bijoycherian/jtbd-skills

A Claude agent workflow that helps you understand what customers are trying to accomplish, identify underserved needs, and decide where to focus.
Great products start with understanding people deeply. As Clayton Christensen put it, we “hire” products to help us get a job done.
But understanding customers is only the beginning. You still need to decide which needs to address first.
I built the JTBD Opportunity Landscape agent to help with that decision: create an initial opportunity map from your research, then refine it with new interviews and customer ratings.
The workflow draws on Jobs to Be Done and adapts Tony Ulwick’s Outcome-Driven Innovation (ODI) opportunity landscape and scoring algorithm. It turns customer research into a reviewable job list and an opportunity map, with supporting evidence and a visible basis for each score.
This is an adaptation, not a full ODI implementation: it scores customer jobs and can start with clearly labelled AI estimates before you collect customer ratings.
Next time you have some customer research and priorities are unclear, start with what you have: a few interview notes, support tickets, customer quotes or an existing job list.
Run /jtbd in Claude, add your research, and use the first map to see where the evidence is strong, where the agent made assumptions, and what to validate next.
## How it works
Claude starts with two questions:
What do you want to learn about your customers?
What decision will this inform?
Those answers frame the work. The workflow then moves through five stages:
Listen → Shape → Map (AI estimates) → Validate → Decide
1. Listen. Drop in your interviews, support tickets, or notes. It extracts customer evidence and checks the extracted text against the original sources.
2. Shape. It turns the extracted evidence into jobs that describe what customers are trying to accomplish—not features you could build. You review the framing and confirm the job list.
3. Map (AI estimates). Where customer ratings are missing, the agent estimates how much each job matters and how well it is served today, then plots those scores on the opportunity map, labelled as AI estimates.
4. Validate. Use the pre-filled survey to collect customer ratings and replace estimated scores.
5. Decide. You choose where to focus. Decisions recorded in the workflow link back to supporting quotes, and changes to job scores are logged.
## What you get
An editable opportunity map. Open a job to see its supporting evidence.
Export the scores and evidence, and use the pre-filled survey to replace estimates with ratings from your customers.
Each score is labelled as an initial AI estimate, an AI estimate revised using customer evidence, your judgement, or a survey result. The ranking informs your decision; it doesn’t make it for you.
## Quick start
Claude app — desktop or web
1. Open Customize → Plugins → + Add.
2. Add the marketplace bijoycherian/jtbd-skills.
3. Click Add on JTBD Opportunity Landscape, or select Try in chat to try it first.
4. Start a new chat and type /jtbd.
5. Attach your research—transcripts, tickets, or notes—and say what you want to learn and what decision it will inform. Using real customer data? Read Data and privacy first.