Personas That Work

I’ve been spending a fair bit of time with personas recently, and it reminded me why they get such mixed reactions. Some people love them, some think they’re a waste of time. Both camps are right. It depends entirely on how they’re done.

When they work, personas cut through noise and help teams design with real users in mind. When they don’t, they turn into glossy posters that never get mentioned again. The difference is simple: evidence, depth, and relevance.

• They humanise research into something memorable.
• They shift the lens from us to them, so decisions aren’t just personal opinion.
• They force focus. You can’t design for everyone at once
• They speed up alignment because the team shares a common reference point.

• Speculative personas → built on assumptions, not research.
• Demographics overload → age or hobbies rarely drive behaviour.
• Too shallow → profiles without context or scenarios.
• Static artifacts → PDFs that never resurface in real work.

The fix is simple: keep them grounded in real goals and jobs to be done (more on this another time). That’s often the most valuable part of a persona, it cuts past fluff and gets to why someone actually uses your product.

I’ve always seen personas as more than a design thing. When you frame them around real jobs to be done, they give every part of the business a clearer line to the customer.

Marketing gets sharper because it’s working off real motivations, not just audience segments. Sales can pitch in a way that makes sense to different decision-makers. Engineering finally gets the “why” behind a feature, not just a ticket in Jira. Support teams can spot common frustrations before they escalate. And leadership stays grounded in customer reality instead of running on assumptions.

One thing I’ve been exploring is how AI can speed up the grunt work behind personas. It’s not a replacement for research, but it’s an effective, low-cost way to explore ideas early.

A few ways I’ve used it:

• Role-play testing: I'll tell the AI “You’re a small law firm partner. Your job is to manage the practice while also bringing in new clients.” Then I’ll walk it through an early product concept and see what questions or objections it raises. It’s not perfect, but it helps pressure-test assumptions.
• Jobs to be done framing: Feed AI a rough job like “reconcile the firm’s trust account at month end so the books are accurate”, and it generates scenarios, pain points, and edge cases I may not have listed yet.
• Quick feedback loops: Before I have a polished prototype, I’ll describe a flow and ask AI to respond in character. It’s an easy way to catch friction points before real user testing.
• Synthesis at speed: Drop in raw notes from interviews, chats, or support tickets, and ask for clusters based on goals, blockers, or motivations. It won’t beat deep analysis, but it gives me a head start.

The important part is treating this as experimentation. The output is never the final persona or insight. It’s a starting point, and I'll always go back and validate with real users.

Personas only matter if they’re used. They should evolve as your understanding grows, and they should show up in conversations where decisions get made, not just in workshops.

The real value comes when personas stop being a design checklist and start becoming a tool the whole organisation relies on.

Designers can use them to build empathetic experiences.

Engineers can use them to understand context.

Marketers can use them to sharpen messaging.

Sales can use them to tailor pitches.

Leadership can use them to stay grounded in customer reality.

Get this right and personas stop being theatre. They start doing real work.