In practice, researchers can now utilize these platforms to generate screeners based on predefined topics and questions, saving considerable time and effort. Moreover, this automation ensures adherence to established best practices in screener development, empowering both seasoned researchers and newcomers to optimize their research endeavors.

The Advantages:

The pros of this approach are in time savings and accuracy, assuming you have a level of trust in the platform and the knowledge of how to interpret the generative AI output. But like many domain specialists, being able to edit a good start is preferred over starting from scratch.   

Furthermore, if you work in a shop such as ours where research is being conducted often and generally with some urgency, preparing a screener quickly and accurately accelerates time to field and ultimately time to insight. 

Lastly, while the above touches on screener development, research platforms are extending this into sample design, questionnaire development, and even some of the analysis steps. 

All in all, research platforms are focusing on generative AI to automate time-intensive tasks. As an agency, this should translate into faster turns on market research and, over time, the ability to reduce costs on market research efforts. 

The Watchouts: 

Similar to web analytics platforms regressing skills and knowledge around the fundamentals of the discipline, new research teams may be able to use this platform perfectly well, but they may struggle on how to develop the right prompts and what to look for in the output. While these AI tools are meant to help researchers of all levels, just because you can code an AI-generated research instrument does not mean it will be a good one. 

For teams adopting new AI capabilities, consider these capabilities still in beta and be very careful with your review and QA. As the research team, you are still responsible for the end product. 

The Emergence of Synthetic Audiences:

Another intriguing advancement in AI-driven market research is the concept of synthetic audiences. This innovative approach leverages existing data to simulate audience responses, bypassing the need for traditional primary research methods.

For instance, platforms like Vizit analyze consumer preferences based on vast datasets, enabling businesses to predict audience responses to various stimuli, such as product images. By utilizing AI models to assess imagery performance, companies can streamline the image optimization process, accelerating feedback turnaround times.

Synthetic audiences extend beyond image optimization to broader consumer insights, where AI models analyze diverse datasets to generate personas reflecting consumer behavior and preferences. While this approach challenges traditional notions of primary research, it offers unprecedented speed and efficiency in gathering insights.

Navigating Trust and Authenticity:

This is about the time that people say, “Wait - are you telling me we don’t need to talk to humans to do primary research?” And the answer is yes. In fact, we do this today through what we call secondary research. However, the concept is taken way further in this context. 

The idea here is the convergence of secondary research with AI in such a way that it replicates human insights similar to what is produced from a primary research effort. If this takes hold, one can quickly surmise this will be very controversial for many reasons. 

First, the question of trusting “fake” audiences will be front and center. However, the research platforms being developed in this area are testing their analysis from synthetic audiences by fielding similar surveys with real audiences. Anecdotal evidence is suggesting that match rates are north of 95% with regards to accuracy. 

A question to consider is how much is the extra 5% worth. Is that where the real insights live? 

Another area of contention will be around authenticity. In the words of CourtAvenue’s Chief Experience Officer Kayla Brodman, “With synthetic audiences, we lose authentic human emotion and thereby empathy. There is a known and existing concern around AI models having bias due to being trained by humans and on historical data which may not represent all audiences.” If the job of a customer experience professional is to tap into real and authentic human behavior, there may be an inherent conflict of interest here.  

Recognizing that having synthetic audiences is a fairly new concept, it will be interesting to watch how this method of research emerges in the years ahead while tackling the above concerns. 

Looking Ahead:

While market research is often seen as a somewhat sleepy industry, the last few years have seen a lot of big advances. Traditional providers are being encroached upon by a technology first platform, and now, the second part of AI influencing market research is just starting to take hold. 

Both of these are welcome advances, but the jury is certainly out on the value of and concerns around synthetic audiences. Like any new technology advance, it’s important to experiment and see for yourself.