Three startups, three shutdowns, three different missing checks. After the third one I realised the same framework would have killed every bad bet before it cost a year. This is the framework. None of the three checks is novel,versions live in Zero to One, in CB Insights' corpus, in Wilbur Labs' founder survey. What's rare is having to assemble it from your own failures.

If you're building right now, this is a pre-mortem. Three sentences before you commit. If you can't complete all three with specifics,name the dimension, name the new job class, name the margin,you have a check that is not yet green. Do not commit until it is.

The framework in one paragraph

Every startup that survives early-stage death has structurally cleared three independent obstacles: it is so much better than its closest substitute that switching makes obvious sense (10x), its TAM compounds as it succeeds (scaling), and the per-active-user economics are positive without subsidy (margins). Each obstacle has its own failure mode. Each one is sufficient to kill a company without help from the others.

The framework is descriptive, not prescriptive,it does not tell you what to build, it tells you what conditions must be true for what you've already chosen to be viable. Most founders run it implicitly during the postmortem. The cost of running it as a pre-mortem is one afternoon. The cost of running it as a postmortem is twelve to twenty-four months.

Check 1,10x value

Peter Thiel's formulation in Zero to One: "Proprietary technology must be at least 10 times better than its closest substitute" (Volt Equity, summary of Zero to One). The point is not that ten is a magic number; it's that "marginally better" is invisible to users in markets with strong incumbents.

The test is concrete: name the dimension, name the multiple. "10x faster on deploy time" is a 10x. "10x cheaper at the same quality" is a 10x. "Easier to use" is not a 10x,it's a feature. "Gentler in tone" is not a 10x,it's a tone. MindWave failed this check. The pitch was about tone. The user-perceived gap from the incumbent was small. The product was useful, not category-shifting.

Where founders deceive themselves on this check: confusing founder-perceived gap with user-perceived gap. The founder, who has lived inside the product for a year, sees every detail. The user, who is comparing in 30 seconds against an existing alternative, sees the elevator pitch. The 10x has to be visible to the user in 30 seconds. If it isn't, it isn't 10x.

Three checks. Any one missing is sufficient cause of death. 10x value Scaling Margins Viable Without scaling = niche Without 10x = trivial Without margins = race to zero
The three-check framework. Two-out-of-three intersections still leave a fatal gap.

Check 2,Scaling potential

Scaling is not "user growth". It's "TAM expansion",does success mean more of the same job, or does it open new jobs? An aggregator product grows users without entering new categories. A product company grows by entering new categories of work.

The test: name a new job class success opens. Super AI failed this check. Each new user routed to one of the same models,sales, support, finance, ops, all funnelled into the same routing job. The kind of growth that opens new jobs was not there. By contrast, an inbox-triage agent that succeeds opens lead-followup, then KPI-reporting, then customer-followup,each a new vertical, not a deeper version of the same one.

This check matters most when the underlying infrastructure is commoditising. In those markets, aggregator products lose margin defence as the thing they aggregate becomes cheaper. The product company keeps its margin because it owns the outcome, not the routing. Specific beats horizontal in commoditised markets.

Check 3,Sustainable margins

Per-user economics positive without subsidy. If a free-tier user costs more in compute than a paid-tier user pays in revenue, growth deepens losses. The math does not fix itself with scale.

The test: name the per-active-user margin at scale. Vibe AI failed this check. Long contextual conversations were compute-expensive at scale; the audience would not pay subscription prices that matched the cost. The retention curve and the cost curve crossed before any paid model converged. The macro pattern across the largest generative-AI products is similar: industry coverage of compute economics has consistently documented per-user costs running multiples of revenue at consumer price points (CB Insights, "Why Startups Fail", 2026). Negative unit economics is the cleanest predictor of capital exhaustion.

Where founders deceive themselves on this check: assuming scale fixes a margin gap. It does not. If your unit economics are negative, scale just makes the loss bigger faster. Capital can subsidise the gap; it cannot close it. Negative-unit-economics businesses do not become positive-unit-economics businesses by adding users.

It's an AND, not an OR. Two-out-of-three is still dead.

The three checks are an AND, not an OR. MindWave had scaling potential and decent margins; missing the 10x killed it. Super AI had 10x speed and decent margins; missing scaling killed it. Vibe AI had 10x emotional value and scaling; missing margins killed it. None of the three required heroic execution,they required the right premise.

That's the test. Two-out-of-three is still dead,just slower. The slow version is more painful because it gives you more time to convince yourself you can fix the missing check. You usually cannot. The check is structural; execution does not move it. The answer is to start with all three green or to start with a different idea.

How to run it as a pre-mortem

Three sentences. Write them out, in this order, before you commit:

  1. "My product is at least 10 times better than [closest substitute] on [single measurable dimension]." If you cannot fill in both blanks with specifics,a named competitor and a named metric,you do not yet have a 10x. Tone, vibe, and aesthetic do not count.
  2. "Success in [vertical 1] opens [vertical 2], which opens [vertical 3]." If your three verticals are the same job applied to different users, you have user growth, not scaling potential.
  3. "At scale, the per-active-user margin is [positive number] dollars at [price point], with [cost-of-goods] per user." If you cannot name a positive number and a defensible cost model, you do not yet have sustainable margins.

If any sentence is missing a specific, the check is not yet green. The fix is sometimes to keep building until you can answer; more often the fix is to choose a different premise. The framework's job is not to make you wait,it's to make sure that when you commit, you commit to something that can survive its own success.

For Gravity, this exercise is the reason bet four feels different. 60-second deploy is a measurable 10x against days-to-weeks competitors. The TAM is every recurring task in every knowledge worker's day,sales, support, content, ops, personal,each a new job class. Capability-based pricing keeps the per-active-agent margin computable in advance. Three sentences. All three filled in with specifics. That's the bar. The full synthesis is in three startups, three shutdowns.

Frequently asked questions

What are the three checks every startup must pass?

Three checks: 10x value (the product is at least ten times better than its closest substitute on a measurable dimension), scaling potential (success opens new job classes, not just deeper use of the same one), and sustainable margins (per-user economics positive without subsidy). All three are required. The absence of any one is sufficient cause of death.

What is the 10x rule for startups?

Peter Thiel's 10x rule from Zero to One holds that a new product must be at least ten times better than its closest substitute on at least one dimension to escape competition. Anything less is perceived as marginal and gets crushed by incumbents in crowded markets.

Why isn't user growth the same as scaling potential?

User growth is more users using the same job. Scaling potential is whether success opens new job classes. Aggregator products grow users without expanding TAM; product companies grow by entering new categories of work. The distinction is what determines whether your TAM compounds or saturates.

Why are unsustainable margins fatal?

If per-user economics are negative, growth deepens losses. CB Insights' 2026 corpus of 431 failed startups found unsustainable unit economics as an underlying cause for 19% of failures, with capital exhaustion as the surface cause for 70%. Negative unit economics is one of the cleanest predictors of capital exhaustion.

How do I run this as a pre-mortem instead of a postmortem?

Three sentences before you commit: name a single dimension where your product is 10x better than the closest substitute, name a new job class success opens, and name the per-user margin at scale. If you cannot complete all three sentences with specifics, you have a check that is not yet green. Do not commit until it is.

Three takeaways before you close this tab

Sources