Persona Development SOP

A persona development SOP turns "our customer is everyone" into one named, specific person your whole team can picture, and you can build a solid first version in two to three days. Alan Cooper introduced personas for software design in The Inmates Are Running the Asylum, and since then most startups have made them wrong: a stock photo, an age range and a favourite coffee order. This SOP builds a persona around behaviour, triggers and money, then tests whether it predicts what real people do.

What is the persona development SOP and when should you use it?

A persona is a one-page profile of a fictional customer who stands for a real group you have researched. A good startup persona answers five questions: what triggers the need, what they do about it today, what that costs, who decides on the purchase and where you can reach them. Demographics matter only when they change behaviour.

Use this SOP:

• On day one, as a proto-persona that lists your assumptions, a format Jeff Gothelf describes in Lean UX.
• After 10 or more interviews, to replace those assumptions with evidence.
• When your team or your ads target different people, and you need one shared picture.

A persona describes who the customer is. To map what they think and feel at one moment, use customer empathy mapping. To map their steps over time, use customer journey mapping.

What do you need before you start?

• Your current guess about the customer, written down honestly.
• 10 or more interview transcripts or notes from the customer interview SOP. Without them you can only write a proto-persona.
• Survey data, if you have it, to check how common each trait is.
• The template below in Notion, Google Sheets or Miro.

How do you run it, step by step?

1. Write the proto-persona in one hour

Fill in the template from your own assumptions, alone or with your co-founder. Mark every field as "guess". This makes your beliefs visible so the research can prove or break them.

2. Pull behaviour from your research

Go through your notes and copy every fact about triggers, current workarounds, money spent, decision process and channels onto sticky notes or rows. Ignore personality adjectives unless several people showed the same trait in their actions.

3. Find the behaviour clusters

Group people by what they do. Two 45-year-old shop owners who handle stock in opposite ways belong to different clusters. Pick the cluster with the strongest problem and the clearest spending; that becomes your lead persona. Keep at most one other as a backup.

4. Rewrite every field with evidence

Replace each guess with a fact and note the source: "Pays a freelancer €200 a month for bookkeeping (interviews 3, 7, 9, 12)". Keep a field as "guess" if you have no evidence. Add two or three real quotes.

5. Run the reachability test

Try to find 20 people who match the persona in one week through the channel the persona lists: a LinkedIn search, a subreddit, a trade association, a local meetup. If you cannot, the persona is wrong or the channel is. A persona you cannot reach is useless for a startup.

6. Run the prediction test

Before your next 10 conversations with new matching people, write down what the persona predicts they will say about their workaround and spend. Compare after the calls. 7 or more hits out of 10 means the persona works.

7. Share it and date it

Put the persona where the team sees it, with a "last checked" date. Update it after every 10 new conversations.

Template you can copy

PERSONA: [Name + one-line label, e.g. "Busy Bookkeeper Bea"]
Based on: [# interviews, # survey answers]   Last checked: [date]

Situation      Role / life stage, only the facts that change behaviour
Trigger        The event that makes the problem urgent
Job to be done "When [situation], I want to [motivation],
                so I can [outcome]."
Today's fix    Tools, people and workarounds used now
Cost today     Time per week / money per month spent on the fix
Decision       Who chooses, who pays, who can block
Budget signal  What they already pay for in this area
Channels       Where we can reach 20 of them in a week
Objections     Top 2 reasons they would say no
Quotes         "..." (interview #)   "..." (interview #)

Evidence status per field: [evidence / guess]
Prediction test: [# hits] out of 10

Worked example: a scheduling tool for dog groomers

This is an illustrative case with invented numbers. A founder's proto-persona reads: "Women aged 25 to 40 who run a grooming salon and love dogs." After 14 interviews, the facts tell a sharper story.

Age and love of dogs predict nothing. What splits the group is how they work: 9 of 14 are mobile groomers working from a van, and they lose 3 to 5 hours a week reshuffling the day's driving order when clients cancel. The 5 salon owners have staff who handle bookings and barely mention scheduling. Mobile groomers already pay for trip-planning apps or an assistant, between €30 and €150 a month.

The new persona, "Mobile Groomer Mia", lists a trigger (a same-day cancellation), today's fix (WhatsApp plus a paper map), cost (4 hours a week) and channels (two Facebook groups and a grooming trade association). She finds 26 matching groomers in five days. In 10 new calls, the persona predicts the workaround and spend correctly 8 times.

What should you measure, and when do you stop?

Stop building and start using the persona when it passes all four. If the prediction test fails twice, go back to step 3 and split your clusters differently.

Which mistakes should you avoid?

• Leading with demographics. Age and gender rarely explain why one person buys and another does not. Start with triggers and spending.
• Creating five personas. Early startups serve one. Extra personas split your messaging and your product.
• Inventing details to make it feel real. Hobbies and pet names without evidence send the team chasing fiction.
• Never testing it. A persona that cannot predict behaviour or be reached is a poster.
• Freezing it. Update it every 10 conversations, and date every change.

How can AI speed this up?

AI tools write a polished persona in seconds, and that is the danger: they fill gaps with plausible inventions. Use the free user persona generator for the proto-persona in step 1, then paste your anonymised interview notes into an AI assistant and ask it to fill each field only with facts it can quote, marking the rest as "no evidence". Check every quote against the source before it goes on the page.

Where does this SOP fit in your validation plan?

The persona comes first and keeps getting sharper.

1. Write the proto-persona before your first customer interview.
2. Rebuild it from evidence after 10 or more calls, then check trait frequency with an online survey.
3. Count how many people match it with the market sizing SOP.
4. Use it to plan channels and messages; the free marketing plan generator takes a persona as input.

All 15 Customer Discovery SOPs build on a clear persona.

FAQ

How many personas does a startup need?

One lead persona, and at most one backup. More than two splits your focus before you have a single paying customer.

What should a startup persona include?

Trigger, current workaround, cost of the problem, who decides and pays, budget signals, channels to reach them and real quotes. Demographics only when they change behaviour.

What is a proto-persona?

A persona written from your team's assumptions before research, with every field marked as a guess. Jeff Gothelf popularised the term in Lean UX.

How do I know my persona is accurate?

Test it. Find 20 matching people in a week through the listed channel, and check whether it predicts what 7 of 10 new contacts say and do.