Failure mode one: three conversations and a conclusion
Three people, all of whom were easy to reach, all of whom said something encouraging. You now have a belief with a citation attached, which is more dangerous than a belief with none.
Failure mode two: research as deferral
Forty conversations and counting. Every new interview adds nuance and no new decision. The research has stopped being a way to find out and started being a way to avoid committing.
The tell is that you cannot say what the next five conversations would need to show in order to change your mind.
The saturation signal
Stop when new conversations stop producing new categories. Not new details — new categories. When interview nine describes a variant of a pattern you first heard in interview four, and interview ten does too, you have your shape.
For a focused decision with a tight profile, that usually lands somewhere between eight and fifteen conversations. If you're at thirty and still hearing new categories, your profile is too broad, not your sample too small.
Set the threshold before you start
The step that matters is the fourth, and it only works because of the second. A threshold set after the data arrives is not a threshold.
- 1Write the decision you're making.
- 2Write what you'd need to see to go each way — a count, not a feeling. 'Six of ten have already paid for something in this category.'
- 3Run the conversations.
- 4Compare against the threshold you wrote before you had a preference.
The honest stopping rule
Stop when the next conversation would not change what you do on Monday. If you cannot imagine an answer that would alter the plan, you already have your decision and you're collecting comfort.