By: Meaghan Willis
Meaghan is a senior qualitative research strategist with nearly 20 years of experience partnering with category leaders and globally recognized brands to shape how insight informs business strategy. She has deep expertise across healthcare, financial services/fintech, CPG/retail, luxury, automotive, and beyond, working with audiences ranging from B2B and HNW to Gen Z and Gen Alpha. Read More.
We are entering a strange phase in market research.
An industry built on understanding people is becoming increasingly comfortable operating without them.
Synthetic data. Virtual personas. Digital twins. AI-generated respondents = faster, cheaper, scalable, always on.
And honestly, I understand why this is gaining traction across the industry. Research teams are under pressure to move faster, do more with less, and deliver confidence in increasingly uncertain environments. But we need to look more carefully at what may be getting lost as more of the work shifts from human understanding to synthetic approximation.
Virtual personas feel closer to customers. They are not.
Virtual personas are often positioned as a way to get closer to customers at scale.
Build a model. Feed it data. Ask it questions. Simulate the customer.
It is easy to see why this is becoming such a compelling proposition. But there is an important distinction that deserves more attention:
“Virtual personas do not generate understanding. They reconstruct it.”
What they produce is shaped by what already exists in the source material: the biases in the data, the blind spots in the methodology, the framing of the original questions, and the assumptions researchers may not even realize they are carrying forward. In that sense, they are not a shortcut to customer understanding. They are an extension of what has already been captured and defined.
And that matters because some of the most important things in research are the things people do not say cleanly. What feels emotionally true but sounds inconsistent. What is changing before it becomes easy to measure. What people reveal only when trust, tension, or context shifts. Real people surface that. That is still where some of the deepest understanding comes from.
We’re mistaking coherence for understanding
One of the biggest challenges with synthetic outputs is that they can feel highly credible. They are coherent, clean, reasonable, and predictable — but that is also the risk. Human behavior is rarely that tidy. Real people are contradictory, emotional, contextual, and often difficult to reduce into a single stable pattern.
That messiness is not a flaw in the work, it is often the insight. But many synthetic systems are designed to smooth over that messiness. The result is something that may feel plausible enough to trust while quietly reinforcing the assumptions already built into the system.
That is part of what makes this moment so important. Poor insight does not always look obviously wrong. Increasingly, it looks polished, professional, and coherent enough to move forward. The more research becomes synthetic, the greater the need for research grounded in real people.
The bigger problem: we are losing ground truth
We are moving into an environment where it is becoming harder to know what content is real, who is human, what responses are synthetic, and how much automation sits between organizations and the people they are trying to understand.
That ambiguity matters, especially in insights work. Market research has always depended on one foundational assumption: that somewhere beneath the methodology, there is still a real human signal. When that assumption begins to weaken, the implications are much bigger than a methodological footnote.
What happens when synthetic respondents begin blending into real sample? When “n=1,000” no longer clearly means 1,000 people? When simulated reactions start to become interchangeable with lived experience? These are no longer abstract questions. We are getting very close to those lines, if we have not started crossing them already.
This is why qualitative research matters more than ever
In this kind of environment, qual becomes something more foundational: a way to validate, ground, and interpret what the rest of the system is telling us.
It helps identify where data feels off, pressure-test conclusions before they scale, uncover what structured systems flatten or miss, and surface changing motivations before they become easy to measure. In that sense, qualitative research is not peripheral to the work, it is one of the clearest ways to reconnect it to reality.
Most importantly, it puts us back in direct contact with people. In an age of synthetic everything, that becomes increasingly valuable. Not because it is nostalgic or resistant to change, but because real human conversation remains one of the strongest forms of calibration we have.
Synthetic data is a tool. It cannot become a substitute for reality.
These tools have legitimate use cases and will likely become part of how many organizations work going forward. But there is a growing risk in confusing simulation with proximity.
Synthetic systems can help scale patterns, support exploration and extend what is already known, but they cannot replace human understanding. The more synthetic the ecosystem becomes, the more valuable real human conversation becomes, not as a nice-to-have, but to keep insight work grounded in lived experience. Because ultimately, the goal is not just to generate answers, it is to understand people.
If you enjoyed this article, also check out:
Why Customer Understanding is the Real Competitive Advantage in Banking
More Data, Less Clarity: How Sidelining Qualitative Research Undermines Strategic Decision-Making