top of page

Qualitative research in the age of AI

Aug 19
3 min read

A few months ago, an AI-run study landed at a scale no human team could sensibly attempt – and the whole research industry had a moment. Jon Suarez didn't. In the latest Truth Revealed episode, Truth Consulting's head of behaviour and culture explains why the number of interviews was never the point.


The 13th interview

Every researcher knows that somewhere around the 12th or 13th interview, the returns start to flatten. You're still gathering, but you've largely stopped learning. Sometimes a client asks for 20 or 25 and the sample grows anyway, because bigger feels safer.

So when a new AI format arrives promising 250 conversations, the honest question isn't whether it can be done. It's what those extra 237 conversations are actually for.

Jon's answer isn't "nothing." AI has opened up a middle ground that's faster, considerably cheaper and genuinely useful for a lot of briefs – early proposition testing, directional reads, work where you need a steer rather than a revelation. Truth already blends that format in where it earns its place. The team just doesn't call it “qual”.


The one voice in the room

Six people, an hour in, and one of them says something slightly sideways. It isn't the consensus, it might be the only time anyone says it. And the room shifts – everyone recognises it at once, and you know you've landed on something. That moment is what reframes a category or gives a client an edge their competitors haven't spotted.

Now run it through synthesis at scale. One voice in six is a minor theme. One voice in 250 is a rounding error. The mechanism that makes scale efficient is the same mechanism that smooths out the outlier – and the outlier is often where the prize is sitting.


Why the team isn't called the qualitative research team

At Truth, it's called behaviour and culture. "Qualitative research" describes a method, and method language makes the work sound off the shelf – a thing you buy in a standard size. But what's actually being handled is people: messy, idiosyncratic, patterned in ways that don't sit still, and never operating in a vacuum. Behaviour, and the culture pressing on it.

So the team would rather talk about the matter than the method. Which sounds like semantics until you hear what it changes about how the work runs, and Jon's view that behaviour and culture has a considerably healthier future than qual does on its own, because it's a richer space to be standing in.


The optimistic bit

Jon describes the last couple of years as an AI rollercoaster – loving it, being let down by it, wondering what it does to jobs, accepting the dust hasn't settled. Truth already uses it to underpin work tasks: immersing into unfamiliar categories in hours rather than weeks, sharpening research design, taking the drudgery out of the parts of the job nobody romanticises.

His read on where that leaves the industry is that AI will push everyone to raise their game. It's harder to get away with mediocre work now; the standard, in his word, is kaleidoscopic.

Whether that reads as a threat or an invitation probably depends on how good your last debrief was.



Listen to the podcast on the following platforms:


 
 
 

Comments

bottom of page