the shifts are instructive. Agriculture has among the lowest failure rates — under 7% in year one — because the people who enter it tend to understand its economics intimately before they start. Mining and oil-and-gas extraction see roughly 60% gone by year five, because they depend on commodity prices no operator controls. The lesson buried in the variance is that survival correlates with how well the operator understands the machine they’re running, not with how exciting the sector looks from outside.

When I was advising companies — Ernst & Young, Capgemini, later Toyota’s digital practice — I sat inside enough businesses to see what the failure statistics don’t tell you. Companies almost never die of a bad idea. The idea is usually fine; sometimes it’s excellent. They die of operational causes: pricing that never quite covered the true loaded cost of delivery, hiring that lagged growth by a year so quality quietly eroded, systems held together by the founder’s personal heroics until the founder burned out, cash conversion cycles that nobody was actually watching until the account hit zero on a Friday.

These are not dramatic failures. That’s what makes them dangerous. A company doesn’t collapse in a single bad quarter — it accumulates small operational debts, each individually survivable, until one ordinary shock arrives and there’s no slack left to absorb it. The post-mortem always names the shock. The real cause was the missing slack, built up over years of decisions that each seemed reasonable at the time.

Then I stopped advising and started building, and I learned the same lesson again — this time with my own money on the line, which is a more honest teacher. At Dignity Living, we operate in care services, a sector where operational failure isn’t an abstraction on a slide. People depend on you daily; a missed detail is not a metric, it’s a person. The discipline that keeps a care home running turns out to be exactly the discipline that keeps any company alive: know your unit economics cold, build the system before you need it rather than after it breaks, and never let growth outrun your ability to actually deliver what you sold.

Companies rarely die of a bad idea. They die of operational causes.

This is why SwellPoint underwrites the operator as much as the opportunity. The five-year survival line isn’t fate handed down by the market — it’s a filter, and the variable it filters on is discipline. The companies that cross it tend to share one trait above all others: someone inside who treats operations as the strategy itself, not as the unglamorous chore you get to after the strategy is set. In the businesses that last, the operations are the strategy.

When we diligence a company, that’s what we’re looking for underneath the financials. Not just whether the numbers are good today, but whether the person running the company understands why they’re good — and whether they’d notice, early, if they started to slip. A founder who can explain their own cash cycle from memory is telling you something a spreadsheet can’t.

And when we invest, that’s what we help build. Not a logo on a deck and a quarterly check-in, but real work on the parts of the business that determine survival: pricing that holds, hiring that stays ahead of demand, systems that don’t depend on any one person being awake. We can help with this because we’ve done it — missed the Friday payroll deadline by hours, shipped anyway, and built the discipline that made sure it never happened twice. The survival curve is steep. It is also, more than the data alone suggests, a choice.