Grappling with Synthetic Data in Market Research

The Insights industry is finally having its Jurassic Park moment – and I’m loving it! As Ian Malcolm famously warned in the film:

Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should.

That line feels eerily relevant today. Market research (and adjacent) professionals are grappling with AI in all aspects of their work. For decades, our industry has relied on human respondents as the backbone of opinion research – but that foundation is being challenged. One of the most fundamental applications of AI is synthetic data. LLMs have made it easier and more accessible to produce human-like results without the human. And as insights pros are constantly searching for faster and cheaper solutions, the notion of replacing some or all human respondents with synthetic respondents is now a real choice.

Some agencies have decided that it is acceptable to replace a portion of human respondents with synthetic ones, while others have deemed it inappropriate. Personally, I am in camp with the latter. I very clearly see tremendous upside to the use of synthetic data, but as a simulator for unanswered questions and not a means for bypassing human opinions (i.e. augmenting vs. replacing).

But the fact that my view is not universal is exciting! Some go further to say synthetic data isn’t “data” at all and thus shouldn’t be used for anything other than internal testing. And others go the opposite direction and ask why we’re even still collecting human data. Discussions and open debates are occurring regularly at conferences, on social media, and within company offices. Among practitioners, academics, and between the two.

The status quo can get boring, at least for me. And while small advancements do occur with regularity, there hasn’t been this much conversation, anticipation, and experimentation dedicated to a single topic in opinion research since the shift from phone to web data collection. And that was a quarter century ago!

We needed this to push our profession forward – the discussion, the exploration, the skepticism. So where do you stand? Are you with the daring John Hammond, the owner and creator of Jurassic Park, embracing synthetic data as a faster and cheaper alternative to human insight? Or do you side with Ian Malcolm, urging caution and thoughtful application? Either way, the conversation is far from over, and that’s a good thing. Share your perspective. Let’s keep this dialogue alive.