After a bulletin board closes, researchers face hundreds of answers, photos, and videos. The client does not want to read all of it. They want to know what it means and what to do next. This post explains how to turn board data into coded themes, which are named patterns supported by evidence, and into a debrief clients trust.
What a coded theme is
A theme is a pattern in what people said or showed, such as “Convenience beats price at the weekend” or “People hide the product when guests visit.” Coding means tagging responses with those labels so you can see how many people mention each one and find strong quotes and images. A good theme has a clear name, a short explanation, and evidence.
Topics versus themes
“Price” is a topic. “People accept paying more when the product saves them a trip” is a theme. Topics tell you what people talked about; themes tell you what you learned. Clients act on themes, so push each topic to a statement that could be true or false.
Judgment still belongs to the researcher
Clients do not buy raw transcripts. They buy judgment. AI themes accelerate judgment. They do not replace the researcher who knows the category.
A four-step path to trustworthy themes
- 1Watch themes during field.
- 2Validate against outliers and media.
- 3Pressure-test with a second researcher.
- 4Export the story, not just the codes.
What each step means in practice
- Watching during field lets you ask follow-ups while respondents are still active.
- Validating against outliers checks whether one dramatic answer is distorting the theme.
- Photos and video often contradict or enrich text answers, so look at both.
- A second researcher can challenge labels that sound clever but are not supported.
- A story has a beginning, a tension, and a recommendation, not just a list of codes.
A test for each theme before it reaches the client
- 1Can I name at least three respondents who support it?
- 2Can I name one who contradicts it, and explain why?
- 3Is there a quote or image that makes it vivid?
- 4Does it change what the client might do?
If a theme fails the last question, it may be interesting but is probably not a headline. Move it to an appendix.
What a client-ready output includes
| Element | Why clients value it |
|---|---|
| Short headline per theme | Quick scan for busy stakeholders |
| Supporting quotes and images | Makes the finding feel real |
| Who it applies to | Prevents overgeneralizing |
| What it means for the brand | Turns insight into action |
| Open questions | Shows honesty about limits |
Being honest about strength of evidence
Qualitative samples are small, so avoid percentages. Use calibrated words: “nearly all” for patterns almost everyone showed, “many” for a clear subset, “a few” for notable but limited cases. Name segments where the pattern differs. Clients trust researchers more when they see you separate strong findings from hunches.
Example: a team studying breakfast habits notices an early theme that mornings are rushed. A closer look at photos shows that people who eat well plan the night before. The team probes this on Day 3, and the final story becomes about preparation, not time.
FAQ