‘Sweating the information harder’: How AI is making its mark on market research

A highly engaged, opinionated customer base might be every brand’s dream. But it can also be a “double-edged sword” when it comes to managing market research, admits Martha Jensen, brand lead at premium bean brand Bold Bean Co.

Each time the small team sends out a survey to gather feedback on the products, how they’re used and what customers would like to see next, they’re faced with a barrage of replies.

“Our consumers are very happy to spend seemingly hours typing up responses,” she says. “[But] that means going through survey responses can take up a massive amount of our time.”

It’s one of the reasons Bold Bean Co has embraced AI as an invaluable new part of its market research toolkit, using the tech to sift through responses in a fraction of the time it would take the team to do manually. They prompt the AI to pull out key themes, quotes and insights, which that have helped the brand flesh out and evolve its established customer personas.

“It’s been a massive gamechanger for us in understanding our customer base and what they want,” says Jensen.

We’re still at a transitional moment where just because we’re using [AI] doesn’t mean that businesses are either acting on it or really believing in it.

Tash Walker, The Mix

The fast-growing food brand isn’t the only one taking this approach. Marketers are increasingly integrating AI solutions into their market research efforts, be it combing through responses, carrying out interviews or even using synthetic personas that simulate viewpoints with astonishing accuracy.

In fact, according to Marketing Week’s 2025 Language of Effectiveness data, almost half (48.6%) of the more than 1,000 brand marketers surveyed are using AI to help conduct market research. This is the second most popular use case for AI reported in the survey – conducted in partnership with Kantar and Google – after the application in creative and campaigns (57.5%).

So, how exactly is AI changing the field of market research?

As the Bold Bean Co found, AI can be a powerful tool in helping teams manage the vast quantities of data that make up market research insights. At frozen food specialist Nomad Foods, AI has enabled the team to carry out data modelling and forecasting internally “really quickly” and at far lower cost, says Alex Hardy, director of consumer and market insight, and analytics.

“We wouldn’t have had the capability [otherwise] or we might’ve got an agency to help us and the cost would have racked up into tens, hundreds of thousands for the work that’s been learned by one person in my team. It’s enabled something we wouldn’t have done before,” he explains.

‘Eroding trust’: Marketers debate the role for ‘AI-enhanced’ creative

Hardy and his team have also used AI to help test new pack designs, using tools that simulate eye-tracking to provide feedback on whether customers are likely to engage with small changes in the right way.

“Incredibly quickly it will come back and say the right people aren’t seeing this part, for example,” says Hardy.

Due to the affordability of AI, the team are able to test out incremental design tweaks which otherwise wouldn’t have justified in-depth market research.

“We might have done that at a high level before, but if you’ve got a range of 50 different packs, you’re not going to test every single pack,” says Hardy.

AI can also democratise access to research and insights, adds Jensen. By feeding transcripts from follow-up interviews with customers into the AI, the team at Bold Bean Co not only ensure no useful titbit is missed, but it also means the content can be easily shared beyond marketing, such as with product development.

Unlocking insights

Other brands have gone even further in putting AI to work. At J.P. Morgan Payments, the team has worked with platform Evidenza to create synthetic or AI-generated personas of its existing or target customers.

“Traditionally, recruiting for qualitative or quantitative research can be time-intensive and expensive,” explains Karina Vivas, executive director of brand and creative at the JPMorgan Chase-owned brand.

“With synthetic research, we’re able to simulate these audiences and pressure-test ideas, campaigns, or messaging early on. It’s helped us move faster and work smarter, while still keeping rigour in the process.”

The Evidenza platform works by taking in an enormous corpus of data – the training data for LLMs (large language models) – until it understands who is saying what about a particular industry or niche topic, explains co-founder Peter Weinberg. The system uses this insight to create hundreds of thousands of AI-generated or simulated personas, each with their own granular characteristics from level of seniority to family to non-work interests.

“It’s AI impersonating a type of person or a persona – you can think of it as an ‘impersona,’” says Weinberg.

[AI has] been a massive gamechanger for us in understanding our customer base and what they want.

Martha Jensen, Bold Bean Co

The use cases are extensive, he adds, although generally speaking Evidenza is working with marketers trying to sell to “very specific audiences”.

Marketers can survey these AI-generated personas for both quantitative and qualitative research. The accuracy of what they come back with can be validated by both what Weinberg calls an informal “sanity check” – i.e. does it make sense given what we know about a category – and also more formally via head-to-head tests, comparing results from traditional surveys versus AI.

