Technology

Economists have a tidy way of describing one of research’s oldest problems. Stated preference is what people tell you they’d choose. Revealed preference is what their behavior shows they actually chose.
For insights and UX teams, the distinction isn’t academic. Most research tools are built to collect stated preference at scale. Most business decisions depend on revealed preference. The space between the two is where many launches, campaigns and redesigns go wrong.
Why stated and revealed preference diverge
People aren’t trying to mislead researchers. The divergence comes from how people think and how studies are set up.
People predict their future selves badly. Asked whether they’d buy a healthier snack, people answer as the person they intend to be.
Research removes the context of real choice. In a survey, a concept has full attention. In a store or an app, it competes with everything else for a second or two.
People can’t report what they didn’t notice. If a shopper never looked at the claim, they can’t tell you it didn’t register.
Social desirability shapes answers. Most people want to sound reasonable, kind and consistent.
Where each one is useful
Stated preference isn’t wrong. It’s the best way to learn what people believe, value and intend, and why. It’s essential for understanding motivations, language and unmet needs.
Revealed preference is the better guide to what will happen when the choice is real: on the shelf, in the feed, at checkout.
The strongest research uses both and pays attention to where they disagree.
How to capture revealed preference in a study
You can’t always observe real purchases, but you can get much closer to behavior than a survey alone.
Choice tasks in context. Show the product next to the alternatives people will actually see, and ask them to choose.
Task-based usability testing. Measure whether people complete a task, not just whether they say it was easy.
Attention measures. With consent, webcam eye tracking shows where people looked, in what order, and what they missed.
Reaction over time. For ads and video, second-by-second reaction shows where interest drops.
Behavioral data you already have. Analytics and sales data show what happened, and research explains why.
How to read the two together
When stated and revealed results agree, you can act with more confidence. When they disagree, that’s usually the most important finding in the study. High stated intent with low attention on the key claim means the claim isn’t doing its job. An “easy” rating with a failed task means the flow needs work, whatever people say.
How Decode helps
Decode runs studies with real people and, with consent, measures where they looked and how they reacted in the same study as the questions. Eye tracking runs at 96% accuracy through a standard webcam. That puts stated and revealed signals side by side, in one report.
For the full story of what happens when stated answers go unchecked, read The Polite Yes, or sign up free.
Keep reading: The Evidence Gap: why good research still loses the room · What is an evidence gap in research? A definition and a self-check
Explore on askdecode.ai: Consumer Insights · Freddy, purchase decisions · Consumer Goods
Measured, not modeled.
Keep reading


