How Synthetic Personas Help Teams Scale User Understanding 🤖✨

[#1]

Understanding your users at scale is a challenge every enterprise team faces — especially when customer needs vary across regions, roles, and responsibilities. Like many teams working in the Atlassian ecosystem, we spend a lot of time trying to better understand our users: who they are, what they need, and how we can support them more effectively.

But getting those insights isn't always easy — especially at scale. That's why our team started exploring a new approach: synthetic personas. These are AI-powered profiles designed to think, ask, and respond like real users. They're not meant to replace actual customer conversations—but they're helping us work smarter, move faster, and stay more in tune with the people we build for.

While synthetic personas might sound like something for the marketing team (and yes, that's where the idea started 😉), their impact goes much further. Because as your team's scale, so does the need for clarity, empathy, and speed in how you build and communicate across products, platforms, and people.

How We Built a Synthetic Persona

Many of our customers are active here in the Atlassian Community — asking questions, sharing feedback, and solving challenges. So we thought: what if we could build something that learns directly from those conversations?

We trained a GPT model using carefully selected community discussions and real customer questions. The result? A synthetic persona that reflects the language, mindset, and common concerns of our users.

It's not perfect — but it's already proven to be incredibly helpful.

How You Can Use Synthetic Personas 💡

Synthetic personas aren't just a concept — they're a practical tool you can apply across teams and workflows. Here are a few ways to get started:

1. Improve Documentation Clarity

Run help content through your synthetic persona before publishing. If it shows signs of confusion or asks unexpected follow-ups, take that as a cue to clarify. It's a quick way to identify gaps and make docs more user-friendly.

2. Refine Product Messaging

Use the persona to pressure-test your feature descriptions, visuals, or onboarding flows. It can highlight jargon, unclear language, or missing context — before it reaches your audience.

3. Support Team Training

Let new team members interact with the persona to simulate realistic user questions. It's a safe, repeatable way to practice responses, explore edge cases, and build confidence — especially useful for support, product, or sales teams.

4. Explore and Validate Use Cases

Use the persona to walk through new features or workflows. Ask questions like "Would this be useful?" or "What might be unclear here?" The responses can help uncover assumptions or usability challenges you hadn't considered.

5. Think Beyond the Customer

You're not limited to end users. Try creating personas for internal stakeholders, partners, or reviewers. These profiles can help you explore internal collaboration, alignment, and training needs more effectively.

Synthetic personas aren't a shortcut. They require high-quality inputs, ongoing refinement, and a deep understanding of your audience. But when used well, they can help you:

• Make faster, more confident decisions
• Communicate with greater clarity
• Keep user needs front and center

They're also a powerful reminder that understanding your audience isn't a one-time effort — it's an ongoing practice.

If you work on a large team, across multiple products, or with complex user journeys — synthetic personas can help. They won't replace interviews, analytics, or direct feedback, but they can help you scale empathy in a consistent, repeatable way.

Here's how to begin:

1. Choose one audience segment (like admins, project managers, or security leads)
2. Gather real inputs (support tickets, user forums, analytics tools like GA or Hotjar)
3. Test in small ways (internal docs, product copy, onboarding guides)
4. Review often and balance AI insights with real-world validation

A Note on Limitations ⚠️ 

It's worth acknowledging: synthetic personas are only as good as the data and training behind them. They can surface patterns, simulate realistic behavior, and speed up internal feedback — but they're not infallible. AI can sometimes reflect outdated information, reinforce biases in source material, or give answers that sound confident but miss the nuance of your specific context.

Think of it as a helpful collaborator — not a final decision-maker. It works best when paired with human judgment, user research, and a healthy dose of curiosity. Use it to guide and inspire, not to replace the critical thinking your team brings to the table. 😉

We're still learning. If you've been exploring this space too — or are curious to start — I'd love to connect. The more we share what's working, the better we all get at building for scale, complexity, and real human needs.