The Startup Pre-Mortem: Kill Your Idea Before the Market Does

A post-mortem asks why something died. A pre-mortem asks the same question before you spend the money.

The technique comes from psychologist Gary Klein, and the mechanism is simple: instead of asking "what might go wrong?", you assert that it has gone wrong and work backwards. It is eighteen months from now. The company is dead. Write the story of how.

That reframe matters more than it sounds. "What could go wrong" invites your brain to defend the plan. "It failed — why?" invites your brain to explain it. People are dramatically better at explaining than at forecasting, and Klein's research found the prospective-hindsight framing increases the number of correctly identified causes by around 30%.

Why founders are structurally bad at this alone

You cannot easily pre-mortem your own idea, for three reasons that have nothing to do with intelligence.

Sunk cost accumulates fast. By the time you have a name and a logo, killing the idea costs you identity, not just time. The instinct to defend arrives long before any real investment does.

You know too much. You have already unconsciously answered the hardest objections, so they no longer register as objections. The gaps in your reasoning are invisible precisely because you filled them in months ago.

Your friends are not adversaries. Asking a supportive person to attack your idea produces a polite, hedged version of an attack. You need someone whose job is to find the flaw, not to spare your feelings.

How to run one in an hour

Step 1 — Set the scene concretely (5 minutes)

Pick a date far enough out that failure is plausible: twelve to eighteen months. Write one sentence in past tense. "In February 2028 we shut down after failing to reach 200 paying customers."

Be specific about the failure mode. "It did not work" produces vague causes. "We could not get past 200 customers" produces sharp ones.

Step 2 — Write causes for ten minutes without filtering (10 minutes)

Every reason you can think of. Do not evaluate, do not rank, do not dismiss anything as unlikely. Volume first. Most people produce fifteen to twenty-five causes; the useful ones are usually in the back half, after the obvious answers are exhausted.

Step 3 — Sort into the four real categories (15 minutes)

• Market: nobody wanted it, the segment was too small, the timing was wrong.
• Product: we built the wrong thing, it was too hard to adopt, the core mechanic did not deliver.
• Economics: CAC exceeded LTV, the price was wrong, churn ate the growth.
• Execution: we ran out of money, the team broke, we were too slow.

The distribution itself is informative. If nearly all your causes are execution, you are probably not being honest about market risk — execution failures feel more flattering, because they imply the idea was fine.

Step 4 — Score each by likelihood and cost (15 minutes)

Two numbers, one to five. How likely is this cause, and how expensive is it if it happens? Multiply. Anything scoring 15 or higher is not a risk to monitor — it is a thing to resolve before you build.

Step 5 — Decide: kill, de-risk, or accept (15 minutes)

For every high-scoring cause, choose one explicitly.

Kill if the cause is fatal and unfixable — a regulatory blocker you cannot price in, a market that structurally does not exist. Killing an idea in hour one is the cheapest win available to you.

De-risk if you can convert the unknown into a test. "Nobody will pay $19/month" becomes "call ten walkers this week and ask." Most high scores should land here.

Accept if the risk is real but you are choosing to carry it. Write down why. Accepted risks that were never written down are the ones that kill companies, because nobody remembers deciding.

What a pre-mortem reliably catches

Across early-stage ideas, a handful of causes show up again and again — and founders are consistently surprised by the same ones.

The payer is wrong. You built for the person with the pain, not the person with the budget. Extremely common in marketplaces and anything selling into a workflow.

The switching cost is invisible to you. Your product is better, but the customer has three years of data in a spreadsheet and no appetite to move it. "Better" loses to "already here" more often than founders expect.

CAC assumptions are decorative. Most early models assume a customer acquisition cost that has never been tested. If your entire viability rests on a number you guessed, that is your top-scoring risk by definition.

Onboarding drop-off. The product works, but there are six steps before value appears and most users leave at step three. This is a product failure that reads like a marketing failure, which is why it goes undiagnosed for months.

A regulatory or platform dependency. Anything built on a single API, app store, or licence regime has a failure mode that is entirely outside your control.

Pre-mortem vs. validation — they are not the same thing

Validation asks "is there a real market here?" A pre-mortem asks "assuming there is, what kills us anyway?" You need both, in that order. There is no point stress-testing the execution of a product nobody wants, which is why the 5-minute validation framework comes first.

The two also fail differently. Validation failures send you to a new idea. Pre-mortem failures usually send you to a modified version of the same one — a different segment, a different price, a narrower first release.

Frequently asked questions

When should I run a pre-mortem?

Before any expensive commitment: writing real code, hiring, signing a lease, raising. And again after any major pivot, because a pivot invalidates most of your previous answers.

Does it work with just one person?

Partially. Solo, you will find the causes you were already half-aware of and miss the ones your optimism is hiding. The technique depends on adversarial distance, which is exactly what you cannot supply about your own idea.

How is this different from a SWOT analysis?

SWOT is a static list that tends to flatter you — threats get one quadrant out of four. A pre-mortem forces a causal narrative ending in failure, which is far harder to fill with comfortable answers.

What if the pre-mortem says the idea is dead?

Then it did its job for the price of an hour. Most founders run three to five ideas through this before one survives, and the survivor is materially stronger for the ones that did not.

Get an adversary that has no reason to be kind

The hard part of a pre-mortem is not the process. It is finding someone who will genuinely attack the idea rather than perform an attack politely.

Forze's Shadow Board convenes three adversarial personas against your venture — a skeptic, an evangelist, and an operator — and returns a Survival Score out of 100, the critical blind spots each one found, and the pivots worth considering. It reads your actual brief, not a template, so the objections are about your business rather than startups in general.

Run your Shadow Board free →

Survived the pre-mortem? Here is how to decide between an AI MVP and hiring a developer.