One canvas or five?

A product org accumulates dashboards the way a kitchen accumulates jars: each one made sense on the day it was labelled. Six months later nobody remembers which lid belongs to the Monday review.

The question we pose in Signal Architecture is not “how many dashboards does a mature team have?” It is “how many first screens does this decision need?” Most weekly product arguments need one. A marketplace with genuinely separate rider and merchant businesses may need two. Five usually means five unresolved fights about what success is.

Split when the decision owner changes

If the person who can change rider incentives is not the person who can change merchant fees, a shared first screen becomes a courtesy visit. Split along ownership, not along chart type. Two canvases with the same four jobs — decision, health, mechanism, annotation — beat one mural with a tab for every squad.

Do not split because a tool made folders free. Folders are not strategy.

Do not split to hide a metric tree fight

We have seen teams create an “activation dashboard,” a “retention dashboard,” and a “growth dashboard” that all open on slightly different versions of the same north star. That is not decomposition. That is conflict avoidance. Sit in one room, pick one sentence, date it, and demote the rest to page two of a single canvas.

Executive viewers can have a thinner first screen. They should not have a parallel universe of numbers. Elena’s guest critique in week seven is mostly this point, delivered without mercy.

A practical test

If you disappeared for three weeks, could a colleague find the weekly review surface without asking Slack? If they would have to choose among five similarly named files, you do not have a dashboard system. You have a search problem. Collapse until the path is obvious, then split only when ownership truly diverges.

Read the Telemetry Canvas sketch, then decide whether you need a studio seat or just a quieter Tuesday.

← All field notes