Two days after Christmas in 2024, more than 35,000 U.S. small businesses that relied on Bench for their bookkeeping found a notice where their books used to be. Bench, a Vancouver accounting startup that had raised about $113 million from investors including Shopify and Bain Capital Ventures, said its platform was no longer accessible. Customers were given days to download their records, weeks before tax season. Staff were let go without notice. If you want to understand why startups fail, it is a useful place to start.
Bench was rescued within days by a buyer, which makes it an unusual ending. But the shape of the story is common. A company that looked healthy from the outside was, on the inside, far closer to the edge than anyone realized.
So why do startups fail? The popular answer, that most of them die in year one, turns out to be wrong. The real data is more interesting, and more useful if you are running a company or thinking about joining one.
Why startups fail, according to the post-mortems
The research firm CB Insights has tracked startup shutdowns for years. Its latest analysis, published in March 2026, looked at 431 venture-backed companies that shut down since 2023 and found identifiable reasons for 385 of them. The leading causes:
- Ran out of capital: 70%
- Poor product-market fit: 43%
- Bad timing or macro conditions: 29%
- Unsustainable unit economics: 19%
The numbers add up to more than 100% because most failed companies cited several reasons. And CB Insights makes an important point about the top one: running out of money is almost always the final cause of death, not the underlying disease. Money runs out because customers didn’t want the product enough, because the timing was off, or because each sale lost money. Those are the root problems.
The list of recent casualties is a reminder that funding is no shield. CB Insights’ sample includes digital freight broker Convoy and healthcare automation company Olive, each of which had raised roughly $1 billion.
A caveat on the data
Post-mortems are written by the people who were there, often in a hurry and sometimes with a reputation to protect. “Bad timing” can be a polite way of saying “we built the wrong thing.” Treat the categories as signals, not verdicts.
The survival numbers: what Statistics Canada and the BLS actually show
You’ve probably heard that 90% of startups fail, or that half of all businesses close in their first year. Official data tells a different story, at least for businesses in general.
In Canada, Innovation, Science and Economic Development Canada publishes survival rates using Statistics Canada microdata. Its 2025 edition of Key Small Business Statistics, covering businesses with 1 to 99 employees created between 2001 and 2022, reports these average survival rates:
| Years after creation | Share still operating (Canada) |
|---|---|
| 1 year | 94.8% |
| 2 years | 86.9% |
| 5 years | 68.0% |
| 10 years | 48.2% |
Goods-producing businesses tend to survive longer than service businesses, a gap that widens over time.
The U.S. Bureau of Labor Statistics measures something slightly different: private-sector establishments, meaning individual business locations, tracked from the year they open. Its survival table shows that of the establishments that opened in March 2015, 79.6% were still operating a year later, 50.2% after five years, and 34.7% after ten. The cohort that opened in March 2020, right as the pandemic hit, did better than many expected: 80.9% after one year and 51.4% after five.
Why does Canada look better than the U.S.? Partly definitions. The Canadian figures count businesses with at least one paid employee, which excludes many very short-lived ventures, while the BLS counts establishments. Don’t compare the two directly. The takeaway from both is the same: roughly half of new businesses are gone within about five to ten years, and the danger doesn’t end after year one.
Here’s the catch for founders: venture-backed startups aren’t typical businesses. A dental office or a landscaping company aims to be profitable and stay small. A VC-backed startup is built to grow very fast or die trying, and investors expect most of their portfolio to fail or return little. That’s why the CB Insights data is more relevant to tech founders than the national survival rates, even though the national figures are more rigorous.
Case study: Quibi and the limits of money
Few startups have ever launched with more resources than Quibi. Founded by Hollywood veteran Jeffrey Katzenberg and led by former eBay and HP chief Meg Whitman, the mobile-only streaming service raised $1.75 billion, according to Crunchbase News, from investors that included major studios. It launched on April 6, 2020, and announced its shutdown on October 21 the same year.
The pandemic is the usual explanation, and the founders pointed to it. Quibi was designed for short “quick bites” of video on the go, and suddenly nobody was going anywhere. But timing alone doesn’t explain it. The service asked people to pay for short-form video when TikTok and YouTube offered endless amounts for free. Projections had called for far more subscribers than it ever reached. Quibi’s library was sold to Roku in early 2021.
Quibi is a near-perfect example of the CB Insights pattern. It didn’t really run out of money; it ran out of reasons to keep spending it. The product-market fit wasn’t there, and no budget could buy it.
