Is 5 a magical number for prototype testing?

The book is full of design best practices to ideate, test, and make decisions in an accelerated way, one of these best practices is the 5 users prototype test.

But why 5? Are only 5 users enough? Is it trustworthy? If you as a practitioner asked yourself this question, that’s great! You are a critical thinker. You could or could not ask yourself if 5 is enough, but the one who will totally ask you if it is enough with no exception is your client.

“Five is the magic number” is the subtitle of the Testing Chapter of the Sprint Book, and I quote:

Then, the book quotes the Nielsen Norman Group, a graph that shows on the Y-axis the total number of problems that your prototype can have, and in the X-axis the number of users to be tested.

This graph suggests that each user finds the 31% of the total prototype problems, so just testing with one user gives you a 31% of learning, 2 users 52%, until you reach the 5th user, which lead you to the 85% of the problems. This is interesting because the function shows that the learning generated with the first user is huge, and the more users you test, the less you learn, that’s when we reach the sweet point of the 5th user, the learning generated from the 5th to the 6th user is marginal, you are entering now in the diminishing results threshold, that’s why the book suggests interviewing only 5 people.

This number comes from the function N(1-(1-L)^n) and I quote Nielsen Norman Group:

So, since each user finds 31% of the prototype total problems (N), the Function goes: N(1-(1-.31)^5 then, N(1-(.69)^5 so N(.8435) or 85%. Ok, that’s fine and make sense but now… Where does this .31 come from?

Virzi’s final suggestion: Even if in the first 5 users the 80% of the problems are uncovered, as a practitioner, you should stop doing research when you stop learning from users, this is a mix of the 5 magic number and the practitioner own criteria.

Based in Virzi’s experiment, The Nielsen Gorman Group made the first of a set of experiments with 100 people in 1992, having as a result that the 85% of the problems were uncovered by the first 5 people, and 15 to find “all” -Mathematically, the limit neve ends and you never reach the 100% ;)- giving L the Magic .31

For me, the .31 gives you a practical and standard way to test and find out the most prominent problems, iteration after applying feedback in batches of 5 people help you to discover and solve new problems

Arturo Ortega TCUX innovation founder.-

Related posts:

Nesses dias um pensamento peculiar cruzou a minha cabeça: o tanto que a Débora de hoje, o que quer que eu tenha me tornado, que eu represente ou signifique, foi construída também por inspiração de…

What is this black draped body. “My newborn self” is published by Martha Clayton Banfield.