AI in Market Research

AI has been the hot digital topic of the past year. Virtually every social media platform, company website, and computer program has an AI chatbot, AI customer service, or AI-assistive text. When using Google, in fact, 72% of searchers often or occasionally use the AI summary to answer their query. The AI wave is here, and it’s not slowing down any time soon. Market research is no different; from digital respondents to LLM-powered solutions, AI has infiltrated the market research industry and changed the way we do research.

Digital Respondents

In the market research industry, the hot AI debate takes the form of digital respondents. Digital respondents, sometimes referred to as synthetic respondents, virtual respondents, or just synthetics, are people created by AI-based Large Laanguage Models (LLMs) to take surveys. Data from previous research, or data simulated by a separate AI program, is fed through this LLM and told to create a group of people that will accurately answer survey questions based on the demographics, attitudes, and behaviors of the real data it’s told to replicate. It’s an AI development that’s meant to lower costs and speed up research times, all while remaining accurate.

There are certainly many mixed feelings about digital respondents. Primarily, there is great debate about how, or if, they should be used. Some AI startups offer AI-only research, made up entirely of synthetic respondents. However cheap and quick these respondents may be, though, there are some who question the validity, accuracy, and effectiveness of only using AI respondents. Others take a more mixed approach, using primarily real respondents and filling in the gaps with digital respondents. Still others shirk digital respondents entirely, focusing their effort on real, human panelists. Though there’s no right or wrong answer to how digital respondents should be used, their use will likely continue to grow and change as AI develops.

AI Moderators

AI moderators, in a way, are the direct opposite of digital respondents. While digital respondents are AI built for answering human-asked questions, AI moderators are AI built to ask questions to human audiences. Moderators are people who run focus groups. They ask the questions, push for deeper insights, and generally help to facilitate the discussion between respondents. In this case, the moderator is a large language model that has been fed the information to fish for and sent into a focus group to get it. Similarly to digital respondents, AI moderators are seen as a way to cut costs and speed up research. And just like digital respondents, they have their proponents and their critics.

AI moderators don’t seem to be nearly as controversial as digital respondents; researchers that use them don’t have the same fears of inaccuracy and validity as synthetics, and so don’t fear the prospect of tainted or incorrect data. AI moderators’ biggest supporters say that the AI helped to speed up research, as well as allowed their teams to conduct even more research than if it had just been human-led. It’s also very quick at picking up surface-level insights and patterns. However, the number one hesitation around AI moderation is, of course, the human element of moderation. An AI does not have empathy, cultural understanding, or a curiosity to ask deeper questions than the ones it was programmed to. Thus, most researchers would not want to use an AI moderator without the oversight of a human moderator, though there are some that say it’s only a matter of time before AI can develop that type of deeper insight.

AI Analysis

The largest, and perhaps quietest, use of AI in market research comes in the form of AI analysis. Market research is all about analysis, taking a series of data points and turning it into something understandable and comprehensible. For many, AI speeds that process along. For some, it’s as simple as asking an AI what the population of a target city is. For others, AI is used to comb through hundreds of open-ended questions to summarize the responses and find common themes. 

AI analytics and AI-powered insights are certainly the most common buzzwords in the market industry right now. There doesn’t seem to be much doubt about their efficiency and effectiveness, though whether or not the various softwares offering these AI solutions can actually follow through is a matter of debate. Still, there are a range of approaches to AI analysis. Some researchers tout AI analysis with human oversight, while others proclaim fully AI-driven research. However, there are very few that brag about a complete rejection of AI analysis, suggesting that those that don’t embrace AI analytics will get left behind. 

Like everywhere else, AI is here, but still not fully understood. AI in market research is sometimes met with eagerness, sometimes with skepticism. Like any other AI program, though, there is a generally-held belief to treat it like a tool. Just like any tool, it can be effective, efficient, and well-intentioned, but it has to be used and moderated by a human. This way, we can ensure that speed and accuracy that AI promises, while still keeping an empathic, human touch along the way.