What Percent of Businesses Fail — and What Our 38 Failures Show

We do not publish a failure percentage, because we do not have the sample to compute one honestly — what we have is 38 documented failures kept inside an index of 406 businesses, each with the specific thing that ended it. National survival percentages come from government data, and the U.S. Bureau of Labor Statistics Business Employment Dynamics survival data is where to read them.

What that percentage cannot tell you is the mechanism. This page is the mechanism: what was built, what it earned, and what killed it.

Table of contents

• ·What percentage of businesses actually fail?
• ·What the official failure rates measure — and what they miss
• ·What actually killed the ones we documented
• ·How much is each of these numbers actually worth?
• ·Why a list without failures is worse than useless
• ·How to use this without fooling yourself
• ·Frequently asked questions

What percentage of businesses actually fail?

Our answer is a count, not a percentage: 38 of the 406 businesses in our index launched, were documented, and did not work. Six of those 38 disclosed a clean monthly figure before they stopped, median $8K/mo, across a range of $29/mo to $400K/mo. Twenty-three were one-person operations.

We deliberately do not convert that to a rate. Our index is built from businesses people chose to talk about, so any percentage derived from it would be wrong in a predictable direction — too optimistic. For the population-level number, use the BLS series or the SBA Office of Advocacy small business FAQ.

What we add is the part those datasets structurally cannot contain: the cause, named, with the revenue attached and the evidence class stated. That is the whole reason we keep failures instead of quietly deleting them.

What the official failure rates measure — and what they miss

Government survival statistics measure whether an establishment still exists after n years. They are the right source for odds and the wrong source for mechanism, because two businesses counted identically as "failed" can have died of opposite causes — one from no demand, one from too much of the wrong kind.

Consider the spread inside our 38. One entry spent about $35,000 building software and sold exactly one licence. Another peaked at roughly $8,000/month and then churned down. A third claimed $60K MRR with zero proof. All three are one tick in a survival table.

The percentage also cannot flag the failures that look like successes on the way past. The 97% audit reports that 97% of 1,000+ tracked vibe-coded apps were dead, abandoned, breached, or under $500/mo after six months (creator-relayed) — most of those apps still technically existed.

If you want to reason about your own odds, the useful move is to read causes and match them to your plan, using something like the SBA's guide to planning a business as the structure.

What actually killed the ones we documented

Six named cases, what they earned, and the specific ending. This is the section most directories delete.

LeadLedger is the cleanest postmortem we hold. Product one sold exactly one licence, at $300/year, to a developer in England. Product two was priced at $29/month, $99/year or a $199 lifetime licence and sold off a single webinar. Total build cost about $35,000 — $15,000 for the SDK, $20,000 for the connector. The money ran out, the partners lost interest, the site is dead.

[Yorbie](/projects/yorbie-viral-content-remix-saas) failed differently: it worked first. It peaked at roughly $8,000/month gross volume in March, dropped to about $6,700/month over the following four weeks, and ran $0 revenue for the first two months after the founder quit his job. Churn, acknowledged on camera, was the mechanism.

Cleo is the unproven tier made concrete: $60K MRR in 53 days claimed with zero proof, from a team whose earlier product Mentions did $20K MRR in month one — and the video making the claim was one link in the launch funnel.

How much is each of these numbers actually worth?

Inside this failure group the split is 5 third-party verified, 21 founder-reported, 6 creator-relayed, 6 unproven — 68% verified or founder-reported. Failure claims deserve the same scrutiny as success claims, and six of these 38 have nothing behind them at all.

The NoFap app's $6K/mo in its first month with roughly 1,100 paying users is third-party verified. The 97% figure is creator-relayed — a real tracked sample, reported by the person selling consultancy against it. Those two belong in different sentences, and most writing on failure rates puts them in the same one.

Our grading criteria are published in full at how we grade the evidence, and every entry in all indexed ideas carries its class on the card.

Why a list without failures is worse than useless

A directory that keeps only winners is selling survivorship bias with a search box. We keep 38 failures inside 406 entries on purpose, because a reader who only sees successes cannot compute anything — every pattern looks like a formula when the counter-examples have been deleted.

Here is the caveat we owe you, said plainly: even with the failures kept, this index over-represents businesses that worked and whose operators chose to publish. The medians here — $8K/mo among disclosing failures, $27K/mo across the index — are comparison tools for judging claims you read elsewhere, not expectations.

Failure detail also changes what a success means. Leftclick reports approximately $400K/month across businesses (founder-reported), with 1 Second Copy peaking at $92K/month and Leftclick scaling to $72K in a month — and its own founder argues the underlying skill is expiring. A winners-only list would print the $400K and stop.

The same applies to channel-dependent wins like the AI Faceless Facebook Content Monetization Page at $32.5K/mo with 314M views in 28 days (third-party verified). Platform-dependent revenue is real revenue with a specific failure mode, which we cover in the Risks & Rules hub.

How to use this without fooling yourself

Four checks, each tied to a number above.

1. 1.Write down your failure mode before your launch date. Match your plan against the six causes in the table: no demand, churn, no proof, wrong positioning, borrowed identity, no defensibility.
2. 2.Cap your pre-revenue spend explicitly. LeadLedger spent about $35,000 to sell one $300/year licence. Decide your ceiling now, in writing.
3. 3.Track churn from month one. Yorbie's peak was roughly $8,000/month and the decline started immediately after. If you are building recurring software, learn the metric definitions first — see What Is a SaaS? Definition, Business Model, and Real Revenue Data and What Is a SaaS Company?.
4. 4.Refuse unproven benchmarks. Six of the 38 failures and 44 entries index-wide are unproven. If a claim like Cleo's $60K MRR has no dashboard, treat it as advertising, because that is what it was.

Frequently asked questions

Is it true that 90% of startups fail?

We cannot confirm that from our data and will not repeat it as fact. Our index holds 38 documented failures inside 406 entries, and it over-represents businesses that survived and disclosed. For a population-level survival rate, read the BLS Business Employment Dynamics series or the SBA Office of Advocacy FAQ instead.

What business has a 90% success rate?

None that we index, and we would distrust any list claiming one. The closest thing to a category-level number we hold is a failure benchmark: 97% of 1,000+ tracked vibe-coded apps were dead, abandoned, breached, or under $500/mo after six months — creator-relayed, not verified.

What are the odds of a business failing?

Odds come from government survival data, not from us. What we contribute is 38 documented endings with causes attached: 23 of them one-person businesses, 6 disclosing a clean monthly figure at a median of $8K/mo, spanning $29/mo to $400K/mo before they stopped.

How reliable are these what percent of business fail figures?

The failure group splits 5 third-party verified, 21 founder-reported, 6 creator-relayed, 6 unproven — 68% verified or founder-reported. Index-wide it is 57 verified, 184 founder-reported, 121 creator-relayed, 44 unproven across 406 businesses in 22 countries. Every figure carries its class.

How many of the 38 indexed businesses actually disclose a monthly number?

Six, at a median of $8K/mo. That low disclosure rate is itself a finding: people publish dashboards when revenue is climbing and stop when it is not, which is exactly why failure data is scarce and why we keep every case we can document.