Defaults, form layouts, and the order of options are redesigned so people drift toward the preferred behavior, while the stated policy and the set of available choices stay the same.
An example
Sales reps at a commercial printing and packaging company gave an average of 9 percent off list on reorders customers would have placed anyway. The pricing manager left the discount policy untouched and redesigned the quoting tool over one quarter. The discount field now defaulted to zero, where it had previously carried over the customer's last discount. List price appeared first and in larger type. A discount above 5 percent required a one-line reason from a short menu, and the most common reason, matching a competitor, asked for the competitor's name. Reps could still give any discount the policy allowed. The pricing manager announced the changes in a short note to the sales team, with the reasons. Over the next two quarters average reorder discounts fell to 4 percent, and margin on reorders rose by about $1.2 million.
Why it works
Changing the setup can work all four ways. The preferred choice looks better when it comes first and most prominently, and easier when it is the default that needs no action. Alternatives look worse when they sit lower or in smaller type, and costlier when they require an extra step or a written reason. No choice is removed and no rule changes, which is why the redesign rarely draws opposition. It depends on owning the tools where decisions are actually made: forms, templates, systems, the order of an agenda.
The thinking behind it
Richard H. Thaler Nudge: Improving Decisions About Health, Wealth, and Happiness (2008), by Thaler and Sunstein, defines choice architecture as the way options are presented, argues there is no neutral design, and calls defaults the strongest example. Changing the discount field to default to zero, putting list price first, and asking for a reason above 5 percent is choice architecture. The reps kept every option the policy allowed, which is what separates a nudge from a rule.
William Samuelson Status Quo Bias in Decision Making (1988), by Samuelson and Zeckhauser, found that people disproportionately keep whatever is framed as the current state. A default carried over from the customer's last discount made the old discount the status quo for every reorder. Resetting the default to zero moved the status quo itself, so the bias that had favored discounting began to favor list price.
When it fits
Repeated, routine decisions made by many people through forms, tools, or templates suit this: discounts, expense choices, staffing requests, approval routes. You need ownership of the tool or influence over it, a clear view of which behavior creates value, and data to see whether the change worked.
How to use it well
Identify the decision, where it is made, and what people currently choose by default. Then change the setup and leave the policy alone.
- Set the default to the choice you can defend as best for the company.
- Put the preferred option first and make it the simplest to complete.
- Add a small, honest step to the alternatives, such as a reason, without blocking them.
- Announce the change and the reason, and name an owner for each default.
Measure behavior before and after, and adjust if the default turns out to be wrong for a real group of cases.
What it costs and where it backfires
Defaults are policy decisions, and people who discover one was changed without explanation feel handled, even when the change was sensible. A default that is wrong for some cases costs those cases every time. Transparency keeps it clean. Hiding a default change, designing friction to block choices the policy allows, or burying an option where people cannot find it turns design into a trick, and the policy everyone agreed to stops meaning what it says.
When it is used on you
Trace each default to whoever set it and why, compare the old and new layouts side by side, and treat significant changes to defaults as policy decisions that need an owner and a stated rationale. Watch for behavior that shifts with no new rule to explain it.