A Guess, Written Down Honestly
Everyone on your team already carries a customer in their head.
The engineer pictures a power user who lives in the keyboard. The salesperson pictures the last prospect who ghosted them. The CEO pictures the friend who inspired the whole idea. Nobody says these pictures out loud, so nobody notices they don’t match. Then the roadmap argument starts, and everyone thinks they’re arguing about features when they’re really arguing about a person nobody can see.
A proto-persona ends that argument by putting the person on the table.
The term comes from Jeff Gothelf and Josh Seiden’s Lean UX, and the definition is refreshingly plain: a proto-persona is your team’s best guess about who is using the product, sketched on paper with everyone in the room contributing. No field studies. No survey panel. No budget. You gather the people who hold the assumptions, and you make the assumptions visible. Gothelf is careful about what he’s offering — in his own words, these were “not going to be research-proven customer archetypes” but reference points a team could align around and then go test.
That last clause is the whole idea. A proto-persona is not a lie and it is not research. It’s a hypothesis with a face. And a hypothesis you’ve written down is a hundred times more useful than one you haven’t — because now you can check it.
When a Guess Is Allowed
Assumptions get a bad name they don’t always deserve.
The problem was never that teams assume things — assuming is unavoidable. The problem is assuming silently, then treating the guess as fact. A proto-persona does the opposite: it assumes out loud, in ink, with your name next to it. Done that way, assumption-based personas are not a shortcut around good practice. In four situations, they’re the correct first move.
1. You’re starting from zero. New project, no research, and a decision due Friday. You can freeze, or you can capture your best current thinking and start testing it Monday. Gothelf’s point is that a documented assumption becomes a target for research instead of an invisible bias steering every meeting.
2. Executives each picture a different customer. Gothelf built proto-personas partly as a corporate-alignment tool — a way to pull leaders out of their “inside-out” functional views and into one shared picture of the human they’re serving. Getting six executives to agree on one flawed persona is more valuable than letting them hold six private perfect ones.
3. You need to aim your research. You can’t interview everyone. A proto-persona names who you think matters most, which tells you who to recruit first. The Nielsen Norman Group describes exactly this loop — proto-personas guide recruiting for the studies that will eventually replace them.
4. You’re teaching the habit. A team that has never made a persona will learn more from building a rough one in an afternoon than from reading three articles about the perfect one. The proto-persona is training wheels for customer-centered thinking.
Notice what unites all four: the proto-persona is a way to start moving and start learning. The moment it becomes a way to stop — to skip the interviews because “we already have personas” — it’s failed.
The Evidence Ladder
A proto-persona sits on the bottom rung. The job is to climb.
Every fact on a persona card has a pedigree — a level of proof behind it. A useful way to think about proto-personas is as a persona whose every field currently sits low on that ladder, waiting to be lifted. Here’s the climb, one rung at a time.
- 1Assumption“Our user is a busy ops manager who hates spreadsheets.” A belief. Zero evidence. Written down anyway — that’s the point.
- 2HypothesisReworded so it can be wrong: “We believe ops managers abandon our import because it feels like a spreadsheet. We’ll know we’re right if…”
- 3SignalCheap first evidence: three support tickets, one analytics drop-off, a sales rep’s hunch. Enough to raise or lower your confidence.
- 4FirsthandYou talk to real people. Five to eight interviews, watching what they do, keeping their words. The guess meets the human.
- 5PatternThe same behavior shows up across people and sources. Now it’s not a guess. It’s a research-backed persona — dated, sourced, trusted.
A proto-persona is a card full of rung-1 claims. A great persona is a card full of rung-5 claims. The work between them is not glamorous, but it’s the whole job: turn each belief into a question, then answer it with a real person. Our guide to persona research methods maps where each kind of evidence hides, and the interview questions give you the script for rung four.
The Honesty Rules
A proto-persona is honest right up until someone pretends it isn’t.
The failure mode is famous and quiet. A team runs the workshop, sketches three plausible characters, drops in stock photos, gives them tidy demographics, and prints them on foamcore for the wall. Six months later a new hire points at “Marketing Mary” and asks, “How do we know she exists?” and the room goes silent. The guess has been laundered into a fact. Nobody meant to lie. Everybody stopped checking.
Four rules keep a proto-persona honest:
Label it. Write “proto-persona — assumptions, not research” right on the card. The stamp costs nothing and protects everyone downstream from mistaking scaffolding for a building.
Date it. Assumptions rot. A proto-persona from before you shipped is a museum piece. Put the date on it so its age is obvious.
Track your confidence. Mark which fields are pure guess and which have started collecting signal. When you glance at the card, you should see instantly where the research still owes you an answer.
Skip the fake precision. Resist the stock photo, the invented age, the fictional salary, the hobbies. Detail that isn’t known is detail that lies. A proto-persona should look a little unfinished — because it is.
