I am fresh after attending our industry’s oldest Insights Conference “ESOMAR CONGRESS”, held in Amsterdam, Netherlands between 10th and 13th of September 2023. ESOMAR celebrated its 75th year and this conference was the largest ever for ESOMAR with over 1200 attendees, of which we had 250 odd clients. While the numbers are impressive, the true success of the conference lies in its diversity with participation from over 80+ countries. Of all the global conferences I have attended over the years, I can honestly say this was the most diverse conference I have ever attended with participation in significant numbers from all continents, especially the developing and emerging markets.

While as professionals we keep track of the happenings in our industry via our environment, market, clients, partners, technology, etc., a research conference like ESOMAR CONGRESS allows us to see all the actions in one place and see where the most noise is coming from. In addition to that, the post-conference and parallel networking events allow us to get to know each other better. One of these events I attended was a CEO networking event and got to see and hear the voice of CEOs and get a glimpse into the “2023 ESOMAR CEO INDUSTRY INSIGHTS” based on a survey amongst 300+ CEOs/MDs from the largest to niche boutique research agencies globally.

The Survey was managed by Leger (Canada’s largest research agency) amongst ESOMAR members and non-members. It was conducted in August 2023, so was recent enough to give us a good insight into what the leaders of the research world are seeing and where they are taking us.

In addition to renewing connections and making new ones, the objective of a conference is to seek out the next big thing and what the futurists are offering. Even before I reached the conference, the industry was abuzz with talk about AI. The reality from the CEO survey and my talks with the numerous researchers and leaders I met is that very little actual demonstrable work is happening using AI. Even a lot of the work around process improvement and automation, I will classify as Machine Learning rather than AI. This is mainly because the current AIs are mainly large language models, more like a Jack of all trades and master of none. None are optimized and built exclusively for the purpose of market research.

We may not see an LLM exclusively for market research anytime soon. However, we will start to see Small Language Models which will be very specific and deep in knowledge in different areas like Sensory Research or TVC Communication Research, etc. These will be prone to be more accurate, with less hallucinations and have deep knowledge and usage within a narrow sphere of research. They may also be proprietary, limiting their use and reach in the short term.

"We may not see an LLM exclusively for market research anytime soon. However, we will start to see Small Language Models which will be very specific and deep in knowledge in different areas like Sensory Research or TVC Communication Research."

Both these areas fall firmly in optimization and operation efficiency improvements and will continue to evolve. Researchers will need to ensure the checks and balances added to validate these improvements are in place before wider adoption. This is because if the researcher of the future is not involved in certain processes that are handled entirely by an AI, then how can these researchers of the future be able to catch, validate and correct. Currently processes are heavily based on experience of handling and working with data. If this is replaced by AI, then we need to be able to ensure nothing unreliable gets past. So, as we build efficiency with these new AIs at our disposal, we will also need to create new processes to ensure the AIs are doing their job correctly. Giving rise to new jobs and opportunities; “AI isn’t going to replace your job, but someone using AI will.” – bcs.org

Another area of interest that is being looked at as the holy grail of market research is “synthetic respondents”. The synthetic respondents would be a pool of artificial respondents who are developed based on existing databased and publicly available social media content to develop artificial consumer profiles and personas that will mimic real consumer responses. The challenge is not in creating these profiles to begin with but to refurbish, update, upgrade and evolve these profiles in an ongoing manner. Further these personas will need to evolve based on their environment and dynamic ecosystem. While this may change things drastically this will give emphasis and opportunity to areas like synthetic user creation, user testing, user validation, user refurbishment and even user entombment.

Whichever direction a research service provider takes, they will have to answer some fundamental questions around what kind of a business do they see themselves as a “research consulting service” or a “technology product lead service”. The skill sets and resources required would be vastly different for each. The new technology implementations today require specialists and dedicated teams to ensure we get the best out of them. One solution or technology to fit all with minor tweaks is a thing of the past.

And finally, regulation and government policy around AI is still evolving. Regulatory policy may have the final say on how we can and cannot use AI. Historically, market research has threaded a fine line between privacy and delivery of insights. Hopefully, we can be part of the solution in growing this potentially once in a lifetime technology transformation we find ourselves in.