Cell Rules Reference
Overview
Cell Rules let you decorate individual cells (color, label, or full CSS) based on conditions you write directly in the visual — no DAX measure required.
You type a list of rules in the Cell rules text box (Format pane → Cell Rules). Each rule is a condition : output pair. For every cell, the visual evaluates the rules top to bottom on the values it has already computed (percentages, index, base, mean, significance) and on the row/column header labels, and applies the output of the first rule that matches.
The previous version relied on a Cell Rules DAX measure that had to reproduce the visual's own calculations (vertical %, side-by-side recomputations, per-scale means…). That is impossible to replicate faithfully in DAX. The engine now runs inside the visual, directly on the numbers it displays, so a rule always sees exactly what the user sees. The old Cell Rules data-role field has been removed.
Writing rules
Syntax
condition : output
condition : output
default : output
- One rule per line.
- Everything left of the first
:is the condition; everything to the right is the output. - The first matching rule wins (evaluation stops there).
defaultmatches everything — use it as a catch-all (put it last).- Blank lines and lines starting with
//are ignored (comments).
// Frequency traffic light
pctV < 0.10 : #FF4444;Low
pctV < 0.20 : #FF8888;Mid
pctV < 0.30 : #8888FF;Good
default : #4488FF;Excellent
A compilation status is shown under the box:
✅ 4 rule(s) applied, or ❌ Line 3: <message> if a rule is malformed.
Conditions
A condition is a boolean expression combining comparisons with AND, OR, NOT and parentheses.
Numeric variables
These come straight from the visual's own calculations:
| Variable | Meaning | Scale |
|---|---|---|
pctV | Vertical percentage (share of the column) | fraction 0–1 (0.33 = 33 %) |
pctH | Horizontal percentage (share of the row) | fraction 0–1 |
cellPct | Cell percentage (value / the cell's own base) | fraction 0–1 |
indice | Index vs. the level subtotal | base 100 (100 = average) |
value | Cell value: count (frequency tables) or mean (mean tables) | raw |
base | Weighted base | raw |
avg | Mean (mean tables only) | raw |
signif1 | First significance test | 1 = higher, -1 = lower, 0 = not significant |
signif2 | Second significance test | 1 / -1 / 0 |
Operators: =, !=, <, >, <=, >=.
pctV, pctH and cellPct are stored as fractions, not percentages. Write pctV < 0.10 for "below 10 %", not pctV < 10. (indice on the other hand is on a base-100 scale.)
value > 100 AND signif1 = -1 AND pctV < 0.333 : #E63946;Rare & low
indice >= 120 : #2E86AB;Over-represented
Header (text) variables
Test the label of the row/column header at a given level. Level numbering is outermost = 1.
| Variable | Refers to |
|---|---|
col1, col2, col3 | Column header at level 1 / 2 / 3 (top → down) |
colX | Any column level (matches if at least one level matches) |
row1, row2 | Row header at level 1 / 2 (outer → inner) |
rowX | Any row level |
In tables with sub-tables (two row levels), row1 is the sub-table name and row2 is the row label. In a simple table, row1 is the visible row label.
Operators (all case-insensitive):
| Operator | Meaning |
|---|---|
CONTAINS "x" | Label contains the substring x |
IN ("a","b","c") | Label equals one of the listed values |
= / != | Label equals / differs from a value |
String literals that contain spaces or parentheses must be quoted: row1 CONTAINS "(4)". Without quotes, the parentheses break parsing. Also, the substring must appear literally — CONTAINS "(10)" does not match note sur 10) (there is no ( right before 10); use CONTAINS "sur 10)" instead.
col1 = "Total" : #eeeeee;Total column
colX CONTAINS "Framatome" : color:#0a7
row1 IN ("BRAND1","BRAND2","BRAND99") : #0088FF;Target
A numeric operator (<, >, …) on a header variable, or CONTAINS on a numeric variable, simply evaluates to false (no error). On colX/rowX, a condition is true if at least one level satisfies it (OR semantics).
Output & display modes
The Measure content dropdown (Format pane → Cell Rules settings) decides how each rule's output string is interpreted.
For every mode except Custom CSS, the output is #color;label, split on the first ;:
| Output | Interpretation |
|---|---|
#RRGGBB or #RGB or rgb(...) | A color |
Any text | A label |
#RRGGBB;Label text | Color and label (label may contain more ;) |
| Mode | Uses | Effect |
|---|---|---|
Cell background color | color | Fills the cell background |
Text color | color | Colors the cell text |
Label only | label | Prepends a label to the cell |
Label + cell background | color + label | Background + label |
Label + text color | color + label | Text color + label |
Label + badge | color + label | Colored badge (shape set by Badge shape) + value |
Custom CSS | free CSS (+ optional label) | See below |
// Mode: Label + badge
avg >= 8 : #27AE60;Excellent
avg >= 6 : #F39C12;Good
avg >= 4 : #E74C3C;Poor
Custom CSS mode
In Custom CSS mode, the output is a raw CSS declaration block, optionally followed by | label. This gives full control: text color that adapts to the background, bold/underline, alignment, gradients, shadows…
// Mode: Custom CSS
avg < 2.6 AND row1 CONTAINS "sur 4)" : background:#E50158;color:#fff;font-weight:bold | Alert
avg < 5.5 AND row1 CONTAINS "sur 10)" : background:#E50158;color:#fff;font-weight:bold | Alert
default : background:#EEE;color:#333;text-align:right
- CSS uses
;as its separator, so in this mode the;no longer splits color/label — the label (if any) goes after a|. - Every declaration is applied with
!important, so a rule reliably overrides the default cell/row styling (alternating rows, cell format…).
