Why most companies die young

Every industrial town in Europe has a building like this one. Four storeys of red brick, windows out, birch saplings growing from the parapet, a name on the gable that nobody under sixty recognises. It employed a thousand people. It had a pension scheme, a works band and a cricket team. Now it has pigeons.

Whatever killed it, the odds were always against it. Firms don't tend to last, and the ones that do are strange enough to be worth studying separately.

But almost everything you've been told about how quickly they die is either wrong, badly sourced, or measuring something other than death.

The nine-out-of-ten figure is junk

You've seen it a hundred times. Ninety per cent of startups fail. Sometimes it's ninety-five. Occasionally someone adds a year — ninety per cent within five years, ninety per cent within three.

Chase any of these to a source and the trail goes cold almost immediately. They get cited to consultancies quoting news articles quoting other news articles, and eventually to nothing at all. There's no study underneath.

The closest thing to a real number in that neighbourhood comes from Shikhar Ghosh at Harvard Business School, whose analysis of venture-backed companies was reported in the Wall Street Journal in 2012 under the finding that roughly three in four failed to return investors' money. That's a genuine piece of work, but notice what it measures — venture-backed firms, which are a tiny and deeply unrepresentative slice of all businesses, and failure defined as not returning capital, which isn't the same as ceasing to exist. Plenty of companies in that bucket are still trading perfectly happily today.

So when you see the ninety per cent line, treat it as decoration. Somebody wanted a dramatic opening and reached for a number that sounded right.

What the real survival data says

Actual figures exist and they're much less exciting, which is presumably why nobody repeats them.

The United States Bureau of Labor Statistics has tracked business establishments through its Business Employment Dynamics programme for decades. The pattern is remarkably stable across industries and across economic cycles: roughly one in five new establishments doesn't survive its first year. About half are still going at five years. Around a third make it to ten. Beyond that the curve flattens out and keeps declining slowly.

The Office for National Statistics runs a comparable exercise in the UK, and the shape is similar — somewhere around two in five businesses still active five years after birth.

Half surviving five years isn't nine in ten dying. It isn't even close.

What's more interesting than the headline is the shape of the curve. The risk of dying isn't highest on day one. It climbs, peaks somewhere in the first few years, and then falls. Sociologists have a name for each half of that: Arthur Stinchcombe described the liability of newness in 1965, arguing that new organisations die more because they haven't yet built routines, reputation or stable relationships with customers and suppliers. Later researchers added the liability of adolescence — the observation that a new firm often has a stock of initial resources, goodwill and patient money that buys it a grace period, and the dangerous moment comes when that runs out and the business has to stand on what it actually earns.

That's why year three tends to be nastier than year one. Year one is still living off the launch.

Death is not one event

Here's where most of the confusion comes from. "The company no longer exists" covers at least six situations that have nothing in common.

• It ran out of money and was liquidated, with creditors unpaid. This is failure in the ordinary sense.
• It was bought. The name disappears, the staff and the products often don't, and the owners may have done extremely well out of it.
• It merged, and the combined entity carries a different name.
• It was wound up voluntarily, solvent, because the owner retired, got ill, emigrated or simply got bored.
• It filed for reorganisation. In American practice, Chapter 11 is a restructuring process that many firms walk out of still trading; Chapter 7 is liquidation. They get reported identically as "bankruptcy".
• It restructured into a new legal entity for tax or regulatory reasons and appears in the statistics as one death and one birth.

The voluntary category is much larger than people assume, because most businesses are very small. A joiner, a hairdresser, a two-person consultancy. When that person stops, the business is recorded as a closure, and it lands in the same column as a collapsed manufacturer with unpaid suppliers. Studies that ask owners why they closed consistently find that a substantial share of exits were solvent and deliberate.

Being acquired is the one that most distorts the popular story. A firm that got bought for a large sum is counted, in a lot of the "companies don't last" analysis, alongside one that went under.

The lifespan statistic that everyone misuses

You'll have met this one. The average lifespan of a large listed company has collapsed — sixty years in the 1950s, under twenty now, and shrinking.

The underlying analyses are real. What they measure is average tenure in a stock market index, and that's a different animal entirely.

Companies leave an index for many reasons. They get acquired. They merge. They spin off divisions. They shrink below the size threshold while remaining large, profitable and completely alive. They get removed because the index committee changed its rules about listings, share classes or domicile. Very few leave because they stopped existing.

There's a second problem underneath: the composition of the economy changed. An index dominated by heavy industry, where firms were capital-intensive and slow to build and slow to fail, is not comparable to one dominated by companies that can reach enormous scale in a decade. Comparing the tenure numbers across those eras is comparing two different populations and calling the difference a trend.

None of which makes the statistic useless. It does mean it can't support the sentence it's usually attached to, which is normally something about how everything is speeding up and only the bold survive.

The planes that came back

The single most useful idea for thinking about any of this comes from a statistician working on aircraft in the Second World War.

Abraham Wald was part of the Statistical Research Group at Columbia, and the military brought him a problem. Bombers were returning from raids with bullet holes, and the holes clustered in certain areas. The obvious move was to add armour where the damage was concentrated.

