Sessions from shopping agents and AI search rarely look like human browsing: they may hit a product page directly, dwell briefly, yet carry clear buying intent. Left unrecognized, this traffic is folded into ordinary visitors, hiding its conversion profile and distorting the denominator of the standard funnel.
Tag them with explainable signals — source parameters, UA shape in request headers, the entry path hit — and record the tagging rules in a changelog. Tags are for grouping and observation only: they do not infer visitor identity, and sessions from different sources are not treated as the same. Rule changes must be traceable, and changes to the tagging definition leave their own trace.
This article covers session identification and grouping only; it infers no visitor identity and judges no source's quality. Record tagged counts as they are, without cross-market absolute comparisons; note flat numbers as flat and do not use them to push new channels or a change of direction.