Fintech Go/No-Go: Kill Criteria Before You Commit
Set kill criteria for a fintech idea before you start validating it, not after. Write down four to six specific, numeric thresholds — on unit economics, regulatory pathway, trust, and demand — that mean stop if you hit them. Set them while you're still neutral about the outcome, because six months and a pile of sunk cost from now, you won't be.
Quick Answer: Pre-commit to explicit stop conditions on economics, regulation, trust, and demand before you build anything. Write the numbers down, share them with someone who isn't emotionally invested, and treat hitting one as a decision already made — not a debate to reopen.
This isn't a pessimism exercise. It's a discipline for making the stop decision before the moment you're least equipped to make it well. The same logic applies to any startup — the general framework for killing a bad idea fast covers the founder-psychology side of that problem. Fintech just raises the stakes and shortens the runway: regulatory review cycles, banking-partner due diligence, and compliance spend can burn a year and six figures before you learn whether the idea actually works. If you haven't yet scoped the regulatory and market-research side of a fintech idea, the complete guide to validating a fintech startup idea covers that ground; this article assumes you're already there and need the stop-loss layer on top of it.
Peter Thiel's Zero to One draws a distinction between definite and indefinite optimism: a definite founder decides in advance what has to be true and builds toward it, while an indefinite one waits to see what happens and reacts to whatever shows up. Kill criteria are how you stay definite under pressure. They're conclusions you draw now, while your incentives are still aligned with clear thinking, so you don't have to draw them later, when a lease, a co-founder's mortgage, and eighteen months of your own identity are riding on the answer.
The Kill Criteria Template
Fill this in before you run your first pilot, not after you're three months into one. Each row needs a threshold specific enough that you (or a co-founder, in six months, arguing the other side) can't talk your way past it.
| Criterion | Threshold (fill in your number) | Evidence Required |
|---|---|---|
| Unit economics | Customer acquisition cost exceeds [X]× projected lifetime value after [N] pilot customers | Cohort-level economics from real transactions — not a spreadsheet model |
| Regulatory pathway | No licensing route or bank-partner path identified within [N] weeks of dedicated research | Written opinion from a fintech-specialist lawyer or compliance consultant |
| Loss / fraud rate | Underwriting or fraud losses exceed [X]% during the pilot, a rate your margin can't absorb | Actual loss data from a live or shadow-mode pilot cohort |
| Trust / onboarding conversion | Sign-up-to-funded-account drop-off exceeds [X]% at identity verification or funding | Funnel analytics from a real onboarding flow, not a mockup |
| Distribution dependency | The model requires one partner (bank, card network, processor) that hasn't committed in writing | Signed LOI, term sheet, or partnership draft — a friendly phone call doesn't count |
| Demand from paying users | Fewer than [N] qualified prospects would pay or switch, measured in structured interviews or a waitlist | Interview transcripts or pre-launch commitments, not friendliness in a call |
Two things about this table matter more than the exact thresholds you fill in. First, every "evidence required" column insists on something you can point to — a document, a data export, a transcript — not a feeling. Second, the table is deliberately short. A twenty-row kill checklist gets skimmed once and ignored; six rows you've actually filled in with numbers get checked. If you want a broader pre-build scoring pass across more dimensions — demand, regulation, economics, data access, trust, and moat — pair this template with the fintech idea scorecard, which is built for exactly that wider sweep before you narrow down to these stop conditions.
What Each Kill Criterion Actually Means
Unit economics
A fintech idea with an acquisition cost that can never be recovered from lifetime value is not a growth problem — it's a business-model problem. Financial products often carry regulatory and compliance overhead that consumer apps don't, which pushes the cost of acquiring and servicing a customer up in ways that are easy to underestimate from a spreadsheet. Set the threshold on the ratio (CAC relative to a multiple of LTV), not on an absolute dollar figure, since absolute figures shift with pricing experiments while the ratio tells you whether the model works at all.
Regulatory pathway
Most fintech ideas that die, die here — not because a regulator issues a formal rejection, but because nobody can find a bank willing to sponsor the program, or a state-by-state licensing effort turns out to cost more than the company has raised. Set a hard research window: if a specialist can't identify a plausible path to operate legally within that window, that's information, not a temporary setback to push through with more effort.
Loss and fraud rate
Every consumer credit, lending, or payments product loses money on a subset of transactions; the business only works if that loss rate stays inside a range your margin can absorb. Set this threshold before you see pilot data, because after you see it, you'll be tempted to rationalize a number that's actually too high by telling yourself the pilot cohort was unrepresentative.
Trust and onboarding conversion
Fintech asks users to hand over information — identity documents, bank credentials, income data — that most software never touches. A high drop-off rate at identity verification or account funding isn't a UX polish problem you fix with a better button; it's often a trust problem, and trust problems are the hardest and slowest thing to fix in this category. Treat a bad conversion number here as a signal about the product's premise, not just its interface.
Distribution dependency
If the entire business depends on one bank, one card network, or one processor agreeing to work with you, and that partner hasn't put anything in writing, you don't have a business yet — you have a pitch to a partner who could walk away. A verbal "this sounds interesting" from a business-development contact is not evidence; a signed letter of intent is.
