Inductive coding
Inductive coding is a bottom-up approach to qualitative analysis in which codes are created from the data itself, without a predefined framework. Researchers read the transcripts and label what they find, so the categories emerge from what participants said.
How it works
Analysts read closely, often line by line, and give each meaningful passage a short label that describes it. As labels accumulate, similar ones are merged, broad ones are split, and a codebook takes shape. Later passes group codes into higher-level categories and themes. The process is iterative: each new transcript can add codes or reshape old ones.
Inductive coding suits exploratory research, a new market or a problem nobody has studied, because it can surface what the team did not know to look for. The trade-off is time and consistency, so teams often move to deductive coding once the codebook stabilizes.
A worked example
With no hypothesis about why, a newspaper interviews 18 readers who canceled their digital subscription. Inductive coding produces 96 codes, grouped into 7 themes. The biggest surprise is "reading on someone else's login", with 8 of the 18 canceling because a relative's subscription already covered them.
Related
Automate this in Pulse
An inductive coder can read Pulse's themes as a second reading of the same interviews, noting where a pattern they built from the data was split differently or never surfaced.