Your customer satisfaction scores are excellent. Your NPS is strong. Your concept test showed 72% purchase intent. So why aren’t the numbers moving?
Welcome to one of the most costly and consistently underestimated problems in market research: the intention-action gap.
It is not a data collection problem. It is not a sample size problem. It is a fundamental feature of human psychology and until your research methodology is built around it, you will keep making confident decisions based on evidence that doesn’t predict what you think it predicts.
What is the intention-action gap?
The intention-action gap is the systematic divergence between what people say they will do and what they actually do.
It shows up everywhere in consumer behaviour research. People say they intend to switch banks – and don’t. They say sustainability influences their purchase decisions and then buy the cheaper option. They rate a new product concept as “definitely would buy” and walk past it on the shelf without another thought. They say price doesn’t matter that much – until it does.
The gap is not random noise. It is not explained by dishonesty or poor survey design. It is structural, predictable, and rooted in how human cognition actually works.
Understanding it is not an academic exercise. For any organisation making strategic decisions based on stated consumer intentions, it is a commercial imperative.
Why the gap exists: the cognitive architecture of decision-making
To understand the intention-action gap, you need to understand the architecture of human decision-making, specifically the distinction between two fundamentally different cognitive systems.
System 1 is fast, automatic, and largely unconscious. It operates through pattern recognition, emotional association, and habit. It handles the vast majority of everyday decisions, including most purchase and brand choices, with minimal conscious effort.
System 2 is slow, deliberate, and effortful. It is the system that activates when we reason through a problem, weigh options explicitly, or respond to a survey question.
The critical insight, established through decades of cognitive and behavioural science research, is that System 1 and System 2 don’t always agree. When we respond to a survey, we’re using System 2 to report on decisions that will eventually be made by System 1. And System 1 doesn’t always follow through on System 2’s stated intentions.
This is not a design flaw. It is an adaptation. The automatic, associative nature of System 1 thinking allows us to navigate a complex world without deliberating over every micro-decision. But it means that the stated intentions captured in traditional research are generated by a cognitive process that is categorically different from the one that will actually drive behaviour.
The five drivers of the intention-action gap
The gap between stated intention and actual behaviour is not monolithic – it has identifiable causes, each with different implications for research design.
1. Social desirability bias
When answering survey questions, people are motivated, often unconsciously, to present themselves in a socially acceptable light. This is why sustainability research consistently shows that far more consumers say they prioritise environmental factors in purchase decisions than actually demonstrated through buying behaviour.
The respondent isn’t lying. They are reporting their values and aspirations rather than their likely behaviour. The survey instrument, by its very nature, invites this kind of response.
2. The hypothetical bias
Concept testing and purchase intention questions ask people to imagine a future scenario: “If this product were available at this price, how likely would you be to buy it?” The answer is generated by imagining, not deciding. And imagining is a fundamentally different cognitive process from deciding.
When we imagine a future choice, we tend to underweight the friction of real-world switching, overweight the novelty of the new option, and entirely neglect the pull of existing habits. The result is systematic over-prediction of new behaviour.
This is why concept testing has such a consistent track record of over-predicting new product success. The methodology is generating hypothetical enthusiasm, not actual purchase likelihood.
3. The intention-habit gap
Even when someone genuinely intends to change their behaviour, existing habits are extraordinarily powerful. Behaviour that is deeply habitual – such as buying the same brand, filling the script at the same pharmacy, using the same bank – operates below the threshold of conscious decision-making. It doesn’t require intention because it doesn’t require a considered decision.
This means that expressed intention to switch is not a reliable predictor of switching. Intention exists in System 2; the habitual behaviour it’s competing with exists in System 1. And in everyday contexts with no external prompt to make a deliberate choice, System 1 wins.
4. Context collapse
Survey responses are generated in one context. The behaviour they’re meant to predict occurs in a different context. These differences matter enormously.
The respondent filling in a survey at home, with no time pressure, with full cognitive attention on the question being asked, is not the same person standing in a supermarket aisle, distracted, running late, reaching for the familiar brand. The cues available in the moment of real decision-making – shelf placement, packaging, peer behaviour, price promotion, cognitive load – are simply absent from the survey environment.
Stated preferences don’t travel reliably across contexts. This is not a problem that can be solved by better survey design. It is a fundamental limitation of asking about behaviour in an environment stripped of the contextual triggers that drive it.
