Colour as Information
A discipline, not a decoration — one palette, driven by limits and enumerations, meaning the same thing everywhere.
On an Infohub dashboard, colour is the fastest-read channel there is. Someone crossing a control room registers red before they register any number. That makes colour the most valuable thing on the screen and the easiest to waste.
The whole of this article reduces to one rule: colour must mean something, and it must mean the same thing everywhere.
Reserve colour for status#
If red means "out of specification", then nothing else in your dashboards may be red. Not a chart border, not a heading, not the third series in a line chart's default palette.
The moment a reader encounters red that does not mean a problem, red stops meaning a problem. The signal is spent, and it cannot be recovered by making the real alarms redder.
This is the discipline that most distinguishes a dashboard set people trust from one they skim. It also means the default series palettes need attention — a line chart whose second series happens to be red is quietly undermining every widget on the same screen.
One convention across everything#
Agree a palette once and apply it to every dashboard, every report, and every printed pack.
| Colour | Means |
|---|---|
| Red | Out of specification, action needed |
| Amber | Marginal, watch it |
| Green | Good, no action |
| Grey | No data |
| Blue | Neutral or informational, not a status |
A reader should never have to relearn the scheme when they open a different page. Where several people build dashboards, write the palette down and keep it with the model conventions.
Grey deserves its place in that table. "No data" must be visually distinct from both "zero" and "good". A missing value shown as green is actively misleading — it tells the reader everything is fine when in fact nothing is known. This single distinction catches a class of problem that otherwise goes unnoticed for weeks.
Drive it from limits and enumerations#
Colour should be a consequence of configuration, never a manual choice on a chart.
For measures, drive colour from limits. Set the thresholds on the measure, bind the limit, and the widget, cell, or series colours itself according to where the value sits.
For categories, drive colour from enumerations. Assign colours to enumeration options and every timeline, Pareto, and table showing that attribute uses the same colours automatically.
A hard-coded colour is a statement that will become false. Setting a widget to green because the value is currently fine produces something that stays green when the value goes out of specification. It looks correct on the day it is built and lies thereafter, silently, for as long as nobody checks.
Every colour on a dashboard should be able to change. If a colour cannot change, it is decoration, and it should not be using a status colour.
The other benefit of driving colour from configuration is consistency at scale. Twelve area dashboards coloured by their measures' limits agree with each other by construction. Twelve coloured by hand will not, and the divergence appears gradually as different people build them.
Do not rely on colour alone#
Around one in twelve men has some degree of red–green colour deficiency. On a plant floor, that is not a hypothetical.
Pair colour with something else so the information survives without it:
- A value. A widget showing
94.2%in red conveys the number regardless. - A word. The three-section format string pattern —
'Pass';'Fail';'Fail'— puts the status in text. - Position. A value plotted against a visible limit line shows the breach spatially.
- A symbol. An arrow or a marker alongside the colour.
The test: cover the colour and ask whether the dashboard still communicates. If it does not, add a second channel.
This also matters for printing. Many recipients print reports in black and white, where red and green become identical greys. A report whose entire message is carried by colour arrives blank of meaning. See Exporting and Scheduling Reports.
Choosing enumeration colours deliberately#
Because enumeration colours flow into every chart of that attribute, a few minutes spent on the palette pays off across a whole dashboard set.
For downtime reasons, a scheme that works:
| Category | Colour | Reasoning |
|---|---|---|
| Unplanned failure | Red | The thing to reduce |
| Planned maintenance | Blue | Necessary and scheduled — not a failure |
| Changeover | Amber | Necessary but reducible |
| No demand / idle | Grey | Not a plant problem |
| External | Purple | Outside the plant's control |
The point is that a timeline coloured this way is readable without a legend, and it separates "we have a problem" from "we did something planned" — which is the distinction that stops planned maintenance appearing as the plant's biggest failure on a Pareto chart.
Keep related categories in related hues, and reserve red for the things you actually want people to react to.
Dark and light#
Flow Server supports light and dark themes, and readers use both — control rooms often prefer dark, offices light.
Test a dashboard in both. Colours that read clearly on white can be weak on a dark ground, and mid-greys can disappear entirely. Where a palette has to work in both, mid-tone saturated colours are safer than very light or very dark ones.
Sparingly, on tables#
Limits can colour table cells, which is powerful for exception reporting — and a table where every cell is coloured has no signal left.
Colour the exceptions. A table of twenty-four hours with two red cells communicates instantly which two hours to look at. The same table with every cell shaded green, amber, or red communicates that someone configured limits.
A short checklist#
Before publishing a dashboard set:
- Does red appear anywhere it does not mean "act"?
- Is every status colour driven by a limit or an enumeration, rather than set by hand?
- Is "no data" visually distinct from "zero" and from "good"?
- Does the dashboard still work with the colour removed?
- Does it work printed in black and white?
- Does it work in both light and dark themes?
- Would a reader moving between two dashboards see the same colours meaning the same things?