The Go/No-Go decision framework
Borrowed from aerospace, sharpened for founders. The 7 criteria, the meeting structure, the common mistakes, and how to apply it to startup launches and pivot decisions.
What is a Go/No-Go decision?
A forced choice between committing resources and stopping, made before the commitment rather than after it. It comes from NASA mission control, where any single controller could halt a launch. Three outcomes are allowed: GO, NO-GO, or WAIT.
- Seven criteria, scored 1 to 5. Three of them are critical and are not compensated for by strength elsewhere: is the problem real, does a market exist, will anyone actually pay
- GO needs an average of 4.0 or better AND every critical criterion at 4 or better. A critical sitting at 3 caps you at WAIT; a critical at 1 or 2 is a NO-GO whatever the average says
- An average below 3.0 is a NO-GO regardless of how the rest scored
- WAIT counts only with a named experiment and a re-decision date. "Let us give it another month" is a hidden NO-GO that costs a month of runway and decides nothing
- Run it before committing resources, not after: once you have started building, sunk cost makes saying NO-GO almost impossible
Where Go/No-Go comes from
The Go/No-Go decision was formalized by NASA during the Apollo program. Before each major sequence — launch, lunar descent, re-entry — Mission Control polled every system controller in turn. Booster: GO. Guidance: GO. Surgeon: GO. A single NO-GO halted the sequence. No discussion, no overrides. Every voice could veto the next minute.
The framework spread to aerospace at large, then to project management, then to corporate stage-gates. It became a standard tool for one specific reason: it forces a binary decision in moments where humans naturally drift toward "let's wait and see."
For founders, that drift is the most expensive habit you can have. Most startups don't fail from one bad decision — they fail from a hundred un-made decisions. Go/No-Go is the framework that makes you decide.
When to run a Go/No-Go
Anytime the cost of being wrong outweighs the cost of running the meeting. In practice that means before:
- → Starting a new startup or product line
- → Building a major feature (>2 weeks of engineering)
- → Entering a new market (geography, vertical, or segment)
- → Raising a round of funding
- → Hiring a senior team member (>$100K commitment)
- → Pivoting product, customer, or pricing
- → Signing a multi-year contract or partnership
For smaller decisions, the framework is overkill. The bar is: is this commitment something I'll regret if it fails? If yes, run Go/No-Go.
The 7 criteria
Each criterion gets a 1–5 score. GO needs an average of 4.0 or better and every critical criterion at 4 or better — critical means it is not compensated for by strength elsewhere. A critical at 3 caps you at WAIT; a critical at 1–2 is a NO-GO whatever the average says. Below 3.0 average is a NO-GO. High-weight criteria can be partially offset by the others, and medium-weight criteria are tiebreakers when the rest are borderline.
Is the problem real?
CriticalSpecific people, with the problem in the last 30 days, who can describe it without prompting and have tried to solve it. If you can't name 5 such people, the problem isn't real enough.
Does a market exist?
CriticalTAM big enough that 1% gives you a real business. SAM reachable enough that the first 100 customers are plausible. Disqualifying signals: saturated, declining, or hostile to new entrants.
Will customers pay for your solution?
CriticalNot "would they?" — will they. Pre-orders, deposits, calendar bookings, signed LOIs. Verbal interest doesn't count. If no one has put skin in the game, you're measuring curiosity, not demand.
Does your team have the right capabilities?
HighDomain expertise, technical execution, distribution. Missing pieces aren't fatal but must be named. "We'll figure it out" is a NO-GO disguised as confidence.
Do the unit economics work?
HighCAC, LTV, gross margin, payback period. You don't need precision at the idea stage — but you need a defensible thesis. "We'll charge $X because competitors charge $X" without margin analysis = not a thesis.
Do you have enough runway?
HighMonths of cash divided by realistic time-to-validation. If validating this idea takes longer than your runway can sustain, the answer is either fundraise first or kill the idea. Don't start what you can't finish.
What are the disqualifying risks?
