Marcus Andrey Study
Ler em português Library

UX Research · Step 6 · 8 min read

Personas and journeys from data

A persona or a journey map is only as good as the evidence under it. How to go from themes to behavioural segments, how to build a persona and a journey that point back to their sources, how jobs to be done fit in, how to show confidence and gaps, and when to update them.

The problem with most personas isn't that they're made up. It's that nobody can tell which parts are. A persona is a fictional person on purpose: it's easier to design for "Sam" than for "segment B". But when the evidence under each line is invisible, a guess from a workshop and a pattern from twelve interviews look exactly the same. This lesson is about building personas and journeys that show their working.

What you'll be able to do after reading:

  • Go from themes and participants to behavioural segments
  • Build a persona where every line points to its evidence
  • Build a journey map with evidence and confidence for each stage
  • Decide between personas, jobs to be done, or both

1. Artefacts are summaries of evidence

NN/g defines a persona as "a fictional, yet realistic, description of a typical or target user of the product", and is firm about the basis: personas might be made-up people, but they should be based on information about real people (NN/g, Personas Make Users Memorable). The same goes for journey maps: NN/g says the actor's actions "should be grounded in research data", and anticipated scenarios based on assumptions "need to be clearly identified" (NN/g, Journey Mapping 101).

Personas and archetypes and Journey maps, in UX Design Foundations, cover what goes into each. This lesson is about the step in between: how the themes from your synthesis turn into these artefacts, without losing track of where each part came from.


2. From themes to segments

A theme is about the data; a segment is about people. To get from one to the other, look at which participants share which themes.

  1. Make a grid: participants down the side, themes and key behaviours across the top. Mark who shows what.
  2. Look for people who overlap on most of the important themes, not on all of them. NN/g describes exactly this for qualitative personas: look for broad patterns, not perfect matches between individuals (NN/g, 3 Persona Types, 2020).
  3. Name the group by what they do, not by who they are. "The quiet creditor" tells a designer something; "Millennial Maria" doesn't.
  4. Check the differences matter for design. If two groups would get the same design, merge them. NN/g's persona guide says the same: merge personas that are too similar, and drop details that wouldn't change a decision.

Why behaviour and not demographics? NN/g points out that demographics provide no behavioural or attitudinal data and are mostly suited to marketing decisions, which is why demographic "personas" often fail to help design (NN/g, Personas vs. Jobs-to-Be-Done, 2017).


3. A persona that shows its working

The quiet creditor

Qualitative persona · 6 interviews · March
Behaviour

Pays shared bills up front and waits to be paid back.

P2P3P4P54 of 6
Goal

Get the money back without a tense conversation.

P2P3P4P54 of 6
What they do today

Lets small amounts go; asks only when the total gets big.

P2P3P43 of 6
In their words

"It felt like the app was me, nagging Ana."

P4similar: P5
For design

Reminders that speak as the app, on a date the group agreed.

From the theme "asking feels rude"
Not known yet

Whether couples behave the same way: none were in the sample.

0 of 6
A persona whose every line points back to the participants it came from. The name describes a behaviour, there's no stock photo, and the gaps are written down instead of filled in.Diagram: Marcus Andrey, CC BY 4.0. A made-up study.

Each line in the persona carries its evidence: which participants, and how many of the total. Three things to notice:

  • Counts aren't percentages. "4 of 6" is honest; "67% of users" isn't, because six interviews can't estimate a share of your user base. NN/g says it plainly: with qualitative personas you can't say what proportion of your users each persona represents (NN/g, 2020).
  • The gaps are written down. "No couples in the sample" stops someone from applying the persona where it was never tested.
  • The "for design" line comes from an insight, so the persona connects to the decision, not just to the data.

How much research is enough? For qualitative personas, NN/g suggests interviewing 5 to 30 people, as a rolling sample of about five per group, until each new interview brings few new insights. If you need to know how big each segment is, you need statistical personas: qualitative research first, then a survey of at least 100 people, ideally 500 or more, and a statistician to cluster the results. NN/g calls that overkill for most teams (NN/g, 2020). Sample size and margin of error explains why the numbers grow so fast.


4. Personas and jobs to be done

Jobs to be done looks at the same people from another angle: what they "hire" a product to do. Clayton Christensen and colleagues built the theory in Competing Against Luck; User needs and jobs to be done covers the basics and the milkshake story.

