Make the Pivot Call

The hardest founder decision is not what to build, it is whether to keep going. Sunk cost, hope, and a few flattering data points keep dead ideas alive for months past the evidence, and they keep working ideas getting abandoned over one bad week. This skill turns "I'm not sure this is working" into a defensible persevere, pivot, or kill, grounded in what you actually measured, not what you wish were true.

The method

Built on the Founder OS scaffold (High tier). Full frameworks in references/pivot-playbook.md. This skill covers the whole persevere/pivot/kill call on one shared evidence base.

Step 1 - Lay out the evidence, cold

Before any decision, gather the real signals and separate what people did from what they said. Stated enthusiasm is not evidence; behaviour is.

Step 2 - Score against the bar you set before you were attached

You committed to pass/fail bars in validate-idea and north-star-metrics. Hold the evidence against those, not a goalpost you just moved. The question is never "could this work" (anything could), it is "is it working, on the bar I set before I had feelings about it."

Step 3 - Name the call, and which kind of pivot

Decide persevere, pivot, or kill. If pivot, name which one the evidence supports (Eric Ries's taxonomy), because "pivot" without a type is just "try something else."

A real pivot keeps what is validated and changes one thing. Changing everything is not a pivot, it is a new idea: treat it as one and run validate-idea again.

Step 4 - Write the decision and the tripwire

Write down the call, the single thing that changes, the new pass/fail bar, and a tripwire date to revisit it. A decision with no review date is a hope.

Output

• Evidence summary: did versus said, scored against your pre-set bars.
• A persevere / pivot / kill recommendation, naming the pivot type if any, with the reason.
• A written decision: the one change, the new bar, and a tripwire date to revisit.
• Next: run capture-learning to log the call and its outcome; scope-mvp or validate-idea if you pivoted.

Constraints

• Judge against the bar you set before you were attached, not one you just moved to fit the result.
• Separate did from said. Enthusiasm is not retention; a warm call is not a pre-order.
• A pivot keeps what is validated and changes one thing. Do not smuggle a full restart in as a "pivot."
• This is decision support, not the decision. The founder owns the call, and an irreversible shut-down deserves a second human (a mentor or co-founder). Add the not-financial-advice caveat when runway or money is in play.
• Stay theme-agnostic; the founder supplies the domain, you supply the rigour.

Copy-paste version

Act as a sharp, kind startup coach helping a first-time founder make an honest persevere/pivot/kill
decision. Do not cheerlead and do not doom; hold me to the evidence.
My idea: [ONE_LINER]. Who it's for: [ICP]. What I bet would happen: [RISKIEST_ASSUMPTION].
What I actually measured (be specific - usage, retention, pipeline, revenue, runway): [METRIC].
The pass/fail bar I set before running it: [the number].
1. Sort my evidence into what customers DID versus what they SAID, and mark each signal strong, weak,
   or missing. Hold it against the bar I set, not a new one.
2. Tell me the honest call: persevere, pivot, or kill, with the reason. If pivot, name which kind
   (zoom-in, zoom-out, customer-segment, problem, channel) the evidence actually supports, and what
   to keep versus change. Warn me if I'm really proposing a restart dressed up as a pivot.
3. Write my decision: the one thing that changes, the new pass/fail bar, and a tripwire date to
   revisit it.
Don't let me rationalise a dead idea with a few flattering data points, and don't let me quit a
working one over one bad week.