What is Grounded Theory?

Grounded theory is a qualitative methodology for generating theory directly from data through iterative cycles of collection, coding, and comparison, rather than testing a theory brought in from outside. The name is the claim: the resulting explanation is grounded in the empirical material, built concept by concept from what participants actually said and did. It arrived as a manifesto: Barney Glaser and Anselm Strauss's 1967 book The Discovery of Grounded Theory attacked sociology's habit of verifying grand theories armchair-first, arguing that rigorous procedures could make theory generation as disciplined as theory testing. The methodology has since branched into schools (Glaser's classic strand, Strauss and Corbin's more procedural version, Kathy Charmaz's constructivist approach), but the core machinery is shared.

The Machinery

Constant comparison. The engine: every new fragment of data is compared against existing codes and concepts (does this incident fit, extend, or contradict what's emerging?), so categories sharpen continuously instead of being imposed once and filled.

Coding in ascending abstraction. Open coding fractures the data into labelled concepts; focused/axial coding organises them into categories and relationships; theoretical coding integrates the categories around a core explanation. Memos (the researcher's running analytical notes) do the connective thinking throughout and become the theory's first draft.

Theoretical sampling. The signature move: the emerging analysis decides who to study next. If early interviews suggest that switching costs drive tolerance of a broken workflow, the next participants are chosen to probe exactly that (recent switchers, near-switchers, the locked-in), a deliberate contrast with fixed-in-advance sampling.

Theoretical saturation. The stopping rule: collection ends when new data stops changing the categories, not when a pre-set n is reached, which is the origin of the saturation concept that qualitative research at large now borrows.

Grounded Theory and Its Neighbours

Against thematic analysis, the difference is ambition and iteration: thematic analysis identifies patterned meaning within a dataset; grounded theory aims at an integrated explanation and lets analysis redirect collection mid-study. Against content analysis, it is interpretive and generative rather than enumerative. In product research, full grounded theory is rare (its open-ended timelines strain sprint cadences), but its working parts transfer superbly: iterative collection-analysis loops across waves of a study, evolving interview guides as understanding sharpens, comparison-driven coding, and saturation as the honest answer to "how many participants?". Continuous-discovery teams running small weekly rounds of interviews or open-ended video studies are, knowingly or not, running a lightweight grounded-theory loop, and platforms that make each next small round cheap (a fresh five-person Ballpark study per iteration) make the loop practical at product speed.

Doing It Credibly

1. Bracket, don't blank: enter with questions rather than hypotheses, and memo your assumptions so their influence is visible, the working answer to the myth of the theory-free observer.
2. Keep the loop real: analyse between collection rounds, and let findings change the sampling and the questions, logged transparently.
3. Compare relentlessly and memo everything: the trail from raw data to categories to theory is the method's rigour made inspectable.
4. Claim saturation honestly: report what stabilised and what a larger or different sample might still move, respecting the limits of theory built from one context.

The Takeaway

Grounded theory is theory-building with an audit trail: data first, constant comparison, sampling steered by the emerging analysis, and a stopping rule earned rather than scheduled. Use it in full when the goal is genuine explanation of an unmapped territory, and borrow its loop (collect, analyse, redirect, repeat) for everyday discovery work, where its central lesson applies universally: the next best question comes from the last round's answers.