A familiar first day on a new product: the team writes the audience. Pete, 20, designer, lives in a large city, likes technology and coffee. It looks like progress and it is beautifully formatted, and it will not help anyone decide what to build.
Pete will not buy the product because he is twenty and works in design. He will reach for it when something happens to him that he needs handled. The situation is the useful part, and the persona quietly discards it.
The job is the unit of work, not the person
People hire a taxi to be somewhere on time. They hire a VPN to reach a service or to work safely on a network they do not trust. They hire an expense tracker to stop guessing where the salary went.
In each case the trigger is a moment, not an identity. Describing the moment tells you what the product must do, what the first screen must say, and what the ad has to promise. Describing the person tells you which stock photo to use.
When [situation], I want [motivation], so that [result]. Everything downstream gets easier once that sentence is honest.
One product, several different jobs
The same product is regularly hired by people who share nothing demographically. A designer, an agency owner and a parent on leave can all use one VPN, and in a given moment they are hiring it for the same job.
The reverse trips teams up more often. Two twenty-five-year-old designers can be entirely different markets. One needs access to services for work and will pay for reliability. The other wants to watch something and will leave the moment it costs more than the alternative. Same persona, different trigger, different requirements, different reason to pay.
This is why segmenting by attributes produces roadmaps that please everyone slightly. Segmenting by job produces a product that is clearly worth paying for to a smaller group, which is the only kind that grows.
The questions that replace the persona
Instead of asking who the audience is, I want answers to five things. They are harder to fill in, which is the point: a blank here is a real gap, not a missing detail.
- At what moment does someone start needing this?
- What are they trying to get done?
- How are they handling it today?
- What specifically fails about that current solution?
- What result would they pay for?
Seven days, and none of them start with a persona
If the constraint is a week, the sequence below is the one I would keep. It spends the first four days on decisions and only then opens a design tool, because building is the expensive way to discover you were solving the wrong job.
Day one, name the job: the situation, the motivation and the result, in one sentence. Day two, the business model: who pays, for which recurring result, how often, and why they keep paying. Day three, the funnel end to end: where someone hears about it, where they arrive, the first action, the moment of value, the payment, the reason to return. If that path cannot be drawn as a single line, the product is still fog.
Day four, choose one scenario. Not twenty-five features labelled as a minimum viable product. One path someone can complete alone, without a call with the founder and without a fourteen-minute explanation. Day five, build that path and nothing beside it. Day six, instrument it: how many arrived, how many started, how many finished, where they stopped, how many came back. Day seven, launch and check the only thing that matters, which is whether people hire it for the job named on day one.
What day seven is actually for
The launch is not a demonstration that the team was right. It is the first cheap test of the day-one sentence. If people arrive and complete the scenario for a different reason than expected, that is not a failed launch, it is the most valuable information available and it arrived in a week rather than a quarter.
The failure mode is a launch with no instrumentation, because then the team has opinions instead of an answer, and the loudest opinion wins.
A product does not begin with a logo, an interface or a repository. It begins with knowing which job someone will trust it with instead of what they use now. Without that answer, what exists is a set of features with a good name.