What if your next research project started with the voices no one asks for?
AI isn’t just a shortcut. Well, it is. But it can also model, with varying accuracy, the people research was never built to include.
The ones who don’t pass screeners. Who avoid institutions. Who don’t recognise your brand or reject it on sight.
These edge cases aren’t background noise. They’re warnings. Missing voices. Pressure points.
Insight starts where the screener ends.
Use AI to put them back in the room before your fieldwork begins.
🧠 The Silent Majority Still Speaks
Case: Researching digital ID in the UK. Your sample includes renters, commuters, people on broadband. What about travellers outside Bristol? Pensioners using cash? Teens with no fixed address?
Prompt: “Write a response to digital ID cards from someone living off-grid in rural Somerset, who avoids digital systems, prefers cash, and has strong views on personal freedom and government surveillance.”
Why this matters: Products fail when you pretend these people don’t exist. They don’t show up in your quotas, but they’ll show up in your headlines if you get it wrong. Simulate the risk before it plays out.
🗿 Let the Brand Get Heckled
Case: Testing an eco campaign for a fast fashion brand. Ocean shots. Green fonts. Recycled polyester. The sustainability box is ticked. Something still stinks.
Prompt: “As a climate activist from Just Stop Oil, respond to this campaign. Focus on contradictions, emotional hypocrisy, and the use of environmental aesthetics in advertising.”
Why this matters: There’s a gap between what’s said and what’s heard. Better to find the cracks now than get dragged later.
🧨 Aspirational Means Nothing If It’s Not Yours
Case: A skincare brand wants to reach working-class mums in the Midlands. The deck says “aspirational.” The ads say “not for you.”
Prompt: “You’re a 40-year-old single mum in Wolverhampton. You’ve never bought skincare over £5. Tell the brand exactly what you think of this ad.”
Why this matters: Aspiration without connection is alienation. This brings that tension to the surface before money’s wasted on a misfire.
🫥 Resurrect What Got Scrubbed
Case: You ran interviews with young men who dropped out of sixth form and got hooked on gambling apps. The transcripts read like product reviews. The pain’s gone.
Prompt: “Restore the voice. Keep the slang. Keep the silence. Let the regret show up in the pacing. Rebuild the emotional mess that got edited out.”
Why this matters: Clean quotes don’t change minds. Real ones do. Reclaim the rhythm and feeling that got lost in transcription.
🌀 Build Your Own Resistance
Case: A government client wants to explore “inclusive futures.” The brief talks about hybrid work and tech careers. It reads like a TED Talk.
Counterpanel build:
- Ex-miner in South Yorkshire, angry about how work has changed
- Teen in Merthyr Tydfil who left school at 15 and never went back
- Ex-prisoner in Glasgow who sees every job ad as coded rejection
Prompt: “Respond to this policy idea about work and skills in 2030. Say what’s wrong with it. Don’t hold back.”
Why this matters: Every shiny strategy has a blind spot. This brings the people left out into the brief, early and unfiltered.
🚫 Break the Screener
Case: Quotas are full. The sample’s clean. But the data’s flat. The screener’s filtering out tension.
Prompt: “Reimagine this screener from the perspective of someone trying to lie their way in. What does it reveal about the questions?”
Why this matters: If only the polished get in, you’re not doing research. You’re hosting a members club. Break your own entry rules.
Postscript: Not a Solution. Just a Start
This isn’t a framework. It’s not finished. These are prompts, not prescriptions.
But if your research has stopped surprising you, if your participants sound like they’ve been trained by the brand, if the gaps are starting to show, then maybe it’s time to simulate what’s been refused.
Not to speak for the margins. Just to stop ignoring them.
