JTBD analyzer
What it does
Jobs to be Done treats a purchase as a hire. Someone is struggling with something and brings a product in to make progress. The analysis works backwards from the moment of switching: what pushed them off the old way, what pulled them toward the new one, what anxiety held them back, what habit kept them in place. Those are the four forces. Bob Moesta and Clayton Christensen laid out the version most product teams use in Competing Against Luck, and the same work goes by jobs theory or switch interviewing.
This skill runs that coding pass over cleaned transcripts or the rough notes you typed during the session itself. It fixes each person's timeline first, since the forces only read correctly once the sequence of events is settled, then attaches each force to the event that produced it. Every extraction carries its quote and the person it came from. Ask Claude cold and you tend to get the framework's vocabulary applied to your own summary. This one stays on the words that were said in the room.
What you get back
- A job statement for each person, in the situation-and-motivation form Moesta uses, with the transcript line that produced it.
- The four forces per interview, each tied to a dated event on that person's timeline.
- The struggling moment and the switching trigger, quoted in full.
- A timeline per person covering first thought, passive looking, active looking and the decision itself.
- Cross-interview patterns: which forces recur across the set, and which appeared once.
- The places where the transcript is too thin to code, marked as follow-ups instead of filled in.
How to use it
Example Prompts:
"Run a JTBD analysis on these five churn interviews and show me the forces person by person."
"What was the struggling moment in this transcript, and when did she start actively looking?"
"Compare the push forces across the whole set and flag the ones only one person raised."
What it won't do
It won't tell you whether the job it found is worth building for. Sizing demand is a separate exercise. It also depends on the interview having gone somewhere near a timeline. If nobody asked when the person first started thinking about a change, there is no sequence to code, and the forces come back as opinions with labels attached.
Questions
Which school of JTBD does this follow?
The Moesta and Christensen line, built on switch interviews and the four forces. Tony Ulwick's outcome-driven innovation is a separate school with its own outcome statements and quantitative sizing, and this skill does not produce those.
Can it analyse interviews that weren't run as JTBD interviews?
Yes, with a caveat. It will code whatever forces are present in generic discovery interviews, and it will tell you where the timeline is missing. Sessions that never established when the decision happened produce the weakest output.
Do transcripts need cleaning first?
Not strictly. Raw exports with broken speaker labels slow it down and make quote attribution less reliable, so running the transcript cleaner first is worth the extra step on anything that came out of an automatic transcription service.
