Customers Said They'd Buy It. Why Didn't They?
The short version: intentions predict behaviour, but loosely. A review of 422 studies with more than 82,000 participants found an average correlation of 0.53 between what people intended and what they did, so intentions explain about 28% of the differences in behaviour. In one analysis, 47% of people who intended to do something didn't. A later review found that even a sizeable change in intentions produced only about half as large a change in behaviour. A survey that says '40% will buy' is a ceiling, not a forecast.
Product research surveyed 1,000 customers about the new premium plan. 40% said they would "definitely" or "very probably" buy it. The business case assumed 40% of customers would switch, minus a cautious 10% for "realism".
Six months after launch, 4% have switched. The product team wants to know what went wrong with the customers.
What intentions predict
Psychologists have measured the link between intentions and behaviour thousands of times: people say whether they'll exercise, recycle, attend a screening, buy something, and researchers check later what they did. A 2002 review pooled 422 studies with more than 82,000 people.
- The average correlation between intention and behaviour was 0.53: a real link, so intentions explain about 28% of the differences in what people did. That's the variance explained.
- In an analysis of people who intended to act, 47% didn't.
- People who didn't intend to act rarely did, so intentions are much better at predicting who won't than who will.
A later review of experiments that deliberately changed people's intentions found that a medium-to-large change in intentions led to only a small-to-medium change in behaviour, about half the size.
Why surveys overshoot
Saying yes in a survey is free. It costs no money, no effort and no change of habit, and some people answer the way they think they're expected to. Buying requires all of those things at a later date, when the customer is busy, the price is real and the old plan works fine. That distance between saying and doing is the intention-behaviour gap.
What to ask instead
- Ask for a commitment. A pre-order, a deposit or a sign-up with a card is much closer to behaviour than a tick box.
- Test with a real offer. Put the product on a page for a small group and count who actually clicks "buy".
- Calibrate against past launches. If past surveys said 40% and launches got 5%, use that ratio, not a 10% haircut.
Statistics for noobs: say isn't do
A survey measures what people say. Whenever a decision depends on what people will do, ask how well this kind of answer has predicted this kind of action before. The honest answer is usually "somewhat", and the gap is largest when acting costs money or effort and saying costs nothing.
Where it costs you money
Investor surveys say people will buy more when prices fall. Fund flows often show the opposite. When a forecast rests on stated intentions, look for the record of what similar people actually did.
The moral
The product team ran a second survey to find out why customers hadn't switched. 62% said they "definitely" or "very probably" would switch next quarter. The business case has been updated with 62%, minus 10% for realism.
What this can't tell you
- Mostly health and social behaviours. The reviews cover many kinds of behaviour; buying decisions may be closer or further from intentions.
- Averages hide variety. Simple, one-off actions follow intentions better than habits that need repeating.
- Our story is invented. The 40% and 4% are illustration; real gaps depend on the product and the question.
The premium plan and the product team are an illustration. The figures come from a 2002 review of intention-behaviour research (European Review of Social Psychology) and a 2006 review of experiments changing intentions (Psychological Bulletin). Educational content, not business or financial advice.