A kill test for SaaS ideas
Before you build another SaaS, find the reason it will fail.
AI made building cheap. It didn't make markets forgiving. The Kill Test is a structured research method for solo founders and indie hackers: stress-test the incumbents, the distribution path, the API access, and the platform risk behind your idea — in an afternoon, not after three weeks of building.
One-time purchase. No subscription. No account.
Jira Automation Conflict Preflight
- Paid demand
- Pass
- Incumbent coverage
- Estimate 30–40%
- Technical feasibility
- Pass
- Commercial API reality
- Fatal
“Forge and OAuth2 apps cannot access this REST resource.” — Atlassian developer documentation, verified 2026-08-13
The most expensive line of code is the first one you shouldn't have written.
With AI-assisted coding, a working prototype costs a weekend. That's the trap: when building is cheap, the real cost moves to choosing what to build.
Most abandoned side-project SaaS didn't fail on code. They failed on things a prototype can't tell you: a paid incumbent already covered the workflow, the platform's API turned out to be closed to commercial apps, there was no path to the first 100 users without outbound sales, or the support burden could never fit a solo founder's life. You usually discover these after the build — in week four, when the product is live and nobody buys.
Every one of those failure modes was checkable in advance, for the cost of an afternoon of structured research.
The question generic validation never asks
Idea validators, lean canvases, and AI scorecards ask: “Is this a real problem? Is the market big enough?” Useful — and not the question that kills most solo SaaS. Mature markets are full of real problems that are already mostly solved by paid incumbents. The question that decides your outcome is:
After the paid incumbents have solved most of the workflow, is there still a commercially reachable job left — one valuable enough that someone will buy a standalone product for it?
We call the answer to that question the paid white space. Most ideas don't have one. The Kill Test exists to find out before you build.
| Question | Generic idea validation | The Kill Test |
|---|---|---|
| Is the problem real? | Yes — surveys, interviews, TAM | Yes — but with paid-demand evidence, not opinions |
| Do competitors exist? | Binary: yes / no | What percentage of the workflow do paid incumbents already cover? |
| Can you technically build it? | Rarely asked | Can your commercial architecture legally reach the data it needs? |
| Will the platform absorb you? | Not asked | Platform capture risk over the next 6–24 months |
| Can you reach the first 100 buyers? | “Marketing comes later” | Distribution reality without outbound sales |
| Can one person run it? | Not asked | Support minutes per customer, onboarding burden, fork risk |
Watch a promising idea die in section 7
Memo no. 001 analyzes a genuinely attractive idea: a static analyzer that catches conflicts in Jira automation rules before they fire. Real pain, deterministic analysis, a marketplace with built-in distribution. Five of seven tests pass.
Then the commercial API test runs, and one sentence in Atlassian's own developer documentation ends the project:
“Forge and OAuth2 apps cannot access this REST resource.”
The rule data the product must read is walled off from exactly the architecture a self-serve marketplace app requires. The problem is real, the analyzer is buildable — and the business is dead. Finding this took about two hours of structured research. Finding it the usual way takes three weeks of building.
Seven tests. Any one of them can kill.
The Kill Test is not a weighted score — an idea that averages 8/10 but fails one fatal test is still dead. You run seven tests against your idea, in order, using the guided research playbook. Each test produces evidence, and every claim gets classified honestly: Fact Estimate Hypothesis Unknown
| # | Test | The question it answers |
|---|---|---|
| T1 | Paid demand | Is anyone already paying to make this pain go away — or only complaining about it? |
| T2 | Incumbent coverage | What share of the workflow do paid incumbents already solve? Not “do competitors exist” — how much is left. |
| T3 | Paid white space | Is the remaining job frequent, costly, standardized, and valuable enough to buy as a standalone product? |
| T4 | Platform capture | Will the platform owner or an incumbent absorb this gap within 6–24 months? |
| T5 | API / commercial reality | Can your intended architecture actually reach the required data — commercially, not just technically? |
| T6 | Distribution reality | Is there a credible path to the first 100 buyers without outbound sales? |
| T7 | Solo economics | Can one person support, onboard, and retain customers without drowning? |
The output is one of three verdicts, decided by fatal flags rather than averages: BUILD, TEST THIS ASSUMPTION — when exactly one unknown stands between you and a build decision — or KILL.
Run it on your own idea
Complete kill-test system
$49 one-time- The seven fatal tests, with step-by-step research instructions for each
- The research query playbook — where to look, what to search, what counts as evidence
- The evidence ledger — classify every claim as fact, estimate, hypothesis, or unknown
- Fatal-flag verdict logic — BUILD / TEST THIS ASSUMPTION / KILL, no vanity scores
- Memo no. 001 — the complete Jira Automation case, as a worked reference
- Delivered as a structured document workspace. Yours permanently.
14-day refund, no questions asked · no subscription · no account
Not sure yet? Read the full sample memo first — it's the exact document the method produces.
Fair questions
- Can't I just ask ChatGPT to validate my idea?
- You should — and it will usually tell you your idea is promising. Chat answers optimize for plausibility, not for finding the one closed API scope or coverage gap that kills a business. The Kill Test is a research procedure with evidence standards: every claim is sourced, dated, and classified. The sample memo shows the difference — its fatal finding is a single sentence in vendor documentation that generic validation never goes looking for.
- What exactly do I get, and how is it delivered?
- A structured document workspace you own permanently: the seven fatal tests with step-by-step research instructions, the research query playbook, evidence-ledger templates, the verdict logic, and the complete sample memo as a worked reference. Your access link arrives by email after checkout. No account, no app to install.
- Is this an idea generator or a market research report?
- Neither. It's a decision system you run yourself on your own idea. It doesn't hand you opportunities; it tells you whether the one you have deserves your next three weeks.
- What if it doesn't tell me anything I didn't know?
- Then it failed, and you get your $49 back. Fourteen days, no questions, no form to argue your way through — one email to info.customer2026@gmail.com. A method that claims to find the fatal assumption in your idea should be willing to be judged by whether it did. The risk of running the tests is meant to sit with us, not with you.
- Who is this not for?
- Funded teams with sales staff, and builders who are building for fun or learning — if the building itself is the reward, you don't need permission. The Kill Test is for solo founders whose scarcest asset is their next month.