That’s one of the $64,000 questions facing choice architects everywhere. Marketing professors Daniel Goldstein, Eric Johnson, Andreas Herrmann, and Mark Heitmann have an answer. Writing in the December issue of the Harvard Business Review, the quartet lays out some design principles from their “field guide to defaults” that are adaptable to (almost) every business environment.
Accompanying their insights is a decision tree (we recommend clicking on the image below and viewing it in a separate window) for their typology of default rules. The major division in their eight default rules is a smart rule and dumb rule, or what the authors call a “personalized” and a “mass” default. The less you know about your customers, the less you tailor your products to them, the less you need to worry about a complex set of default rules. You’ll probably want to go with something simple, benign, and idiot-proof. If, however, you know something about your customers you’ll want to consider “persistent” default rules pegged to someone’s previous selection, or “smart” default rules that make guesses based on data about past choices.
It might be tempting to want to use forced choices on large, serious decisions, as an Econ might prefer. But, sadly, there is good evidence from behavioral economics that errors are prevalent in these settings. Forced choice is an option to be exercised with caution, as forcing choice can backfire by leading to higher rates of post-choice regret as customers feel like they were unfairly shoved into an option.
(A preview of the article “Nudge Your Customers Toward Better Choices” is here.)
Addendum: A new, interesting example of the power of default rules.
A large national railroad in Europe made a small change to its website so that seat reservations would be included automatically with ticket purchases (at an additional cost of one to two euros), unless the customer unchecked a box on the online booking form. Whereas 9% of tickets included reservations before the change, 47% did after, earning the railroad an additional $40 million annually. This substantial boost in revenue was produced with only a small fixed cost in programming and infrastructure.