Research + Interpret + Produce = Design

A follow up thought to the user personas discussion among Steve, Jared, Joshua, me, countless other people, and in particular to Peter Merholz’s thoughts about the value of personas created through design team conversations.

Let’s begin with a simple premise that I think most practicing UX designers would agree with in a heartbeat: The worst possible way to employ user personas in a design process is for the designers themselves to have no role in the creation of the persona documents themselves.

Or to put it another way: It sucks when the creator of the research artifacts is not also the designer of the product. If personas are created by a specialized “research team” and then handed off to a specialized “design team”, that design team doesn’t actually experience the substantive benefits good personas can provide.

This is true because, in my view, the primary benefit from creating personas is bestowed upon those who actually make the artifacts, via the thinking, collaboration, and conversations that occur during their creation. The best insights emerge during the investigations and discussions about the data.

Typically we think of a research-based design process being boiled down to a simple equation: “research + design”. I think this model is too simplistic. To me, there are three steps, not two, in any good research-driven design process. Between researching users and designing a product there is an additional critical step, something we all do but don’t recognize as a distinct stage in the design process: research interpretation.

The creation of compelling and useful research artifacts, whether personas or modemaps, mood boards or mental models, is a process of interpreting plain data into meaningful structures and systems that are sensible and useful to designers. It is a synthetic and analytic process at the same time. It’s a creative process. It is a design process.

The Second Step

Let’s look at these three components or stages in a research-driven design process, in particular the second step:

1. Gather research data
2. Interpret the research
3. Produce the design

Ideally all three steps would be conducted by the same team, with the same individuals doing the data collection, documentation, and design work. This is easy when the whole team is made up of skilled designers with good research skills.

But on some projects good user research doesn’t yet exist. Someone will need to conduct surveys, observe users, run tests and analysis, interview domain specialists, and do all kinds of of direct, primary research.

Meanwhile on other projects the research may already exist, in great quality and quantity. The only real research necessary is for the design team to ingest these pre-existing reports and data into their design process.

In either case, however, the second step needs to be taken. Somebody needs to transform the data into something that lays the groundwork for the design.

Getting Creative with Research isn’t a Bad Thing

For example, when we create personas we make editorial decisions about how many different types of users we will define. We may choose to represent several types of users in our group of personas. As an example, let’s say for a news web site we define the following four primary personas based on how dedicated they are to visiting the web site:

• The Temporary Visitor
• The Occasional Repeat Visitor
• The New Subscriber
• The Long-Term Subscriber

Does this breakdown of users not immediately suggest a navigation scheme or a UI design model? Doesn’t it seem likely that all four of these user types will want their needs addressed in some explicit way on the web site, something that manifests itself in a big way in the final design?

But what if we chose to define them this way instead, focusing on their content desires rather than on their devotion to the site?:

• The Sports Fan
• The Political Junkie
• The Concerned Parent
• The Well-Rounded Person

Would this alternate way of thinking of users and of interpreting the data not have a fundamentally different effect on the subsequent UX design process? Wouldn’t the resulting designs be different from the design that came from the first set of personas? The data behind these personas may be the same, but the effect of the interpretation of that data on the rest of the design process may be profound.

There are many other ways, of course, to structure a set of research-informed user personas from the same underlying data. My contention is that this process of transforming data is right on the edge, and maybe over it, of being a design process. Sometimes a dataset may reveal clear design solutions (if 30% of your users speak only Spanish, you may want a link to en español somewhere pretty obvious), but more often than not these kinds of structures are far from obvious in the data. Usually it demands creativity and abductive thinking.

Todd Zaki Warfel likes the phrase “Data Driven Design“. I prefer Data Inspired Design. Data-driven implies that the best design solutions are inferred from or deduced from the data, like Michaelangelo removing David from a block of marble. I don’t think design happens that way, even when data is deeply integrated in the design process. In my mind, the data exists to inspire the designers to new ideas, to point them generally in the right direction towards a solution. Not to provide the solution outright.

This is where interpretation comes in. Interpretation and inspiration. This is the magical part of great design, the part where being a good researcher isn’t enough and where being a good designer isn’t enough. It’s where the designer understands research, and where the researcher understands design.

To be a good designer or a good design researcher, you must master the second step of interpretation.