How Should Behavioural Economics Shape Market Research in the Age of AI?

Behavioural economics should shape market research in one main way: it should push you to study the decision, not just the answer. People don’t choose in a vacuum. They choose in context, using shortcuts shaped by defaults, time pressure, what’s in front of them and how options are framed. Ask them afterwards and they’ll give you a tidy, reconstructed story. Good research therefore observes behaviour in the moment, tests choices in realistic settings, and treats stated reasons with healthy scepticism. Two things have changed since behavioural economics became a marketing buzzword. The evidence for many popular “nudges” turned out weaker than first claimed, so testing in your own context matters more than ever. And AI has arrived, both as a powerful analysis tool and as a new force shaping how consumers choose.

What is behavioural economics, and why does it matter in market research?

Behavioural economics studies how real people make economic choices, drawing on psychology rather than assuming everyone calculates costs and benefits perfectly.

The old shorthand was “people are irrational”. That’s not quite right, and it’s not very useful. A better summary is that people are predictably context-dependent. Mental shortcuts are efficient most of the time. They just mean the setting a choice happens in shapes the choice, sometimes more than the product does.

For research, that matters in three ways:

• What you measure: decisions made in context, not only opinions given in a survey.
• How you measure it: methods that capture behaviour as it happens, not as it is remembered.
• What you recommend: changes to the decision environment (the shelf, the sign-up flow, the price ladder), not just the message.

What has changed in behavioural economics since the nudge boom?

The nudge era, sparked by Thaler and Sunstein’s 2008 book, promised that small, cheap tweaks could reliably shift behaviour. Since then, the field has been through a reality check. A widely discussed 2022 re-analysis in PNAS found little evidence of an average nudge effect once publication bias was taken into account.

That doesn’t mean behavioural economics is dead. It means:

• Effects vary enormously by context. A nudge that works for one brand, audience or channel can do nothing for another.
• Structural changes tend to beat messaging tweaks. Changing the default, the order of options or the effort required usually does more than telling people something.

Plenty of practical patterns still show up again and again in real markets. Defaults stick. A middle option looks safer when there are three price points. Popularity reads as quality. Scarcity makes people grab. The point is to treat these as starting hypotheses and test how they play out in your pricing and product trade-offs.

How do consumers actually make choices?

Most purchase decisions involve several linked choices, each with its own shortcuts:

1. Where to look for advice: a friend, a review site, a retailer, and increasingly an AI assistant.
2. What to buy: often decided from a small shortlist, filtered fast.
3. Who to buy from: shaped by convenience, trust and habit.
4. What to say when asked: the one marketers forget.

That fourth choice is the survey answer itself. People don’t open their heads and pour out the truth. They reconstruct memories and opinions on the spot, influenced by how the question is worded, what came before it and how they’re feeling. That’s why stated reasons and real behaviour so often diverge, and why watching real shoppers at the shelf can overturn what a survey seemed to say.

Where is the line between nudging and manipulation?

The line is whether the design helps people make choices they’d endorse, or works against their interests.

Nudge thinking now comes with a twin concept: sludge, the friction that makes good choices harder (think subscriptions that are easy to start and painful to cancel). Regulators have noticed. Australia’s new unfair trading practices law, passed in 2026 and due to start on 1 July 2027, targets manipulative design that distorts consumer decisions and causes detriment. That includes many “dark patterns” such as false urgency and obstructed cancellations.

Research can show whether a design genuinely makes choosing easier, or just makes leaving harder. That’s also where measuring the customer experience earns its keep: loyalty built on friction rarely shows up as satisfaction.

How is AI changing behavioural research?

AI changes behavioural research in four ways, two helpful and two cautionary.

Where it helps

• Analysing observed behaviour at scale. AI can help code video ethnography, in-the-moment mobile diaries and thousands of open-text responses, so teams can study real behaviour across far bigger samples than manual analysis allowed.
• Generating hypotheses faster, before you design fieldwork.

Where to be careful

• Synthetic respondents aren’t real decisions. Large language models can mimic textbook biases, partly because they’ve read the textbooks. A 2026 psychometric audit found AI-generated survey answers looked plausible but had compressed scales and were easy to tell apart from human data. Use them to sharpen questions, not to replace the people you’re trying to understand.
• AI is becoming a choice architect itself. When a shopper asks an AI assistant “what’s the best option for me?”, the shortlist is set before they reach the shelf or website. Understanding how your category is presented there is fast becoming a research question in its own right.

RCC’s take: three reality checks for any behavioural insight

Before we act on a behavioural insight, we put it through three checks:

1. Observed or recalled? If it comes from what people say they did, verify it against what they actually do. Recall is reconstruction, not playback.
2. In context or in a vacuum? A finding from a lab study, a famous paper or a chatbot is a hypothesis. It needs testing in your category, channel and moment of decision.
3. Human or synthetic? AI-generated answers can help you design better research. They can’t stand in for real customers making real choices.

In practice, that means ethnographic and behavioural research to see decisions as they happen. It means online communities that capture choices in the moment, and in-the-moment mobile surveys and other quantitative methods timed to the point of decision. Focus groups aren’t dead either. Behaviourally designed focus groups, using mock shelves, conflict groups and choice tasks, show far more than a polite round-table discussion.

How does behavioural research lead to better decisions?

Bringing behavioural thinking into research pays off where it counts:

• Fewer launches built on stated intent that never turns into purchase.
• Sharper shelf, pack and price decisions, tested against how people actually choose. That makes testing concepts in a realistic choice setting far more predictive.
• Retention fixes that work, because you can see the friction points between trial and loyalty.
• Lower regulatory risk, because you can show your choice architecture helps customers rather than traps them.

For brand and category teams, behaviourally informed FMCG research turns “why didn’t that work?” into “here’s what to change”.

Frequently asked questions

Does nudging actually work?

Sometimes, and less reliably than early enthusiasm suggested. Once publication bias is taken into account, the average effect across studies appears small, and results vary widely by context. Structural changes such as defaults tend to hold up better than information-only nudges, so test any nudge in your own setting before scaling it.

Can AI predict consumer behaviour?

AI is very good at analysing behavioural data and generating hypotheses, but AI-generated “synthetic respondents” are not a reliable substitute for real people. Research has found their answers can look plausible while differing in measurable ways from human data. Use AI to speed up and scale analysis, and real customers to validate decisions.

Why do survey answers differ from real behaviour?

Because people reconstruct their reasons when asked, rather than recalling them perfectly. Question wording, earlier questions and current mood all shape the answer. That’s why behavioural research pairs surveys with observation and in-the-moment methods.

Talk to us

Ruby Cha Cha is a boutique qualitative and quantitative research consultancy in Surry Hills, and we’ve been bringing behavioural thinking into our work for years. If you want to understand how your customers really decide, our consumer insights consultants can help you design research that sees past the tidy answers. Get in touch with the team, call us on +61283585537 or email info@rubychacha.com.