Making sense of evidence starts with a distinction that is easy to blur under time pressure: data, findings and insights are not the same thing.
- Data is what was said or observed – a verbatim quote or a recorded behaviour.
- A finding is a short, evidenced statement summarising a pattern across several data points, for example ‘people think they can click the progress bar to navigate’.
- An insight goes further, explaining why the pattern matters and what it means for the service.
The Government Digital Service’s user research guidance recommends capturing raw observations on sticky notes as exact quotes or behaviours, not opinions, so that early interpretation does not quietly overwrite what people actually said (Government Digital Service, 2016).
Once observations are captured, the next step is sorting them into patterns. Affinity mapping – grouping similar observations on a wall or digital board until clusters emerge – is widely used because it is quick, visual and works well with a team rather than a lone analyst (Government Digital Service, 2016). This sits alongside thematic analysis, a more structured process of labelling segments of data with short codes, then comparing codes to build broader themes (Braun and Clarke, 2006). Rosala (2022) sets this out as a repeatable cycle: read everything before analysing anything, code the text against the research questions, group codes into candidate themes, then step away and return with fresh eyes before checking each theme is genuinely supported by the data rather than by one persuasive participant.
Confidence in a finding depends partly on how much evidence sits behind it. With small qualitative samples, this is less about hitting a target number of participants and more about watching for saturation – the point where further sessions stop producing new themes. Guest, Bunce and Johnson (2006) found meaningful themes were already emerging after six interviews and largely stable by twelve, which is reassuring for teams who cannot run large studies, provided the sample still reflects the range of users affected. Quantitative data brings a different kind of evidence – scale rather than depth – and the two work best together. Whitenton (2021) describes this as triangulation: checking a finding from one source, such as interviews, against another, such as analytics or support records, so the limitations of one method are offset by the strengths of another, and a single small study is never asked to carry more weight than it can bear.
Sense-making also has to guard against bias. Involving several people in analysis, rather than one researcher working alone, reduces the risk that findings simply confirm what the team already believed (Government Digital Service, 2016). It helps to keep asking whether a theme is well supported by the data, whether contradictory evidence has been ignored, and whether colleagues reviewing the same data independently arrive at similar conclusions (Rosala, 2022).
Finally, evidence only earns its keep once it is translated into action. In service design, this usually means using insights to define the problem worth solving before moving on to generate ideas – the hinge point between understanding an issue and designing a response (Design Council, nd). Insights become the raw material for personas, journey maps and problem statements, and should be written so that anyone in the organisation, not just the research team, can see what was found, how confident the team is in it, and what it suggests should happen next (Stickdorn et al., 2018; Hall, 2024).
Action Point
Take one recent set of research notes, interview transcripts or customer feedback. Extract 15 to 20 individual observations as exact quotes or behaviours, then group them into three to five named themes. For each theme, write one insight statement explaining what it means for your service and what you would recommend doing next.