The problem
Robert Cooper’s Stage-Gate, introduced in 1990, solved a real problem: without a forced decision point, a new-product project drifts on momentum long after the case for it has quietly stopped being true. The gate — a defined moment where someone actually decides go, kill, hold, or recycle — was the discipline. It worked, and companies that had nothing before it were right to adopt it.
What it became, in a lot of the rooms it was installed in, was something else: a binder. A status deck due the Tuesday before the gate meeting. A scramble to assemble a deliverable that existed mainly to be reviewed, not to be used. None of that was Cooper’s idea — it was a reasonable answer to a real constraint in 1990, when the analysis behind a gate genuinely took weeks of analyst time, worth managing formally. A process built to manage expensive analysis doesn’t get lighter just because the analysis it’s managing got cheap. It gets emptier.
A gate after every stage is stage-gate; a commitment only where money commits is the lighter thing.
Two levels, not gates: a check is quick, self-serve, recorded but never approved — one person, minutes. A commitment is where real money or credibility changes state. The product already runs one check the way this distinction intends: before an AI reference search that costs real money, six wording checks run — free, instant, in your browser — and flag a market description that’s too short or too broad to search well. Proceed when satisfied. Nobody approves it. That’s the whole shape of a check.
Placed by cost, never by sequence
There is deliberately nothing between the first three questions of the journey — is the need real, is it worth it, can we win it. Moving from one analysis question to the next spends nothing but time, so nothing sits there: no gate, no sign-off, no “proceed to stage two.” Checks and commitments only appear where a step would spend something genuinely hard to get back — funding R&D, booking capacity, opening a customer qualification, launching. Sequence isn’t the trigger. Cost is.
The posture that falls out of this, stated the way it’s meant to be said to a buyer who already runs a formal process: no new process. Nothing here invents a meeting or an approval — it arms the decisions your company already makes. If you run Stage-Gate, this feeds it. If you don’t, you’ll never need to. Not a replacement — the tool does no approvals or governance, so it structurally can’t be one. A standing case, continuously current, that turns the gate meeting into a fifteen-minute read instead of the reason the deliverable-scramble existed in the first place.
Sources
This is one piece of the same trust story behind every AI feature in this tool — you always see exactly what would be sent, and the math is never something the AI invents.
Try the Market Size Calculator →