MediumRegulatory, platform-dependency (one customer or one channel = 90% of revenue), legal exposure, single-point-of-failure technology. Not all risks are deal-breakers, but unaddressed disqualifying risks turn GO into NO-GO mid-execution.
Applying that rule by hand is where it usually goes wrong: the veto on a critical criterion flips more verdicts than the average does. Score the seven criteria here and you get the verdict plus the one sentence that matters — which criterion decided it, and why. No signup; you leave with a filled-in decision record.
The Go/No-Go meeting
Total length: 60–75 minutes. Attendees: 3–5 people who can challenge each other. Avoid stacking the room with people who report to the founder — they'll vote with their job security, not their judgment.
- Pre-meeting prep — 1–2 hours each. Everyone reads the same evidence packet, so the meeting decides rather than discovers.
- Round 1: each criterion, each voice — 45 minutes. Silent scoring 1–5 before anyone speaks, then discussion. Disagreement is the signal you came for.
- Round 2: the decision — 15 minutes. Aggregate against a threshold set beforehand, so the outcome cannot move the goalposts.
- Round 3: commit or kill — 5 minutes. GO names owner, budget and date. WAIT names the experiment and the re-decision date. NO-GO gets a postmortem.
Roles, facilitation, the traps that turn the meeting back into a discussion, and a minute-by-minute agenda are on the dedicated page: the Go/No-Go meeting agenda.
5 common Go/No-Go mistakes
These are the patterns that turn the framework from a decision tool back into a discussion that resolves nothing.
✗ Letting the loudest voice decide
Most founders default to whoever has highest conviction. That biases toward action and against killing bad ideas.
✓ Instead: Silent scoring before discussion. Each criterion gets numerical scores from each person before anyone speaks. Reveals disagreement that polite discussion would hide.
✗ "Let's give it another month and see"
This is a hidden NO-GO. You commit resources without committing belief, and a month later you're still in the same place — minus a month of runway.
✓ Instead: WAIT must come with a specific experiment and a re-decision date. If you can't name what you're testing and when, the answer is NO-GO.
✗ Skipping unit economics at the idea stage
"It's too early to model" is the most expensive sentence in startups. Most ideas die here when you actually do the math, and you save 6–12 months.
✓ Instead: Build a back-of-envelope unit economics model. CAC, LTV, gross margin, payback. Numbers don't need to be precise — they need to be defensible.
✗ Running Go/No-Go after you've already started building
Sunk-cost bias makes saying NO-GO almost impossible. You'll rationalize whatever the data says into GO.
✓ Instead: Run Go/No-Go BEFORE committing significant resources. This is the entire point — the whole purpose is to prevent the sunk-cost trap.
✗ Treating Go/No-Go as a one-time gate
You GO, conditions change, but no one re-decides. The original GO becomes inertia.
✓ Instead: Set a re-decision trigger: a milestone, a deadline, or a specific signal (e.g., "if we don't hit X by Y, we re-run Go/No-Go"). Build the kill-switch in.
The 15-minute Go/No-Go session
We named our company GoNoGo because the framework is what we built. A 15-minute voice session walks you through all 7 criteria — voice intake, market sizing, competitor map, synthetic focus group, unit-economics check, risk surface — and produces a written GO/WAIT/NO-GO verdict with reasoning.
It doesn't replace a real Go/No-Go meeting with co-founders for major decisions. It does what most founders skip: a structured first pass on the idea before you spend weeks scheduling that meeting.
The output is the meeting prep packet. Bring it to your real Go/No-Go and the meeting takes 30 minutes instead of 3 hours.
15 min · up to 20 reports · Your decision packet, ready for the real meeting
Frequently asked questions
What does Go/No-Go actually mean?+
When should I run a Go/No-Go decision?+
What's the difference between Go/No-Go and a regular planning meeting?+
Can I run a Go/No-Go alone as a solo founder?+
What should I do if the score is borderline?+
Go/No-Go deep dives
Related guides