Artificial intelligence (AI) has done more than make it faster for market researchers to do their jobs. Technology has altered how data sets can be cultivated and analyzed to help organizations better understand their target audiences. To do this, many are pivoting toward using synthetic respondents—simulated data sets output by large language models (LLMs) to mimic responses that projected target audiences might provide in surveys or transaction data. The goal of synthetic respondents is to provide larger data sets that offer more meaningful analysis for market researchers, helping them understand the behaviors and tendencies of the audiences most crucial to their business success.
Combining Synthetic and Human Respondents
The team at Sago, known for its market research expertise, recently held a webinar discussing the role of AI and synthetic respondents for research and profiling purposes. Raj Manocha, Chief Client Officer at Sago, shared that the approach to incorporating synthetic respondents can entail something other than an all-or-nothing mentality. He explained, “This could be a big part of how companies operate in the future. This is where a mix of synthetic respondents and human respondents could coexist together. You do some early-stage testing with these personas and then go live closer to the end with actual people.”
The result could be one that enables market researchers to test out data sets with artificial data and refine their approaches for when the time comes to work with true human profiles. This way, market researchers can work with broader data sets without exhausting real-life resources. This scalability can allow researchers to explore options their existing consumer data may otherwise limit them from considering. As projects become more advanced, researchers can also use true consumer data to refine their results and strategies before making critical business decisions.
This blended approach allows market researchers to test data sets with synthetic information first, refining their methods before engaging real human participants. Using synthetic respondents helps researchers work with broader data sets without exhausting real-life resources. This scalability gives researchers the ability to explore options that their existing consumer data might limit. As projects become more advanced, true consumer data can then be incorporated to refine results before making critical business decisions.
The Ethical Considerations of Synthetic Respondents
Key to this discussion is the role of ethics in using synthetic respondents in market research. The Sago research team acknowledged the importance of integrating real-life data alongside artificial personas and highlighted the benefits of using synthetic results when used responsibly.
Manocha explained that using synthetic data is one solution for companies with limited access to larger data pools that could need more quality and diversity to make informed decisions. He stated, “I think quality has become a harder thing for a lot of companies to manage. I also think that a lot of audiences who are underrepresented on panels, in general, would allow you to kind of stretch that as well.”
A Balanced Approach: Not Replacing Human Data
Manocha and Sago also underscored that research should not be solely synthetic. Manocha elaborated, “I don’t think we’re going to rely on synthetic for everything, but I do think there’s places where it adds a lot more value. And I don’t think it’s about replacing your study. I think it’s about how you have a mix of both things happening or in your research process, where should you use them?”
Future Regulations and Industry Guidelines
The industry is also likely to make moves toward regulation and guidance for synthetic panel usage in market research, helping professionals responsibly use the technology to better understand human behavior. This could also manifest in more regulation and protocols as empirical validation studies are developed to better qualify the results produced from using synthetic respondents in data sets.