A third-party archive of documented startup failures, compiled from six public source lists. It is a working record for framing commercial tests. It is not proof of what causes failure or of what successful companies do.
A third-party compilation of 2,715 documented startup failures, one row per company, drawn from six public source lists: Loot Drop, a Kaggle dataset of CB Insights records, CB Insights, and three Failory cemetery lists. GTM Right did not collect these records through original fieldwork, and no customer work is added to it.
A company is included because one of those lists documented it as shut down or discontinued. The lists were reconciled into a single row per company; no company name appears twice in the current file. This is not a random or representative sample of startups: it can over-represent companies that were well known or well funded enough to be written about, and the Failory Google and Amazon lists include discontinued products and acquired companies associated with those two firms, not only independent startups.
The current archive file (2,715 records, last revised 20 July 2026, after the June 2026 Failure Intelligence report). Three records are marked Reclassified because later research found they were not go-to-market failures; they are kept so the published count is unchanged.
Fields are shown as the sources recorded them and are not guessed. Where a source gives no cause, country, sector or closure year, the record stays Unknown for that field. Where a source records multiple reasons, the record is not split or reassigned to one of them.
The archive contains only companies that failed. There is no matched group of companies that survived, so it cannot show what successful companies do differently, and a count of records in a sector is not a failure rate.
A cause of death is what a source reported, not a verified or tested cause. The patterns in the archive are directional, not causal, and nothing here shows that GTM Right would have prevented any failure.
The archive also carries a stage label assigned under an earlier version of GTM Right's stage order. Those legacy assignments are not used on this page to count, rank, cost or otherwise quantify today's six-stage framework.
Records by source list: Loot Drop 1,675, Kaggle/CB Insights 510, CB Insights 442, Failory Cemetery 30, Failory Google Cemetery 29, Failory Amazon Cemetery 29; total 2,715.
2,178 of 2,715 records have a closure year, and 2,035 of those 2,178 fall between 2008 and 2024. 741 of 2,715 records have no recorded cause. 1,475 of 2,715 records list the United States. Record counts are not failure rates.
These six lenses are GTM Right's editorial hypotheses for founders to test, one per current stage. They were not statistically derived from the 2,715 records, and they do not explain, predict or prevent any failure in the archive.
Stage 1: True Validation. The Painless Product (editorial hypothesis): The problem may not be urgent, widespread or painful enough for buyers to act on. Question to test: Is the pain urgent, widespread and actively complained about?
Stage 2: Buyer Precision. The Invisible Buyer (editorial hypothesis): The people who feel the problem may not be the people with the budget and authority to buy. Question to test: Can you name the first reachable buyer specifically enough to find them this week?
Stage 3: Buyer + Offer. The Feature Without a Job (editorial hypothesis): The offer may not replace anything the buyer does today, or be hired for a clear job. Question to test: What does this product replace, and who would hire it to do that?
Stage 4: Revenue Funnel. The Broken Funnel (editorial hypothesis): The path from stranger to paying customer may not hold once acquisition cost and conversion are real. Question to test: Can one customer become profitable without relying on future scale?
Stage 5: Market + Funnel. The Crowded Lane (editorial hypothesis): Incumbents and the real competitive set may leave no entry point you can defend. Question to test: Which competitor's underserved segment is a wedge you can hold?
Stage 6: Commercial Decision. The Regulatory Blindspot (editorial hypothesis): A legal, regulatory or trust requirement may block launch or adoption. Question to test: Could one legal, regulatory or trust issue block launch or adoption?
1. Choose an assumption: Pick the lens closest to the assumption your plan most depends on and that you have not yet tested.
2. Form a falsifiable test: Write down, before you start, what result would show the assumption is wrong.
3. Gather evidence: Collect what buyers and the market actually do, not what they say they might do.
4. Decide what remains uncertain: Record what the evidence supports, what it does not, and the next test that would reduce the uncertainty.
The archive can suggest what to investigate. These guides help turn that hypothesis into a founder-run evidence test.
GTM Right (2026). GTM Right Failure Archive: 2,715 documented startup failures compiled from 6 public source lists. Version of 20 July 2026. https://www.gtmright.com/research
Cite the version date with any figure: the archive file is revised, and a count taken from one version will not match another. The records were compiled from published third-party lists, so cite those lists for the underlying records, and do not describe a count of records as a failure rate.
The commercial failure patterns: The recurring commercial patterns behind failures like these, each with its mechanism and the signal that shows it is present.
Interest is not demand: How to tell a buying signal from a polite one when you are reading your own evidence.
Browse the archive: The full record set, filterable by sector, funding band, country, year and recorded cause.
The methodology: How GTM Right decides what counts as evidence in each stage.
True Validation tests public market evidence only: whether the problem is real, visible and urgent. It does not prove buyer budget, willingness to pay, offer fit, acquisition economics or full commercial viability. Run the free Stage 1 check.