3b. Moving from data to insights

Moving from data to insights

Categorizing your data to extract insights

After offloading your data, the next step in analysis is to categorize your data and look for patterns and themes.

There are many different approaches you can take to categorize your data - in this module we’ll take a look at "bottom-up affinity clustering" and "top-down clustering."

It's always a good idea to try a few different categorization techniques with your data, since each method can provide a different perspective on the problem at hand and potential solution criteria. Before you move on to a different categorization method, remember to document your existing categorization with photos and notes as needed.

Bottom-up affinity clustering

How to conduct a bottom-up affinity clustering

  1.  Organize your data into groups based on similarities you’re noticing between different pieces of data. The groups will emerge organically depending on the content of each sticky note.
  2. As groups emerge you should give them a descriptive title on another piece of paper or sticky note.
  3. For each group, write down insights related to the group, focused on the user. It can help to use different colored sticky notes or pieces of paper to capture your insights. Write statements like: “Users want…” or “Users need…”
  4. To help generate insights for a particular group, you can also ask yourself “Why is this group important?” or “So what?”

At the end of a bottom-up affinity clustering you will have many different insights, and from there you will move on to writing your problem statement and design principles.

Top-down affinity clustering

How to conduct a top-down affinity clustering

  1. Identify the high-level categories that you will use to organize your data. Example categorization frameworks include:

    • POEMS: People, Objects, Environments, Messages, Services (Kumar and Whitney, 2003)

    • NOABS: Needs, Objectives, Activities, Breakdowns, Solutions (Alexis)
    • AEIOU: Activities, Environments, Interactions, Objects, Users (Robinson et al., 1991)
  2. Place your data into the appropriate categories.
  3. There will likely be groups that emerge within each category around particular themes or topic areas. As groups emerge you can give them a title.
  4. For each group, write down insights related to this group, focused on the user. It can help to use different colored sticky notes or pieces of paper to capture your insights. Write statements like: “Users want…” or “Users need…”
  5. To help generate insights for each group, you can also ask yourself “Why is this group important?” or “So what?”

At the end of a top-down affinity clustering you will have many different insights, and from there you will move on to writing your problem statement and design principles.

Synthesizing your findings

Once you have finished categorizing all of your data and generating insights, you will take a look at the insights you’ve generated and write down two lists:

List 1: Problem statements

Problem statements are “How might we…?” statements that describe a challenge or problem that needs to be solved. For example, “How might we encourage friends to bike together?” or “How might we help bikers get non-bikers excited about the benefits of biking?”

Sometimes your user research will uncover new problem statements that are different from the problem you originally thought you would be focusing on - but that are equally, if not more, important to users.

As your analysis process concludes, you will likely notice that particular problem statements stand out as most important to users, most critical, or most urgent. These are the problem statements you should focus on as you move into the ideation stage of the design thinking process.

List 2: Design principles

Design principles are statements that describe what an ideal solution should be like. They often start with the phrase, “The solution should…” or “The design should…”

Good design principles describe what makes a solution successful in terms of what matters to users. Good design principles do not describe what a solution should look like, or how specific features should work.

Which of the following statements would make good design principles?

  • The solution should be orange
  • The solution should reinforce the user’s sense of identity as someone who cares about the environment
  • The solution should have a quick start button
  • The solution should help users take ownership over their own health so they feel empowered to make important decisions
  • The solution should grow with users over time to address their changing health concerns as they age

Share and reflect

Share with your problem-specific learning group:

  • What are the key differences between the two approaches to affinity clustering introduced in this module (top-down versus bottom-up)?
  • What might be difficult about conducting a top-down or bottom-up affinity clustering? What do you think you could do to address these difficulties, to make the process go more smoothly?