NN/g argues the two aren't rivals. A job describes a need shared across many kinds of users; a persona adds the behaviour, context and attitudes that change how each group wants the job done. A professional and a weekend DIYer both have the job "put holes in things"; one cares about durability, the other about price (NN/g, Personas vs. Jobs-to-Be-Done, 2017). Their advice:

  • If your organisation already uses jobs, reference the jobs inside each persona instead of repeating them.
  • If you have thin personas, rewrite their goals as jobs: what is this person trying to get done, and what makes it a success?

For the bill-splitting study, the job is "get even with my flatmates without it becoming a thing". The quiet creditor and the keeper of records share it, and need different help with it.


5. A journey with evidence and confidence

StageWhat they doWhat they think or feelEvidenceConfidence
1 · Pay the billPays the whole electricity bill and adds it to the app"Easier if I just do it"P2, P3, P4, P5High: 4 of 6, said and shown
2 · WaitLets small amounts build up for weeks"It's only five euros"P2, P3, P4Medium: 3 of 6
3 · AskSends a message, late, or turns reminders offAwkward: "like I don't trust them"P3, P4, P5Medium: 3 of 6
4 · Settle upSettles over dinner or on rent day, after checking a spreadsheetReliefP1, P2Low: 2 of 6, told, not seen
A journey built from the same six interviews, with an evidence column and a confidence column. The weak stage is marked as weak, so nobody designs on it as if it were proven.Diagram: Marcus Andrey, CC BY 4.0. A made-up study.

A journey map from research follows one actor through one scenario, stage by stage: what they do, think and feel (NN/g, Journey Mapping 101). To keep it honest, add two columns most templates leave out:

  • Evidence: the participants each cell rests on.
  • Confidence: high when several people said and showed it; low when one or two people told you about it. In the example, "settle up" is low: two people described it, and nobody was observed doing it.

Build the stages from what people actually did, in the order they did it, not from the screens of your product. If the stages match your app's navigation exactly, you've probably mapped the product, not the person. And mark the cells you had to guess, so the next study knows what to look at, ideally through Contextual inquiry and diary studies.


6. Keep them alive, and keep them plain

NN/g recommends treating personas as living documents: update them when the user base or the business changes, not for every new feature, and don't over-invest in visual polish, because beautiful artefacts are expensive to change and so they go stale (NN/g, Personas Are Living Documents, 2023). Two habits help:

  • Put a date and a source list on every persona and map. "6 interviews, March" tells the reader how old and how deep it is.
  • Keep the editable version in your research repository, next to the data it came from, and link to it instead of pasting screenshots into decks.

NN/g also lists uses after the design phase: personas as recruiting criteria for future studies, and as segments for reading analytics (NN/g, Personas Make Users Memorable). Both are good tests: if a persona can't be turned into a screener question, it probably isn't specific enough.


7. Common mistakes

  • A persona built from demographics, with a stock photo, and no behaviour.
  • Percentages from six interviews.
  • Every line written with the same confidence, from guesses and from evidence alike.
  • A journey that follows the app's screens instead of the person's steps.
  • Too many personas to remember.
  • A beautiful poster from three years ago that nobody dares to change.
  • Workshop guesses ("proto personas") presented as research.

8. Vocabulary

Term Meaning
Segment A group of people who share behaviours and needs that matter for design
Qualitative persona A persona built from small-sample research such as interviews
Statistical persona A persona built from clustering a large survey, after qualitative research
Proto persona A persona from the team's assumptions, not from new research
Job to be done The progress a person is trying to make, which they "hire" a product for
Actor The person whose journey a map follows
Scenario The situation and goal a journey map covers
Confidence How strongly the evidence supports a line or a cell

9. Try it (no tools needed)

  1. Take a persona your team uses, or one from a public case study.
  2. For each line, write where it came from: which research, how many people. Mark the lines with no source.
  3. Rewrite one goal as a job: "When ___, I want to ___, so I can ___."

Reflection questions: Which line of that persona has influenced a decision? Was it one with evidence?


10. If you already practise

  • Add evidence and confidence to your existing personas and maps before you make new ones. It's cheaper, and it shows where the next study should go.
  • Turn each persona into two or three screener questions, and use them in your next recruitment.
  • Review personas once a year, or when the user base changes, with a short changelog.
  • Present the gaps out loud. "We don't know how couples do this" is a research roadmap, not a weakness.

Checklist

Learn more

Books, courses and videos to go deeper

Book links go to Amazon. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.

Personas and archetypes · Journey maps · User needs and jobs to be done · Thematic analysis · Screeners and sampling

Saved only on this device.

↑ ↓ to move · Enter to open · Esc to close