Wald pointed out that they were only looking at the planes that came back.

The aircraft hit in the places that showed no holes in the surviving sample weren't undamaged there. They were in the sea. The armour belonged where the returning planes were unmarked, because damage there was what stopped a plane returning at all.

Business writing is almost entirely a study of the planes that came back. A book examines a dozen companies that grew enormous, identifies the practices they had in common, and presents those practices as causes. The companies that did exactly the same things and died aren't in the sample, because nobody writes books about them and their records were destroyed when the receivers cleared the office.

The results are checkable, and they're not flattering. Phil Rosenzweig's The Halo Effect, published in 2007, went back through the celebrated management studies of the previous decades and found that a large share of the exemplary companies subsequently performed no better than average, and some performed considerably worse. He also identified the mechanism: when a company is doing well, observers describe its culture as focused and its leader as visionary. When the same company stumbles, the same culture gets described as rigid and the same leader as stubborn. The judgments follow the results rather than explaining them.

Profitable companies run out of money

This is the failure mode that surprises people who've never run anything, and it's one of the most common.

Profit and cash aren't the same thing and they don't arrive at the same time. You buy materials in March, pay wages in April, deliver in May, invoice at the end of May, and get paid in August if the customer is honest and October if they're not. Every one of those transactions is profitable on paper. All of them consume cash months before they produce any.

Which produces the genuinely counter-intuitive result: growth kills companies. A firm whose orders double has to fund double the materials, double the wages and double the receivables, all before a penny of the new revenue lands. It's the same trap as a household taking on a bigger mortgage on the promise of a pay rise that hasn't cleared yet, except the numbers move faster and the bank is less patient.

Undercapitalisation shows up near the top of every serious analysis of small business failure, and it usually isn't a bad idea or a bad market. It's a good business with a mismatch between when money leaves and when money arrives.

The founder problem

The skills that start a company and the skills that run one at scale overlap far less than the mythology suggests.

Starting something requires conviction bordering on the unreasonable, a tolerance for chaos, and a willingness to do every job badly rather than wait for someone qualified. Running a business of two hundred people requires delegation, process, hiring discipline and the patience to sit through meetings about holiday policy. There's no reason to expect the same person to be good at both, and the evidence says most aren't.

Noam Wasserman's research, gathered in The Founder's Dilemmas in 2012, tracked thousands of startups and found that founders were replaced as chief executive far more often than the folk story allows — a majority had lost the top job well before the company matured, and a large share of those departures weren't voluntary. Notably, the replacements often happened after success rather than after failure, because success is what creates the scale the founder isn't built for.

Family succession runs into a related wall. A firm passing to a second generation is handing control to someone selected by birth rather than by aptitude, and the survival statistics for family businesses across generations are correspondingly grim.

The market moves and you don't notice

Clayton Christensen's The Innovator's Dilemma, published in 1997, argued that established firms often fail not through complacency but through competence. They listen to their best customers, who want improvements to the existing product. They invest where the margins are. A cheaper, worse alternative appears at the bottom of the market, gets ignored because serving it would be irrational, improves steadily, and eventually becomes good enough for everybody.

The theory has taken serious criticism — Jill Lepore's 2014 essay in the New Yorker went at the case studies hard, pointing out that several of the disrupted firms in the original book did fine afterwards, and several of the disruptors didn't. Treat it as a useful pattern rather than a law.

The example everyone reaches for deserves correcting while we're here. Kodak is presented as a company that ignored digital photography. It didn't. One of its own engineers, Steven Sasson, built a working digital camera prototype in 1975, and the company spent enormous sums on digital imaging over the following decades and held valuable patents in the field. Its problem wasn't blindness. It was that its profits came from film and chemicals, and no amount of foresight turns a chemicals business into a consumer electronics business without destroying most of what you already have. It filed for Chapter 11 in 2012, restructured, and still exists.

Nokia gets the same treatment and it's the same error. The handset business was sold; the company carried on as a network infrastructure firm and is still trading.

The ones that don't die

At the far end of the distribution, some businesses last absurdly long, and they're not what you'd predict.

Japan has more century-old firms than anywhere else, thousands of them, mostly small — inns, sweet makers, sake brewers, builders. The oldest documented case, Kongō Gumi, was a temple construction firm founded in 578 that operated for roughly fourteen hundred years under the same family. It didn't fail because it was overtaken by technology. It took on property debt during Japan's asset bubble, ran into trouble in the 2000s, and in 2006 was absorbed as a subsidiary of a larger construction group. The work continues under new ownership.

Which is the point of this whole business, really. Even the oldest company on earth ended not in collapse but in an acquisition, and if you were building a chart of corporate mortality it would appear as a death.

The long survivors share a few unglamorous traits. They're usually in businesses whose demand doesn't disappear — food, drink, shelter, ritual, repair. They tend to stay small deliberately and avoid debt. They pass on craft knowledge through apprenticeship rather than documentation. And several of them practise adoption of successors from outside the bloodline when the natural heir isn't suitable, which quietly solves the succession problem that kills most family firms.

None of this scales, and none of it would survive contact with a growth target. That's rather the trade being made.