Because shoppers respond to the whole physical object, not just the artwork. A stimulus that’s off on color or finish changes the reaction and produces confident but misleading results. As Luke Cahill puts it, in-context research lives or dies on what you put in a shopper’s hand.

The study can read positive on a pack that never ships, so the launch underperforms at shelf. The inaccurate stimulus quietly invalidates the whole study, and the failure is invisible until the product is live and the numbers don’t match the research.

No. The fidelity should match the decision. A quick directional read can run on less. But any study whose outcome will move a launch, a greenlight, a price, or an inventory commitment needs a production-real stimulus, or the data can’t be trusted.

Because the reaction you’re paying to capture is a reaction to a physical object. Change the object, even slightly, and you’ve changed the answer without changing a question. If the stimulus isn’t real, the result isn’t either.