A pleasant meeting is not signal
You will end many conversations feeling good without being able to say what you learned. They were engaged, said it sounded relevant, agreed to stay in touch. That is not signal, that is a pleasant meeting. Filter out three categories of bad data: compliments ("this is really interesting"), hypothetical statements ("I would definitely use something like that"), and wishlists ("it would be great if it also did X"). None of these tell you whether the person will buy or whether your ICP hypothesis is correct.
Find the edges of your hypothesis, not confirmation of it. If you were not a little uncomfortable with at least one question you asked, the conversation was too safe.
Complete this after every conversation
How many times did you ask for an opinion or reaction to your product rather than a fact about their experience? If above zero, review your question list before the next call. Rewrite every question that starts with "what do you think" or "would you consider." Replace with "what did you do" or "walk me through what happened when." Opinion questions produce opinions, you need facts.
How many concrete examples from the past did the other person give you? If fewer than three, the conversation stayed at the level of generality. Start every question with "tell me about a time" or "walk me through what happened when." If you are still getting abstract answers, push back: "can you give me a specific example of when that happened?"
If you identified a disqualification signal during this conversation, log it immediately, see Step Three of the main framework for the log format.
Did you ask at least one question you were slightly reluctant to ask, about budget, timeline, decision-making authority, or what would stop them moving forward? If no, write the question you were reluctant to ask at the top of your notes before the next call. Commit to asking it within the first 15 minutes, before the conversation builds enough warmth to make it feel impolite.
What concrete next step did the other person agree to? No next step, an introduction, a follow-up with an agenda, a second stakeholder, or interest in a paid pilot or Letter of Intent (LOI). If "no next step" is your consistent outcome, you are collecting conversations, not building pipeline. End every call with "based on what we have discussed, is there a useful next step?" A clear no is more useful than an open-ended close.
If you cannot complete this, the conversation confirmed what you already believed. If three consecutive calls produce nothing surprising, your questions are too safe. Write three questions that could actively disprove your current ICP hypothesis and use at least one in every subsequent call.
The rollup check
After 10 conversations, review the completed scorecards and look for:
For pattern analysis across transcripts, use Prompt 3 in Addendum D. If your interviews have been primarily non-switchers, use Prompt 4 instead.