Jobs, not demographics
Lesson 7 of 24, level 01Index
Marketing hands you a persona: “Ananya, 29, urban professional, loves brunch and fintech”. You have to build something on Monday and this tells you nothing.
People hire products to do a job
A demographic describes who someone is. A job describes what they are trying to get done, in what situation, and what they will consider progress. Only the second one tells you what to build.
The format that keeps you honest is: when [situation], I want to [motivation], so that [expected outcome]. The third clause is where the value is, and it is the one people skip.
- Situation
- The trigger. When it happens, in the real world.
- Motivation
- What they are trying to do right then.
- Outcome
- How they will know it worked. This is what you actually build for.
A useful job statement helps you choose what to build. A demographic label rarely tells you what happened just before someone opened the product or what successful progress would look like.
Use your chosen course project throughout. The additional examples below are fictional practice cases; transfer the method to your own evidence.
Reconstruct a moment
Choose one recent episode. Record the trigger, the environment, what the person tried, and what happened next. Someone reconciling invoices before a deadline has different constraints from the same person exploring software on a quiet afternoon. Describe that context before assigning a persona.
Separate progress from a feature
“I need a dashboard” is a proposed solution. Ask what decision the dashboard enables and what happens without it. “Know which invoices need chasing before Friday” leaves room for an email, a list, or a service. This makes the job useful when comparing options instead of merely restating the feature request.
Keep competing explanations
A workaround is evidence that someone acted, but not proof of why. Note the alternative explanations and ask a follow-up question that could distinguish them. Avoid converting a single vivid story into a universal need. Keep the original words beside your interpretation so a teammate can challenge the jump.
From a person to a moment
- Trigger: What changed?
- Progress: What must improve?
- Evidence: What actually happened?
How McDonald's handled it
McDonald's
McDonald's wanted to sell more milkshakes and had tried improving flavour and thickness based on customer surveys, with no effect.
Clayton Christensen's team watched who bought them and found nearly half were sold before 8am, to solo commuters. The job was not dessert, it was a one handed breakfast that lasts a long commute and does not leave crumbs. The competition was bananas and bagels, not other milkshakes.
Segment by job and your competitive set changes completely. Segment by demographic and you improve the flavour of something nobody was buying for its flavour.
Doing it with AI, and where it breaks
Feed a model your support transcripts and ask it to extract jobs in the when/want/so-that format, one per transcript, quoting the line that supports each.
It will generate elegant jobs that no quote supports, because the format is easy to write and hard to falsify. Delete every job you cannot attach a real sentence to. Usually half.
Your AI workbench
Start with your own notes or clearly labelled practice data. Remove private details before sharing. Replace the placeholders, run the prompt in your chosen AI tool, and keep the output beside its source.
Using the anonymised episode below, extract trigger, desired progress, workaround, anxiety, and missing evidence. Cite a source sentence for each field. Propose two competing job statements and one neutral interview question that distinguishes them. Do not invent motivations or population sizes. Episode: [paste your notes].
Before you use the output
- Every motivation has a source or is marked a hypothesis
- The job does not prescribe a feature
- A recent event supports the statement
Stuck? Try this next
If AI produces an elegant but vague sentence, ask which actual decision it would change. Rewrite it using the user's situation, a concrete action, and an observable result.
Keep a brief AI log: input used, useful output, what you checked, and what you rejected. The decision remains yours.
Build it
Three evidenced jobs, and a written argument for the one to build for.
Checkpoint
If you did the build, these take two minutes. If you cannot answer one of them, that is the part to go back to.
Moving on marks this lesson complete. Finish the build first, it is the part that counts.