5. The affect heuristic and post-hoc rationalisation
People frequently make decisions based on emotional responses and then construct rational explanations afterwards. When asked about a decision such as why they chose this brand, what matters most to them in a product, what would make them switch, they are often providing a plausible narrative rather than an accurate account.
The emotional and associative drivers of the actual decision may not be accessible to conscious introspection. What people report as their reasons is largely a reconstruction – coherent, sincere, and systematically incomplete.
The scale of the problem
The intention-action gap is not a small or occasional discrepancy. Research across categories consistently documents a dramatic divergence between stated and actual behaviour.
Studies of purchase intention scales – the standard “definitely/probably would buy” measures used in concept testing – have found that actual trial rates are typically a fraction of what stated intent predicts. Across FMCG categories, the average conversion from high stated intent to actual purchase has been estimated at between 20 and 50 percent, with variance depending on category, novelty, and switching costs involved.
In healthcare, the gap between physician-stated prescribing intention and actual prescription behaviour is well-documented and commercially consequential. Physicians may express genuine openness to a new medication and then continue prescribing the established competitor through force of habit and contextual inertia.
In financial services, intention to switch providers is one of the weakest predictors of actual switching. High rates of expressed dissatisfaction coexist comfortably with very low rates of actual churn – not because the dissatisfaction is insincere, but because habit, inertia, and switching friction are far more powerful than they appear in survey responses.
What closes the gap: a behavioural research approach
Acknowledging the intention-action gap is the first step. The second is designing research that is built around it rather than blind to it.
This is the shift from research that describes to research that predicts, and it runs through everything we do. See how we apply behavioural science across our work.
Measure behaviour-proximate constructs
Rather than asking about intention, measure the underlying variables most closely connected to actual behaviour: automatic brand associations, habit strength, switching friction, and the contextual triggers that activate category consideration. These variables are harder to articulate but far more predictive.
Use forced-choice and trade-off methodologies
Techniques like MaxDiff and conjoint analysis are widely known examples that require respondents to make explicit trade-offs – mimicking the structure of real decisions rather than allowing everything to be rated as important. The preference data this generates is substantially more predictive than stated importance ratings or purchase intent scales.
Test responses, not reports
Implicit association techniques measure automatic mental responses – the kind of fast, associative reactions that actually drive System 1 decision-making – rather than asking people to report on their mental states. These measures are significantly less susceptible to social desirability and post-hoc rationalisation.
Incorporate contextual and behavioural data
Where possible, triangulate stated preferences against observed behaviour – whether from passive data, ethnographic observation, or real-world choice experiments. Actual choices, even in controlled experimental settings, are more reliable predictors of real-world behaviour than hypothetical ones.
Apply behavioural segmentation
The intention-action gap is not uniform across consumers. Segmentation that identifies individuals by their decision-making profile – habitual choosers versus active evaluators, for example – allows research findings to be applied with precision, targeting the consumers for whom a given intervention will actually change behaviour.
The organisational cost of ignoring the gap
The intention-action gap has a financial consequence. Every strategic decision made on the basis of research that doesn’t account for it carries a hidden risk – the risk that the behaviour predicted by the data will not materialise at the scale assumed.
That risk shows up as new products that don’t reach forecast sales targets. As campaigns that generate brand affinity measures without shifting purchase behaviour. As customer experience investments that improve satisfaction scores but don’t reduce churn. As pricing research that understates actual price sensitivity.
The question is not whether this risk exists – it does, in every organisation that uses traditional research to forecast behaviour. The question is whether it is being measured and managed.
The bottom line
The intention-action gap is not a methodological nuisance to be minimised. It is a fundamental feature of human cognition – one that traditional market research was not designed to address.
Behavioural market research is. By incorporating what we know about how decisions are actually made – automatically, contextually, and largely below conscious awareness – it produces findings that more accurately predict what consumers will do, not just what they say they’ll do.
In a business environment where the cost of a wrong strategic call is high and the volume of research data is increasing, the quality of prediction matters more than ever.
The intention-action gap is closeable. But it requires research designed to close it.
Honeycomb Strategy builds behavioural research methodologies that account for the gap between stated intention and real-world action. Our work helps clients make strategic decisions on evidence that predicts behaviour – not just measures it.
Renata Freund
Founder & Director