Precision, Not Accuracy

As a design tool, it is more important that a persona be precise than accurate. That is, it is more important to define the persona in great and specific detail than that the persona be the precisely correct one. This truth is surprising because it is the antithesis of the goal of interaction design, in which accuracy is always more important than precision. The end result is to have a program that does the right thing, and we are willing to accept some friction in the system to obtain it.

In mechanical devices, moving linkages must be without slack. That is, a piston must move with minimal tolerances in its cylinder. If there were play in the linkage, the piston would quickly slap itself into self-destruction. It matters less that the piston is too short or too long for the cylinder than that it fits without looseness. The same is true of personas. It matters more that the persona is expressed with sufficient precision that it cannot wiggle under the pressure of development than it does that it be the right one.

For example, if we were designing the roll-aboard suitcase, for our persona we could use Gerd, a senior captain flying 747s from Vancouver to Frankfurt for Lufthansa.

On the other hand, we can't extend our persona to include any commercial flyer. Sonia, for example, attends classes at Embry-Riddle Aeronautical University in Daytona Beach and will be a professional pilot when she graduates. She flies every day, but only in small, single-engine propeller planes, and never stays overnight away from home. From a luggage point of view, Sonia is an edge-case pilot. As soon as you blur the definition of Gerd to include Sonia, you make him approximate instead of exact. You get into endless, unproductive discussions about whether Sonia is or is not an airline pilot and what special features her baggage needs.

On the other hand, we could certainly design a roll-aboard by using Francine, a newly minted flight attendant on Reno Air, as a persona. She flies the length of California three times a day, serving drinks and handing out peanuts. Gerd and Francine are dramatically different personas, but their suitcase goals and needs are equivalent.

Programmers live and die by edge cases, and they will bring that awareness to the persona-selection process. They will argue that Sonia has a valid claim on persona-hood because she occupies a pilot seat. But whereas programming is defined by cases at the edge of the paradigm, design is defined at the center. If there is any doubt at all about a persona being very close to the center, that persona should be shunted out of consideration.

In the interest of being precise in the definition of personas, averages have to be ruled out. An average user is never actually average. The average person in my community has 2.3 children, but not a single person in my community actually has 2.3 children. A more useful representative would be Samuel, who has 2 children, or Wells, who has 3. Samuel is useful because he is a person. Yes, he is hypothetical, but he is specific. Our parent of 2.3 children cannot possibly be specific, because if he were, he wouldn't have that impossible average.

Average personas drain away the advantages of the specificity of precise personas. The great power of personas is their precision and specificity. To deal in aggregates saps that power.

Personas are the single most powerful design tool that we use. They are the foundation for all subsequent Goal-Directed design. Personas allow us to see the scope and nature of the design problem. They make it clear exactly what the user's goals are, so we can see what the product must do and can get away with not doing. The precisely defined persona tells us exactly what the user's level of computer skill will be, so we don't get lost in wondering whether to design for amateurs or experts.

The personas we invent are unique for each project. Occasionally, we can borrow from previous projects, but because precision is the vital key, it is rare to find two personas exactly alike.