“We do these tests for clients across categories, across markets, across audiences looking at correlation, mean difference, conclusion similarity, to say that these things are essentially telling you the same thing, that they’re more or less identical,” he explains.

“Once you do the first test and see how accurate it is, then all of a sudden you have a hundred other use cases.”

The benefits of this approach are speed, cost and outcomes, says Vivas.

“Synthetic research gives us answers in days – not weeks – and at a fraction of the price. But what’s been equally valuable is what that unlocks for our team, because it’s faster and more accessible. It allows us to test more ideas, challenge our assumptions, and iterate quickly,” she states.

“So yes, it’s more efficient, but it’s also helped us become more informed marketers, with better instincts and stronger go-to-market execution.”

Meta outlines vision for ‘superintelligence’

Used in the right way, AI has the potential to significantly reduce wasted time and investment for brands, says Tash Walker, founder of market insights firm The Mix London.

“Market research as an industry suffers from a really uneven balance in quality,” she says. “There’s a lot of repetition of work, there’s a lot of poorly constructed reports, there’s a lot of utility around testing and validation, which is, quite frankly, a waste of time and budget. It probably gets filed on a server and never ever used again. It shifts the dial by not one jot when it comes to actual marketing services.”

AI is helping to cut out some of that lower quality output, she adds, by providing brands with an alternative that “focuses people’s minds” on whether outsourcing or investing in significant pieces of traditional market research work are really worth it.

The technology can also unlock further insights from existing data, she adds.

“People are looking harder at the knowledge that we already have available to us in the industry. We’re sweating that information harder because it’s available and because AI tools are looking at in a way that no single human being is right now,” Walker explains.

“So, in a way, it’s elevating the role of market research, because people are talking about it and they’re talking about where they’re going to get rid of things, and where they’re really leaning heavily on AI to do a bit of that thinking for them.”

The human touch?

Of course, AI is not without its limitations.

Walker flags a “believability gap” among brands when it comes to the findings provided by AI platforms and tools.

“If it came from AI, is it real? We’re still at a transitional moment where just because we’re using it doesn’t mean that businesses are either acting on it or really believing in it at that very senior exec level,” she says.

“As a consequence, it’s still not quite informing the business decisions in the way that marketers want it to.”

That need to check, double check and check once more echoes the approach at Bold Bean Co. The team always ask AI tools to show its workings, explains Jensen, in order to avoid leftfield responses.

“We do notice that if you don’t continually update it and question it and say: ‘Show your working’ sometimes it wants to please you and will pull out a random thread that you’re like: ‘Where have you got that from?’ So, everyone in the team knows to continually stress test it and use it as a base rather than take conclusions from it,” she says.

“It starts the conversation or does the initial working and a first draft, and then we go from there.”

AI isn’t the always the answer, but it’s helped evolve what’s possible in terms of scale, speed, and accessibility.

Karina Vivas, J.P. Morgan Payments

AI is becoming more sophisticated, but it still lacks the nuance of a human being when it comes to qualitative research, Hardy suggests. When the team experimented with an AI agent or chatbot that asked more in-depth research questions, it was fine “at a surface level”.

“But what you get with real qualitative research practitioners is a depth and real understanding of what people are saying,” he notes. “It can ask the questions, but I haven’t seen anything that’s made me think that’s anywhere as close to being as good as a more traditional approach yet.”

Vivas agrees that AI is complementary rather than a replacement, with the team at J.P. Morgan Payments combining AI and more traditional approaches to market research.

“We continue to value traditional methods, especially when it comes to deep qualitative research or highly specialised initiatives,” she says.

“We aim to use the right tool for the right job. AI isn’t the always the answer, but it’s helped evolve what’s possible in terms of scale, speed and accessibility.”

Given the pace of progress in AI, it’s almost impossible to meaningfully forecast how the technology might shape market research in the years or even months ahead.

For now, AI remains a tool that works best when used in conjunction with the right internal expertise and human ability to drive momentum, turning insights into action, says Robin Karakash, strategic advisor and former marketing lead at Mozilla, Asics and Coca-Cola.

“AI is selling this vision of solving every kind of problem that we have as humanity,” he says.

“You see that in a lot of sales pitches from research organisations coming through, that we solve any kind of problem that marketing has. [But] AI is augmenting your capabilities. If you don’t have the capabilities in-house, for example, to read a lot of data or action against insights, it won’t transform that.”

Karakash argues that for AI insights to be useful brands need people who can judge what’s coming out of the analysis and build a narrative using the data which drives the organisation forward.

Without that capacity, use of AI could simply leave brands right back where they started.