Case study: Theranos and the failure that was fraud
Theranos sits in a different category. The blood-testing company claimed it could run a wide range of tests on a few drops of blood. At its peak it was valued at about $9 billion. Then, in October 2015, The Wall Street Journal reported that the company was running many tests on conventional machines rather than its own devices, and that its technology wasn’t performing as claimed. Theranos dissolved in 2018.
Founder Elizabeth Holmes was convicted of defrauding investors and in November 2022 was sentenced to more than 11 years in prison. Her former business partner, Ramesh “Sunny” Balwani, received a sentence of nearly 13 years.
Most failed startups have nothing in common with Theranos. But it carries a lesson for honest founders too: when a company’s public story drifts away from what the product actually does, the gap tends to widen, not close. Investors, boards and employees who don’t ask hard questions make that drift easier.
Case study: Bench, a Canadian cautionary tale
Bench was a Vancouver success story for most of a decade. It started in 2012, moved to Vancouver the following year, and raised a $60 million Series C in 2021. It sold bookkeeping and tax services to small businesses through a mix of software and human accountants.
The shutdown was abrupt. Its bankruptcy filing in Canada in January 2025 later showed about $65.4 million in liabilities against roughly $2.8 million in cash, according to the company’s publicly reported bankruptcy details. Employer.com acquired the business days after it went dark, and service resumed in early 2025. Founder Ian Crosby, who had earlier been replaced as CEO, publicly blamed the board’s decision to bring in new leadership. That’s one view; the full story of why the company couldn’t raise again or sell earlier hasn’t been told.
Whatever the internal causes, Bench shows how a services-heavy business with real revenue can still fail if its costs outrun what it earns and new funding dries up. It also shows the human cost of a sudden shutdown: customers lost access to financial records, and employees were let go without notice.
Warning signs you can spot early
Founders usually see trouble before anyone else does, but they’re also the most motivated to explain it away. Here are signals worth taking seriously, whether you’re the founder, an early employee, or an investor:
- Runway under 12 months with no clear plan. Fundraising takes months. If you’ll need money before you can show meaningfully better numbers, you’re negotiating from weakness.
- Growth that depends on paid acquisition. If customers stop arriving the moment you pause ads, you don’t have product-market fit yet; you have a marketing budget.
- Every new customer loses money. CB Insights flags unsustainable unit economics in about one in five failures. Scale makes bad unit economics worse, not better.
- High churn hidden behind new sign-ups. Track how many customers are still active after three, six and twelve months. Retention is the most honest product-market fit metric there is.
- Repeated pivots without new evidence. Pivoting is healthy. Pivoting every quarter because the last idea didn’t raise money is a warning.
- Leadership turnover and board conflict. Executive churn and visible tension between founders and investors often precede bad outcomes, as Bench’s story suggests.
- Metrics that keep getting redefined. If the headline number changes every board meeting, ask what the old one was showing.
What to do with all this
Startup failure is normal, and the data is clear that most businesses won’t survive a decade. That’s not a reason to avoid starting one. It’s a reason to design for the most likely failure modes from the start.
- Treat runway as a product metric. Review it monthly. Decide in advance what you’ll cut, and when, if fundraising takes longer than planned.
- Prove demand before you scale spending. Quibi spent heavily before it knew whether people wanted what it sold. Small, cheap tests are underrated.
- Know your unit economics early. Calculate what it costs to acquire and serve a customer, and how long it takes to earn that back.
- Plan for a graceful wind-down. If the worst happens, a planned shutdown that gives customers time to export data and employees proper notice protects your reputation for the next venture. Bench is a lesson in what happens without one.
- Get outside eyes. An adviser or board member who will tell you the uncomfortable truth is worth more than one who cheers.
The startups that survive aren’t always the ones with the best ideas or the most money. More often, they’re the ones that noticed the warning signs early and acted before the bank balance made the decision for them.
Sources and further reading
- CB Insights: The top reasons startups fail (March 2026)
- ISED Canada: Key Small Business Statistics 2025
- U.S. Bureau of Labor Statistics: Survival of private sector establishments by opening year
- U.S. Bureau of Labor Statistics: Establishment age and survival data
- TechCrunch: Bench shuts down, leaving businesses without access to accounting and tax docs
- Wikipedia: Bench Accounting
- Crunchbase News: Quibi is shutting down after raising $1.75B
- Wikipedia: Quibi
- CNBC: Elizabeth Holmes sentenced to more than 11 years in prison
- Wikipedia: Theranos