Demand from paying users
This is the criterion founders skip most often, because fintech ideas can feel self-evidently necessary — "everyone needs to manage money better." Necessity is not the same as willingness to switch providers or pay for a new one. Structured interviews that surface a specific number of qualified prospects willing to commit, not just agree the problem is real, are what this threshold requires.
A Filled Example: Kill Criteria for a Consumer Lending Idea
Here's what the template looks like filled in for a hypothetical short-term consumer lending product — the numbers are illustrative, meant to show the shape of a real answer, not a benchmark to copy:
- Unit economics: Kill if CAC exceeds 3× projected first-year LTV after the first 200 funded loans.
- Regulatory pathway: Kill if no state licensing route or bank-partner structure is identified within 8 weeks of engaging counsel.
- Loss rate: Kill if the pilot's default rate exceeds the rate the pricing model assumes by more than 3 percentage points, sustained across two consecutive cohorts.
- Trust/onboarding: Kill if more than half of applicants who start identity verification abandon before funding.
- Distribution: Kill if no bank-sponsorship or lending-partner agreement is signed within 90 days of needing one to originate loans.
- Demand: Kill if fewer than 30 of 50 structured interviews produce a prospect who says they'd switch lenders for this specific offer, not just that the offer sounds appealing.
Notice what this list does that a vague "we'll know it when we see it" plan doesn't: every line has a number and a deadline. A founder reading this in month four, tempted to keep going because "we're so close," has to argue against a number they set themselves while clear-headed — a much harder argument to win than talking yourself out of a fuzzy hunch.
Adjusting the Template by Vertical
The six categories hold across fintech, but which one fires first — and how tight the threshold needs to be — shifts by vertical.
| Vertical | Criterion most likely to kill first | Why it dominates |
|---|---|---|
| Consumer lending / credit | Loss rate or regulatory pathway | Underwriting risk and state-by-state licensing hit before distribution does |
| Payments / money movement | Distribution dependency | The product is nearly worthless without a processor or network relationship in place |
| Insurtech | Regulatory pathway | Insurance licensing and reserve requirements are often stricter and slower than banking |
| Wealthtech / investing | Trust and onboarding conversion | Users are asked to move existing assets, not just open a new account — the trust bar is higher |
| B2B / embedded finance | Distribution dependency | The whole model routes through a platform partner's willingness to embed your product |
Payments and embedded-finance ideas should weight distribution dependency more heavily than the base template suggests, because the product frequently can't function as a standalone offering at all — it exists as a feature inside someone else's platform, and that someone else's roadmap now controls your timeline. Insurtech and lending ideas should weight the regulatory pathway criterion earliest, often before writing a line of product code, because state and federal licensing timelines are largely outside your control and can eat a runway that would otherwise fund months of building. Wealthtech ideas should set an unusually strict trust threshold, since asking someone to move retirement or investment assets is a categorically bigger ask than a new budgeting app, and a mediocre conversion number here tends to reflect a real trust deficit rather than a fixable funnel step.
None of this means abandoning the other criteria for a given vertical — it means deciding, before you start, which one you expect to be the binding constraint, so you design your earliest pilot to test that one first instead of discovering it last.
Key Takeaways
- Set kill criteria before validation starts, not during it. Once you're emotionally and financially invested, you lose the neutrality needed to set an honest threshold.
- Every threshold needs a number and a deadline, not a vague description — "too expensive to acquire customers" isn't a kill criterion; "CAC exceeds 3x LTV after 200 customers" is.
- Regulatory pathway and distribution dependency kill more fintech ideas than product quality does. Test these early, not after you've built something.
- Loss and fraud rates should be set before you see pilot data, or you'll rationalize a number you'd have rejected in advance.
- A trust or onboarding-conversion failure is rarely a UX problem — treat a bad number here as a signal about the underlying premise.
- The criterion most likely to fire first depends on your vertical — lending and insurtech tend to hit regulation early, payments and embedded finance tend to hit distribution early, wealthtech tends to hit trust early.
- Write the thresholds down and share them with someone outside the decision — a co-founder, advisor, or investor who will hold you to the number when you're tempted not to be.
Frequently Asked Questions
How many kill criteria should a fintech idea have?
Four to six is usually enough. Fewer than four tends to miss a major failure mode (economics, regulation, trust, or demand); more than six gets skimmed once during planning and then ignored under pressure. Pick the smallest set that covers your idea's real risk categories.
Can you change a kill criterion after you've set it?
Only with real new evidence, not with discomfort about the outcome. If a threshold turns out to have been set on a wrong assumption — a regulatory rule you misread, a cost you miscalculated — revise it and document why. If the only thing that's changed is how close you are to hitting it, that's the criterion working, not a reason to move it.
What if I hit a kill criterion but investors still want me to continue?
Hitting a kill criterion is information for that conversation, not a decision investors get to override by preference. Share the evidence you set the threshold to require and let them make the case for continuing with specifics — a new partner commitment, a corrected number — rather than general optimism. If nobody can produce new evidence, the criterion did its job.
Is it possible to pass every kill criterion and still have a bad fintech idea?
Yes. Kill criteria are a floor, not a guarantee — they're designed to catch the failure modes that are easy to rationalize away, not to certify that an idea will succeed. Passing all of them means the idea has cleared the risks most likely to kill it silently; it still needs the rest of your validation process, competitive analysis, and go-to-market work to earn a real go decision.