The panel, which was chaired by Zappi chief innovation officer Steve Phillips, featured Hasdeep Sethi, data science director at Bonamy Finch, Vanja Ljevar, chief data scientist at Kubik Intelligence, and Jasmina Saleh, senior manager, trends and futures (thought leadership) at National Research Group.

Sethi told the conference that “synthetic data is a disruptive force, and it will accelerate”, adding: “We will see improvements to models and techniques, and as it grows in prominence more and more clients will expect it and be curious about it.”

There is a distinction between personas based on real data and synthetic respondent data, with an expected trade off in speed and precision when using synthetic panel respondents, Sethi explained.

He added that for synthetic personas, “clients are becoming warmer to them and expecting to use them for a wide range of use cases.”

“It is up to humans to put in guardrails” on what data AI will use, Sethi added, and that’s a critical role for researchers.

Ljevar said that data quality would be a vital element that determines how successful widespread use of synthetic data would be in the market research industry.

“Previously, we had this huge issue in market research which was data quality,” Ljevar said. “The reality is we often get cases where we can’t get respondents so easily.

“There are some really good things about synthetic data – it protects privacy because you are not relying on real customers, so regulators are happy. It is faster and cheaper, and also it simulates some scenarios that in the real world might not be so easy to test.”

However, Ljevar added: “That’s the good side of synthetic data, but there is always a catch. I love to think of synthetic data as lucid dreaming – it really is the same as reality, and there are only a couple quirks in a lucid dream that help you understand when you are having one.

“Similarly, with synthetic data, there are a lot of existing biases in the data it is trained on. I think market research isn’t going away, but how we do it is evolving.”

Saleh said that synthetic data is still in its infancy, and experimentation is key.

“Where are the limitations, where are the boundaries, what are the things we are maybe not thinking of? There is no fear of failure – let’s experiment, let’s play around,” she said.

Phillips was optimistic about the potential impact of synthetic data: Steve Phillips: “We are researchers – we care about data quality and we worry about application and confidence levels. But what we have already seen is some real progress in the creation of synthetic respondents.”

He added that he believed that market research’s role to be the voice of the consumer in company decisions.

“The truth is we aren’t going to do it just with real respondents, because there aren’t enough of them, we don’t pay them enough and we can’t economically do it,” Phillips said.

“But with synthetic respondents, we can massively increase the number of questions we can ask. We should think of this as, yes, a difficult process, but as a massive opportunity for our industry to be more powerful in business.”