Jobs to Be Done and Behavior Data — Finding What Users Are Really Trying to Do
Jobs to be done asks what progress a customer is trying to make. Behavior data shows what they do on the way. Learn how to combine interviews with session recordings and journeys to uncover the real jobs your product is hired for.
Short answer: jobs to be done treats a purchase as "hiring" a product to make progress in a particular situation. Interviews reveal the job and the circumstances. Behavior data shows which jobs people actually attempt in your product, in what order, and where they fail. Use recordings to see workarounds and unexpected uses, which often point to a job you had not designed for.
Table of Contents
- The Idea
- What Interviews Give You
- What Behavior Gives You
- Finding Jobs in Session Data
- From Jobs to Decisions
- Key Takeaways
The Idea
People do not want a product for its own sake. They want to get something done. The framework's usual form is a statement:
When [situation], I want to [motivation], so I can [outcome].
Examples:
- "When conversion drops after a release, I want to see what users experienced, so I can fix it before it costs a week."
- "When a customer says something is broken, I want to see their session, so I can answer without asking them to reproduce it."
The same product is hired for different jobs by different people. A session replay tool is hired by an engineer to debug, a marketer to improve a landing page, and a founder to understand new users.
The value of the framing is that it shifts attention from features and customer demographics to situations and outcomes. Two customers with nothing else in common can share a job.
What Interviews Give You
Jobs are discovered mainly by talking to people about a real, recent decision.
- The trigger. What happened that made them look for something?
- What they used before. A competitor, a spreadsheet, or nothing.
- What they considered and rejected.
- What almost stopped them.
- What "done" looks like to them.
This is context no analytics can supply (feedback and behavior).
What Behavior Gives You
Interviews are few and reconstructed from memory. Behavior data is plentiful and direct.
| Question | Source |
|---|---|
| Which jobs do people attempt most? | First actions after login; journeys |
| In what order? | Journeys and funnels |
| Where does a job fail? | Funnel drop-off, recordings |
| What workarounds exist? | Recordings |
| What do people use the product for that we did not intend? | Recordings, unexpected paths |
| Which jobs retain customers? | Cohorts by first action |
| How long does a job take? | Time between events |
Finding Jobs in Session Data
1. The first action after arriving
What a returning user does first is the job they came to do. If most sessions begin by going to the same report, that report is the job. Group sessions by their first meaningful action.
2. Repeated sequences
A sequence of five steps performed again and again by the same users is a job your product supports clumsily. It is a candidate for a single action.
3. Workarounds
- Exporting data to finish the task elsewhere.
- Copying values between two tabs of your product.
- Using a text field to store something it was not meant for.
- Creating dummy records to group things.
Each workaround is a job with no direct support. These are the strongest signals for a roadmap.
4. Unexpected paths
A journeys report from a key page often shows a route nobody modelled. DeepSync's docs note that journeys frequently reveal a route you never thought to turn into a funnel (journeys).
5. Abandoned attempts
A user opens a feature, tries something, fails, and leaves. They had a job. The product did not do it.
6. Entry points
The page someone lands on, and the search query that brought them, hint at the job. A visitor landing on a comparison page has a different job from one landing on a setup guide (search queries by landing page).
Tag sessions so jobs can be compared:
deepsync("track", "job_started", { job: "debug_issue" });Then interview a few of the people you watched. "I noticed you export this report every Monday. What happens to it next?" is a question that could only come from behavior, and its answer could only come from a conversation.
From Jobs to Decisions
| Finding | Decision |
|---|---|
| One job dominates first actions | Make it the default screen |
| A job takes eight steps | Reduce to two |
| A common workaround | Build direct support |
| A job nobody attempts | Stop investing in that feature, or check whether it is discoverable |
| Different jobs by role | Different onboarding paths |
| Marketing describes a job users do not do | Change the positioning |
| A job retains customers far better | Guide new users toward it |
Jobs also sharpen positioning. Describing your product by the job it does, in the words customers used in interviews, is clearer than a list of features, and it is what search engines and AI assistants pick up (writing for citation).
A caution: behavior shows the jobs people attempt in your product. It cannot show the jobs they never tried because they assumed you could not help. For those, you need interviews with people who chose something else.
Look for the export button
What users take out of your product, and what they do with it next, is frequently the job you have not finished for them.
Key Takeaways
- A job is the progress a customer wants in a situation.
- Interviews give context; behavior gives frequency, order, and failure points.
- Workarounds and repeated sequences reveal unsupported jobs.
- Use behavior to choose who to interview and what to ask.
Conclusion
Customers hire your product for reasons that are often different from the ones you built it for. Their sessions show what they are trying to do. Ask them why, and build for that.
Frequently Asked Questions
Related articles
Stay in the loop
Get the latest insights on product analytics and user behavior delivered to your inbox.



