Generating Synthetic Focus Group Data using AI

Did you know you can simulate a focus group using AI tools you probably already have access to?

This falls under synthetic data — artificially generated datasets that follow the statistical patterns of real data, without needing the real thing. This is not something new. Synthetic data has been used for years wherever real data was impractical to obtain, for reasons of privacy, cost or scale. What's changed is that AI tools have made generating such data fast and accessible to all, not just data science teams.

For communications and marketing, that's a genuinely useful shortcut: testing how a message might land with stakeholders, or how customers might react to a campaign concept, before committing to the time and cost of real research.

The good news: unless you need massive volumes, you don't need specialised tools for this. The everyday AI tools you already use — ChatGPT, Claude, Gemini, Perplexity — can do it.

All you need is a detailed prompt. The key is to include as much context as possible. Let's test this out with a hypothetical campaign targeting solo female travellers, given that solo female travel is a genuinely rising trend — Google reported in April 2026 that searches for "women solo travel" had hit a 15-year high.

Here are some key elements to include in the prompt:
- AI persona: define the role you want the AI assistant to take on, e.g. "You are a senior market researcher specialised in travel and hospitality".
- Define the task: give the brief, e.g. "simulate a focus group of 10 participants reacting to a new summer travel campaign for solo female travellers".
- Define the stakeholders: pinpoint exactly who you want feedback from, e.g. "working-age women with university degrees in the UK, with an interest in solo travel".
- Set parameters: state what to probe for, e.g. "destination preferences, and how safety concerns affect travel budgets". Keep parameters neutral — asking the AI to test a reaction is different from asking it to validate one.
- Output format: specify how you want the response structured, e.g. "a brief profile per participant, followed by their reaction, ending with a summary of common themes".

The full prompt is included at the start of the attached document.

As a general rule with AI assistants: the more context you provide, the better the output. You can also run another iteration — if something feels off once you've sense-checked the response, just give clarifying instructions and try again.

One thing to bear in mind is that all AI output needs to be checked and reviewed. Moreover, using AI to create synthetic data can be extremely useful when planning messages, slogans and marketing materials, but it doesn't replace real-world research, which should still happen once the content has been refined.

Curious what the output actually looks like? The attached document shows the full responses from Claude and Gemini.

Fares S Ghneim
Stratetic Communications & Corporate Affairs

Dubai Media City

Ground Floor, Buiding 8

Dubai, United Arab Emirates

© 2026. All rights reserved.