If you’re asking participants to guess, predict, hypothesize, or anticipate either their future behavior or that of a friend, you’re not researching. You’re pretending.

Be it an interview or a survey, ask your participants about past actions and behaviors. This eliminates aspirational responses (Yes, I would pay money for this app) and answers designed to please you, the researcher.

A safe bet is to eschew any questions that begin with How likely are you to… ? The framing of this question eliminates any burden on the participant to answer based on evidence or past behaviors.

A simple fix is to replace How likely are you to… ? with Have you ever… ? For example, when my colleagues proposed a NPS (Net Promoter Score) question for a survey, I countered with an evidence-based approach. Instead of:

How likely are you to recommend Polygon to a friend or family member?

We asked:

Have you ever recommended Polygon to a friend or family member?

We followed this up with Why? It’s a small change that provides both a measure of and the impetus for said recommendations.


There is a time and place for research based on guesses and projections, but it’s not in a survey or interview. Design walkthroughs and participatory design sessions, among others, offer opportunities to ask tightly scoped questions and collaboratively work through ideas. But here again, the participants should be those who can anchor their responses in past experience.


We are capable of building amazing products and experiences that exceed the expectations of our users and customers, but those experiences begin with a solid foundation based in reality.