Proto-persona

Start work without research and the word "user" means something different to each person in the meeting. Nobody is lying, and the conversation still goes nowhere, because each person is treating the handful of people they happened to meet as the whole user base.

Pagedeck hit this while redesigning team invitations. Mia was picturing an administrator at a 40-person company who has to split permissions by department. Noa was picturing a three-person team where one link finishes the whole invitation. They talked for an hour before realizing they had been describing two different screens. Finding out properly meant two weeks of user research — going out to meet and observe users to collect evidence — and the invitation decision was due that week.

A proto-persona pulls the user out of everyone's head without waiting for those two weeks. Two hours in a meeting room produces one card. Its value is not accuracy but visible disagreement. When four people each draw a card and the user's team size comes out as 3 on some and 40 on others, the first thing the next round of research has to confirm is settled on the spot. The disagreement is the output here, not a side effect.

Six fields go on the card, and each field carries a mark for whether its content is an assumption or something already known. Without the marks, two months later nobody can tell which parts were guesses. The marks are what keep the card honest.

1. Name and situation — something you can say out loud, plus three or four sentences on where this person works and what they do. Naming a guess still has one purpose: nobody remembers "User A," and everyone remembers Dohyun six months later.
2. Goal — what this person is trying to finish by using the product, written as their job rather than as a feature. Not making a document look good, but getting four teammates' input onto one page before a meeting.
3. Where they get stuck — where you expect this person to stop partway. This is the field teammates disagree on most when they draw cards separately, which is why most research questions come out of it. Two people rarely match here.
4. Tools and comfort level — what they already use and how fast they pick up something new. You can't hand the same screen to someone who has memorized shortcuts and someone who hunts through menus one at a time.
5. Evidence marks — each field gets one of three labels: confirmed, heard from a few people, or pure guess. Draw the card honestly and guesses usually come to more than half.
6. Questions to confirm — every disputed or guessed field gets rewritten as a question. "Do administrators actually split permissions by department?" is answerable; "who is our user?" is not.

A single line at the top says the card was drawn before research. Without that line, someone cites the card three months later, and by then the invented fields and the confirmed ones are indistinguishable.

This isn't a document one person writes. A product manager runs the session and the people building the work each draw their own card. A product manager is the person who decides what gets built and why, and stays accountable through launch. Drawing separately and then laying the cards side by side to find the mismatches is the whole point, so starting from one agreed card buries the disagreement the exercise exists to find.

Three groups read it.

1. The team building the work treats it as the premise behind this round's decision. Whether invitations go out as one link or by department rides on this card. A premise that is written down can at least be argued with.
2. Whoever is preparing research reads only the questions field. It becomes the customer interview guide — customer interviews being the method where you spend about an hour with one person hearing what they actually did recently.
3. Sales and support point at fields that clash with the customers they've met. Even before research, that side already carries memories of real people.

The card doesn't leave the company. Put it in investor material or a customer-facing document and the guessed fields start getting quoted as fact. Its life is short too: once research comes back, this card gets thrown out and rewritten as a user persona, the version of the card drawn from research material about the people who actually open the product.

"Research doesn't start until next month, so let's draw a proto-persona this week and go."

An offer to turn the premise behind today's decision into a document.

"Team size is different on Mia's card and Noa's card. That's question one."

The moment a mismatch between two cards becomes a research question.

"That line is a guess. Don't quote it."

A warning not to treat an unmarked field as fact.

Pagedeck booked two hours the day after the invitation meeting went nowhere. Mia, Noa, and Theo each drew a card, the three went up side by side, and only the differing fields got marked. Three fields split: team size, whether permissions get divided, and who sends the invitation.

Those three became the opening of the interview guide two weeks later. Across eight companies, the person sending invitations turned out to be whoever joined first, not the administrator. All three cards had that field as the administrator, so all three people were wrong together — and drawing the cards is what made being wrong visible.

The most common reading is that this is the cheap substitute for research. Two hours, no participants to recruit, so it looks that way. But the card doesn't replace research; it aims research. Treated as a substitute, the team settles for it, keeps postponing the interviews, and six months later is still drawing screens on the same guesses. Two hours of drawing costs two hours; the interviews it points at cost two weeks and come back with answers.

The other common belief is that producing a wrong card is waste, since committing unconfirmed content to a document feels careless. Being wrong is how the card works. Three people filling one field three different ways narrows what research has to check, and three people filling it the same wrong way narrows it too. Skip the card and the guesses don't disappear — they stay inside meeting remarks, where nobody can point at what needs confirming.