Security & certification
User CSS is sanitized before being applied, and applied only through the CSSOM (style.setProperty), never as HTML — so it cannot execute JavaScript and cannot load external resources. The sanitizer rejects any declaration whose value contains url(...), image-set(...), expression(...), backslash escapes, at-rules (@…), </>, braces or comments.
Gradients (linear-gradient(...)), calc(), rgb()/hsl(), vendor prefixes, text decoration, alignment, shadows — none of these need url(), so they all pass. Only external-resource and code-execution vectors are blocked, keeping the visual certification-safe.
Enable Cell Rules
Setting: Activate Cell Rules — Type: Toggle — Default: Off
Master switch. When off, no rule is evaluated and no decoration is applied.
Setting: Badge shape — visible when mode is Label + badge
| Shape | Appearance |
|---|---|
| Circle | Fully rounded pill |
| Square | Sharp rectangular badge |
| Rounded Square | Rectangle with slightly rounded corners |
Cell Rules badges and Ranking badges are cumulative — both can appear in the same cell (the Cell Rules badge is placed first).
Evaluation order & scope
- Rules are evaluated at the very end of the pipeline, after all calculations and after masking: a masked cell is never rendered, so no rule runs on it.
- If a threshold empties a cell's displayed content, rules still apply (the underlying numbers are still available).
- Rules apply to data cells and to subtotal / total columns (they all go through the same render path).
colXon a grand-total column is empty, but numeric conditions androwNstill work there. - The Base / Total rows (Base, Unweighted base…) are out of scope — they bypass the rule engine.
Compatibility and Conflicts
Some modes are mutually exclusive because they write the same visual property. On conflict, the grid is not rendered and an error is shown.
| Conflict | Reason |
|---|---|
| Cell Rules (background) + Ranking (color mode) | Both write cell background |
| Cell Rules (background) + Significance (background mode) | Both write cell background |
| Cell Rules (text color) + Significance (font color mode) | Both write cell text color |
| Ranking (color mode) + Significance (background mode) | Both write cell background |
To resolve: disable one feature or switch to a non-overlapping mode. Significance shown as icons or borders never conflicts.
Because Custom CSS can set any property, it is not part of the automatic conflict detection — you are responsible for avoiding clashes with Ranking/Significance colors.
Practical Examples
Heterogeneous means (per-scale thresholds)
Same table, some rows scored /4 and others /10 (marker in the row label). Thresholds depend on the scale:
// Mode: Custom CSS
(avg < 2.6 AND row1 CONTAINS "sur 4)") OR (avg < 5.5 AND row1 CONTAINS "sur 10)") : background:#E50158;color:#fff;font-weight:bold
(avg < 2.8 AND row1 CONTAINS "sur 4)") OR (avg < 6.5 AND row1 CONTAINS "sur 10)") : background:#FFC79F
(avg < 3.0 AND row1 CONTAINS "sur 4)") OR (avg < 7.0 AND row1 CONTAINS "sur 10)") : background:#96EEEC
default : background:#7CE0A3
Frequency traffic light (background only)
// Mode: Cell background color
pctV >= 0.30 : #27AE60
pctV >= 0.15 : #F39C12
default : #E74C3C
Highlight significant results
// Mode: Text color
signif1 = 1 : #1B7F3B
signif1 = -1 : #C0392B
Target specific columns
// Mode: Label + cell background
col1 IN ("Framatome","EDF") AND indice >= 120 : #2E86AB;Key account
Troubleshooting
The status shows ❌ Line N: Expected ')' found '…'
A value or operand is missing on that line (e.g. avg < AND … — no number after <). The parser consumed the next keyword as the missing value. Complete the condition.
A header rule never matches
The searched text must appear literally in the label and be quoted if it contains spaces/parentheses. Remember CONTAINS "(10)" ≠ note sur 10).
A percentage rule never matches
pctV/pctH are fractions (0–1). Use pctV < 0.10, not pctV < 10.
Custom CSS: background doesn't change (normal mode)
Already handled — every Custom CSS declaration is applied with !important, so it overrides the default background. If you edited rules, reload the visual to pick up the change.
A url(...) in my CSS is ignored
That is intentional: external resources are stripped for certification safety. Use gradients or inline colors instead.
Decoration appears on some cells but not others
Expected — a cell where no rule matches (and no default is set) is left undecorated.