The same coffee shop,
seen from the outside

The idea in three sentences

• A brand emits signals; you manage the sources, not the perception itself.
• Each observer completes those signals into a perception — so the reading is a property of the observer cohort, not an absolute score.
• The gap between intended and perceived is not noise to be smoothed away — it is the measurable quantity, and different cohorts sit at different distances from what you meant to emit.

Eight dimensions, one profile

The perception is a vector across eight dimensions — not a single equity number. Below is Spectra as the specialty-coffee cohort perceives it: experiential-led (the cup), craft and ethos strong, economic scale and heritage still low for a young Berlin shop. (Illustrative values.)

Switch the observer. Watch the profile re-shape.

The instrument reads the same brand through different observer cohorts. Change who is looking and watch the eight bands move — and watch how close the reading stays to the signal Spectra intended to emit. This is what the Brand Spectrometer measures. (Illustrative values; the real readings come from public artifacts.)

The producer–observer seam

One artifact joins the two demos. The OrgSchema demo consumes perception/signal-requirements.yaml downward — the L1 signal must be satisfied by the processes, inputs, and sourcing beneath it. Here the same file is read outward — the L1 signal is what the instrument measures across cohorts. When the OrgSchema tracer pushes steam temperature out of the 60–68°C window and the “silky, never scalded” experience test (L0) fails, the band that drops here is Experiential — internal specification failure, external perception change, on the same parameter.