Key insight: The danger of a proto-persona isn’t that it’s wrong — being wrong on purpose, and knowing it, is the design. The danger is forgetting it was a guess. An assumption dressed up as research is worse than no persona at all, because it wears the authority it never earned.
How to Graduate a Proto-Persona
Graduation isn’t a ceremony. It’s a habit of checking.
You don’t throw the proto-persona away and start over. You promote it, field by field, until almost nothing on the card is still a guess. The move from proto to real is the move from rung one to rung five — here’s the practical version.
Turn every claim into a question. “Hates spreadsheets” becomes “What does she use today, and what does she say about it?” A card of assertions becomes a research plan.
Rank by risk. Some assumptions, if wrong, sink the product; others barely matter. Test the load-bearing ones first — the beliefs your roadmap is quietly standing on.
Gather cheap signal before expensive proof. Before you book interviews, mine what you already own: support tickets, session recordings, sales-call notes, cancellation reasons. Half your assumptions will wobble in an afternoon.
Then talk to real people. Five to eight conversations per persona is enough to start seeing patterns. Ask about their actual last week, not their imagined future. Keep the exact words — the quote that ends up at the top of a great card comes from here.
Re-draft from the evidence, and let it hurt. The point of research is to be surprised. Delete what the interviews contradicted, even the line everyone loved. Add what nobody guessed. If your persona survives contact with real users completely unchanged, you probably weren’t listening.
Run that loop and the proto-persona quietly becomes a qualitative persona — the type the Nielsen Norman Group calls the best fit for most teams, built on five to thirty interviews. The full step-by-step is in how to create a user persona.
Proto, Qualitative, Statistical
Same card. Three very different amounts of proof behind it.
The Nielsen Norman Group draws the cleanest line here, sorting personas by the evidence underneath rather than the format on top. It’s worth memorizing, because it tells you exactly what your persona is allowed to claim.
| Type | Evidence | Best for |
|---|---|---|
| Proto-persona | No new research — a workshop of assumptions | Lean teams, kickoffs, executive alignment, aiming research |
| Qualitative | 5–30 interviews, clustered by hand | Most teams, most of the time — the sweet spot |
| Statistical | Large surveys (100+) with statistical clustering | Well-resourced teams needing sized, defensible segments |
Read the table top to bottom and you’re reading a maturity path. Most teams should start at the top and aim for the middle. The bottom row is a luxury, and NN/g notes it often lands in the same place as the qualitative work at many times the cost. The lesson: don’t skip the proto-persona out of shame, and don’t stop there out of comfort. Climb one rung.
Run the Workshop This Week
You can have a proto-persona by lunch. The trick is what you do after lunch.
Get the people who hold the assumptions in one room — product, sales, support, a founder. Two hours. For each persona you think matters, agree on four boxes: a rough sketch and name, what you believe about their role and behavior, their needs and frustrations, and what might solve them. Argue where you disagree; the disagreements are the gold, because they mark the assumptions worth testing first.
Then, before anyone gets attached, do the honest thing. Stamp the card “assumptions.” Date it. List the five riskiest beliefs on it. Book two customer conversations for next week against the riskiest one. That’s the difference between a proto-persona that grows up and one that ossifies on a wall.
A blank structure helps you move fast — grab the free persona template so the workshop has a shape. And when you’re ready to compare your guess against real buyers, our notes on B2B buyer personas and worked persona examples show what a graduated card looks like.
The proto-persona is a promise, not a product. It says: this is our best current guess, and we’re going to go find out. Keep the second half of that promise and you’ll understand your customers better than any team that skipped the guess — or any team that never checked it.
Build a persona right now — free, no signup — or create a workspace and track each assumption from guess to proof with your team.
Questions People Ask
What is a proto-persona?
A lightweight persona built from your team’s existing assumptions instead of new research. Coined in Gothelf and Seiden’s Lean UX, it captures your best guess about a user in a short workshop. Its value is honesty: it makes hidden assumptions explicit so they can be debated and tested.
How is a proto-persona different from a regular persona?
Same card, different evidence. A qualitative persona is grounded in real research; a proto-persona is grounded in assumptions. NN/g sorts personas into three types by proof: proto (no new research), qualitative (5–30 interviews), and statistical (large surveys). The proto is the honest starting point.
Are proto-personas a good idea or a bad practice?
Good when labeled honestly and treated as a hypothesis. A proto-persona beats no persona and beats an invisible assumption steering decisions unseen. It turns bad only when a team dresses it up as research and stops there. Assume out loud, then go check.
How do you turn a proto-persona into a research-backed persona?
Climb the evidence ladder. Turn each assumption into a testable claim, then gather signal: five to eight interviews, support tickets, analytics, sales calls. Keep what people confirm, delete what they contradict, add what you never guessed. Re-draft from the evidence and date it.
When is it okay to use assumption-based personas?
When you have no research yet and a decision now: a kickoff, aligning executives who each picture a different customer, prioritizing who to interview first, or teaching the persona habit. Always as a way to start and to aim research — never as a substitute for talking to real people.