Empirical for the people.

Research methods for healthy websites

Why the need for early user research before committing to a website design? Good research helps us create  a website in step with it’s users needs – site owners included.

Pitfalls of overlooking research

  • Let’s say we agree a specific design pattern, crucial to the overall design, and then we find that the pattern seems to put a significant demographic of our target users off.
  • Say we’ve designed a layout optimised for traditional desktop screens, and then find most of our target users tend to browse the web on their phone, or on a tablet?
  • Or, we might assume our target users browse the web on smaller screens, but research might find something different.
  • we decide to optimise our site for iPhones, but if most of our target audience use Android phones, we’ve wasted time and missed out a priority task.
  • we build in extensive integration to a particular external API, say Facebook’s, only to discover most of our users prefer to use Twitter.


We don’t have to slavishly follow such preferences; maybe we don’t want to depend on third party social networks. Facebook’s ‘technical diffficulties’ last year prompted one web entrepreneur to publicly reconsider their Facebook Connect login strategy. But by conducting quality research, we can make informed choices.

Social research can be characterised as either quantitative, producing data that can be expressed numerically, or qualitative, producing data that cannot. Ideally we want to combine methods from both approaches, to give us a triangulation on whatever it is we’re trying to find out. And that’s the first step, before we choose method: we need to decide what we are trying to discover and from whom.

Case study

I worked on a web site for a youth work team with two broad sets of target users. With a tight budget, I decided to focus my research on the first group: the service users, young people between 10 and 25, living in the catchment area.

What I wanted to find out is how they approach their Internet and Web use. On what kind of screens do they browse the Web? What websites do they use most often? Any they specially like? What do they want from our site? What social networking tools do they use? Any special needs to consider? Will different age groups have different needs?

Ideally I would like to gather this information qualitatively, using in-depth, semi-structured interviews with individuals and groups, and participant observation, which produces the richest data of all. I would like to participate in their activities both in real life and online over several weeks, taking notes on their habits and attitudes from their point of view. I want to imagine and understand what the Web is like for the youth of today. For many of them the Web has always been there. I have a totally different perspective, introduced to the Web in my mid 20s, when the Web had been around for three years.

The more qualitative the research, the deeper the picture. Time and budget in this case restrict how much I can do. We must consider a cheaper, quantitative methodology that can produce statistics and analysis:

The Questionnaire

or Survey. But we must be careful, as we would with qualitative methods, to guard against our biases. In analysis of survey results, unacknowledged bias  becomes an unpredictable variable. The questions must be very precisely worded, clear and unambiguous, yet succinct. Even the order of the questions must be considered, since that could affect how they are answered. We must also take care, I feel, not to seem parasitic: respondents should feel valued, not just sources of information. We’re aiming  for a collaboration.

With this in mind,  we can sample a large number of people in our survey and thus see statistical trends in analysis. Comparing different groups, helps generate useful insights for our site already; they can be followed up in more informal interviews, and find richer information we need for the full picture. We can also include space for qualitative responses, using open questions.

It’s crucial however not to start straight away. Any questionnaire must first be piloted: tested out on potential respondents, to iron out any issues and clarify whether the questions are indeed clear.

In this case, the questionnaire must be understandable by ten-year olds without seeming too ‘young’ to older respondents. Piloting with just four or five participants, balanced by gender and age group, allowed fine tuning these aspects before deployment.

Here’s a generic version of the questionnaire I came up with. Feel free to use it as a base for your projects but bear in mind that it is fine tuned for specific circumstances.