Business Idea Evaluation

What does the strongest available evidence justify doing next?

Problem: Founders can collect enthusiasm, market statistics, and feature ideas without learning whether a specific beneficiary urgently wants a better outcome.

Decision: Assign exactly one disposition: ADVANCE, TEST, PARK, or KILL.

This method produces an evidence card, a cheapest falsifying experiment, and a dated disposition.

Evidence Standard

Start with lived friction and a named beneficiary. Test urgency through cost of delay and the workaround already being used. Prefer narrow, deep demand that you can observe over broad stated interest.

Rank signals by how much the beneficiary risks or changes:

1. Repeat use or continued reliance.
2. Payment or another costly value exchange.
3. Commitment of time, reputation, access, data, or a scheduled next step.
4. Tolerance for a rough, manual, or inconvenient version because the outcome matters.
5. Observed workaround or repeated problem-solving behaviour.
6. Specific testimony about a past event.
7. Compliments, hypothetical intent, votes, survey preference, and broad audience metrics.

Lower-ranked evidence can guide a question; it cannot impersonate higher-ranked evidence. Novelty is no evidence. Tedious or unglamorous work is no reason to reject an idea.

Inputs

• An Idea Capture card.
• Sources and dates for known facts.
• Constraints on founder time, capital, access, safety, consent, and reversibility.
• A review point and the decision owner.

Steps

1. Reconstruct the friction. State the beneficiary, struggling moment, desired outcome, current workaround, frequency, and cost of delay. Output: a problem record with facts and assumptions separated.
2. Assess urgency. Look for what the beneficiary has already changed, spent, risked, repeated, or tolerated. Output: an evidence ledger ordered by signal strength.
3. Test demand depth. Ask whether a narrow reachable group shows costly behaviour, not whether a broad group says the idea is interesting. Output: a bounded demand claim and its missing proof.
4. Check founder access. Record why you can observe, reach, understand, or serve this beneficiary better than a generic entrant. Output: an access claim and falsifier.
5. Check distribution. Name the first credible path to the beneficiary, the trust required, and the cost or constraint of using it. Output: a distribution hypothesis.
6. Check value capture. State who receives value, who pays, what they exchange, and whether delivery can retain enough value to continue. Output: a value-exchange hypothesis.
7. Check delivery constraints. Expose regulation, safety, data, dependency, capability, capital, and time constraints before commitment. Output: a constraint ledger with stop conditions.
8. Choose the cheapest falsifying experiment. Target the assumption with the greatest combination of uncertainty and consequence. Declare a context-specific pass threshold, fail threshold, resource limit, and ambiguous-result rule before running it. Output: a bounded test protocol.
9. Complete the evidence card. Record the decision and its review point. Output: one durable evaluation record.

Evidence Card

Field	Entry
Named beneficiary and lived friction
Known facts and sources
Assumptions
Strongest behavioural evidence
Current workaround and cost of delay
Founder access
Distribution hypothesis
Value-capture hypothesis
Constraints
Missing proof
Cheapest next test
Predeclared pass and fail thresholds
Kill signal
Review point and owner
Disposition and rationale	ADVANCE / TEST / PARK / KILL

Disposition Rules

• ADVANCE — the critical demand and access assumptions have behavioural support, no stop condition is active, and the next commitment is bounded and reversible.
• TEST — the idea remains plausible but one consequential uncertainty can be reduced through the named cheapest test.
• PARK — the idea may be worthwhile, but timing, access, capacity, or another prerequisite blocks an honest test now; name the event that reopens review.
• KILL — evidence contradicts a critical assumption, a constraint makes responsible delivery unacceptable, or the cheapest relevant test has failed its predeclared threshold.

Do not average fatal evidence into a score. Do not use fixed interview counts, audience sizes, search volumes, or generic improvement multiples as universal gates. The decision owner must declare thresholds that fit the idea, evidence maturity, downside, and cost of being wrong before the test begins.

Checks

• Every material claim is tagged as known fact, assumption, or missing proof.
• Behaviour, commitment, payment, repeat use, and rough-version tolerance outrank compliments and surveys.
• The experiment could produce a KILL or PARK result, not only confirmation.
• The disposition is exactly one allowed value and includes a rationale.
• PARK includes a reopening trigger; TEST includes a review point; ADVANCE names the bounded next commitment; KILL names the contradicted assumption or active stop condition.

Failure Modes

• Opinion as evidence — enthusiasm or survey intent is reported as demand.
• Vanity threshold — a universal count substitutes for the economics and risk of this idea.
• Novelty premium — technical newness raises conviction without beneficiary behaviour.
• Score averaging — strong presentation hides a fatal demand, access, safety, or value-capture failure.
• Endless testing — no threshold, review point, or disposition closes the loop.

Proof Of Done

An unfamiliar founder can read the record, distinguish opinion from behavioural evidence, identify the cheapest next test, and reach the same unambiguous disposition or point to the exact judgment they dispute.

Action Ladder

1. Do it now — complete the evidence card for one captured idea and assign a disposition.

Changes my mind: The method repeatedly advances ideas without costly beneficiary behaviour, or independent reviewers cannot reproduce the disposition from the evidence card.

Retrieval

Retrieve this method before meaningful build, distribution, hiring, fundraising, or technology commitments are made for an idea.

Version delta: The former staged template is now a source-independent evaluation method centred on behavioural evidence, context-declared thresholds, falsifying tests, and explicit dispositions.

Sources

• Paul Graham, “How to Get Startup Ideas” — supports beginning with problems founders understand and warns against forcing ideas to sound novel.
• Steve Blank, “Build, Measure, Learn?” — supports customer discovery and testing hypotheses before treating execution activity as learning.

Context

• depends-on Idea Capture — begin with a traceable friction record rather than a polished pitch.
• pairs-with Model Selection — choose the operating model only after the idea earns an advance decision.
• pairs-with One-Page Plan — carry supported facts and explicit assumptions into venture planning.
• risk-governed-by Risk Management — bound downside before capital or irreversible effort increases.
• depends-on Purpose — keep the disposition accountable to the beneficiary and value at stake.

Questions

Next question: What would the beneficiary have to do—not say—for you to change today’s disposition?

• Which critical claim has only opinion behind it?
• Which test could change the disposition with the least cost and harm?