Deductive coding
Deductive coding is a top-down approach to qualitative analysis in which researchers apply a predefined set of codes, drawn from a theory, framework or the research questions, to the data. It tests and organizes rather than discovers.
How it works
Before reading the transcripts, the team builds a codebook from an existing model, such as a customer journey's stages or a technology-adoption framework. Analysts then read the data and tag each passage with the codes that fit. Because the categories are fixed, coding is fast, consistent across analysts and easy to compare across studies or over time.
Its limit is that it finds only what it looks for: a theme outside the framework goes unnoticed. Many teams therefore work in a hybrid way, coding deductively against the framework and adding inductive codes when the data shows something the framework missed.
A worked example
A car insurer codes 50 claim-call transcripts against the five stages of its claims journey, from first notice to payment. Each passage gets a stage code. The count shows that 38 percent of all complaints fall in one stage, the repair estimate, and that is where the redesign starts.
Related
Automate this in Pulse
A team coding deductively can set its framework's categories beside Pulse's themes and see which categories the interviews filled and which stayed empty.