A dashboard is a control if someone acts on it and wallpaper otherwise. The distinction is mostly in the design.
Why most are ignored
Too many tiles. Twelve numbers with no hierarchy means no attention.
No normal range, so every fluctuation looks like a problem or nothing does.
Measures nobody can act on.
Refreshed at the wrong frequency — real time for a decision made weekly, or weekly for a decision made hourly.
Aimed at nobody in particular, so nobody owns it.
Vanity measures, included because they look good rather than because they inform.
Design that works
One screen per audience, and different audiences need different screens.
Three to five measures, not twelve.
The normal range marked, so a reader can tell signal from noise without asking.
Trend, not just level.
A named owner per measure.
One line of interpretation on each, written by a person, updated when it changes.
The floor display
Distinct from a management report and frequently confused with one.
Visible from where the work happens.
Two or three numbers, legible at distance.
Aggregate and team-level, never individual, which is a matter of both obligation and cooperation.
Current state, not history: the queue now, the target for the shift, whether anything is blocked.
Actionable by the people looking at it, or it is decoration.
Not a leaderboard. A floor display ranking individuals is a rate-pressure system with a screen, subject to every objection in the injury note.
The supervisor view
The current constraint and where the queue is.
Exceptions needing a decision.
Equipment availability.
Staffing against the demand curve for the rest of the shift.
Not individual rates, which is the temptation and the drift point.
The management view
Weekly, not real time.
Throughput, cycle time percentiles, waiting by stage, with normal ranges.
Safety measures: near-miss reporting rate, detected events, discomfort reports.
Data quality, which belongs on the same page so that figures are read with the right confidence.
Open findings with owners and dates.
One page.
Maintenance
Review quarterly: which tiles has anyone looked at, and which produced a decision.
Remove what nobody uses, which is most of it.
Update the normal ranges as the process changes, or they become misleading.
Fix broken measures immediately. A dashboard with one obviously wrong number loses credibility for all of them.
The test
Ask three people who should be using it what they did differently because of it, in the last month.
A blank answer means it is wallpaper, and the response is to remove tiles until the remaining ones are used, not to add more.
Removing tiles
The maintenance activity that keeps a dashboard alive.
Quarterly, ask which tiles anyone has looked at.
Ask which produced a decision.
Remove the rest, which is usually most of them.
Resist adding one when a question is asked, since the answer is frequently a one-off analysis rather than a permanent tile.
A dashboard that only grows becomes wallpaper within a year, and the shrinking is what keeps the remaining measures noticed.
The interpretation line
One sentence per measure, written by a person, that makes a dashboard readable.
What the number is doing and whether it is inside the normal range.
Why, if known.
What is being done about it, and by whom.
Updated when it changes, not weekly regardless.
Written by the owner of the measure, which is also how you find out whether it has one.
Numbers without interpretation get read as noise, and the reader supplies their own explanation, which is usually wrong.