Preference is built, not retrieved

The conventional premise of market research is that customers have stable internal preferences which careful questioning can elicit. The literature has been moving the opposite way for decades. Jiang, Thomas and Chu (2025), in the Review of Managerial Science, model brand preference as constructed at the point of choice rather than retrieved from an internal store. Preference is downstream of the choice environment. The finding sits inside a constructed-preference tradition that runs from Lichtenstein and Slovic (2006) through the choice-architecture work of Thaler and Sunstein (2008): preferences are calculated at the moment of choice, in ways that depend heavily on the information environment the choice is made in.

This carries the same problem for traditional brand-tracking instruments that the sampling situation carries for surveys generally. Direct attitudinal questioning, conjoint analyses that ask customers to reason about abstracted attributes, brand-equity instruments built on rating scales: these measure something that does not reliably exist outside the act of measurement. The result drifts from the behaviour it is supposed to predict. The instrument is doing exactly what it is built to do, measuring what it elicits. The trouble is that what it elicits is not the thing you want to predict. This is a validity problem, not a reliability one: the instrument can be perfectly consistent and still measure the wrong thing.

Which environment, for which decision

The productive question is which information environment shapes which preferences get formed. That has a different answer depending on the purchase. Low-involvement decisions are routine, low-risk, often habitual: the chocolate bar at the checkout, the familiar soft drink. The information environment there is thin, the decision is fast, and traditional brand-tracking built around recall and salience still does useful work. This is the ground the Ehrenberg-Bass tradition was built for, where what matters is mental availability, the brand coming to mind at the moment of choice (Sharp, 2010). It is well matched to how the decision is actually made.

High-involvement decisions are different. They matter to the customer, carry real risk, and involve active research and comparison across several sources before the decision lands: cars, mortgages, B2B software, high-consideration electronics. There the information environment is rich and consequential, and the substantive question becomes what that environment is telling customers about your brand right now. That is answerable directly, by looking at the environment, rather than indirectly, by asking customers whose preferences will be formed inside it.

So the useful question is no longer what is inside the customer, but what the environment is telling them. And for a growing share of high-involvement decisions, that environment now has a new centre of gravity.