layers-observed-behaviour

This skill is a library of techniques for the observed behaviour layer of a layered product design framework, the layer closest to reality that captures what users actually do rather than what teams believe or wish. It solves the problem of grounding design in evidence by separating two situations, planning a study when no research exists and synthesising material when research already exists, and instructs detecting and stating which mode applies (with partial research, synthesise first, then plan to fill gaps).

Its disciplines keep observation honest: stay close to raw data near the source, ground claims in something seen or heard rather than team beliefs, mark confidence as observed, inferred, or assumed (an observed claim needs quotable evidence in the same note), name research gaps explicitly, and treat workarounds as strong signal. The planning techniques include defining a learning goal, JTBD interviews, contextual inquiry and observation, diary studies, support ticket and review analysis, analytics review, and usability observation, with guidance to plan synthesis up front and a note that six to ten qualitative interviews usually reach saturation. The synthesis techniques cover extracting concrete observations, pattern grouping, forming candidate job stories with confidence ratings, and gap-flagging. Outputs are lightweight residue: key raw observations, patterns with supporting evidence, candidate job stories, and named research gaps that feed the user-needs layer.

It targets product designers, UX researchers, and product teams using this methodology (it assumes layers-intro is loaded). Typical use cases include designing a user research study matched to a specific learning goal and synthesising interviews, tickets, or analytics into grounded, confidence-rated findings. It is pure research methodology guidance with no code or external tooling.