The standard behind every dashboard we build
Our methodology rests on three pillars: a disciplined set of design principles drawn from the notation standard our team is certified in, a purpose-first visual vocabulary for choosing the right chart, and a governed semantic layer underneath it all.
Certified in IBCS®
Our team holds the International Business Communication Standards certification and applies its principles across every dashboard we design.
Pillar 01
Our eight design principles
These eight principles guide the certified notation standard our team applies. We use them as a checklist against every dashboard and report we design, and as the reference point when we review dashboards our clients already have.
Say something
Every chart or report should communicate a clear message, not just display data. If a viewer can't state the takeaway in one sentence, the chart hasn't done its job yet.
Unify
Consistent visual semantics across the board. The same color always means the same thing: actual, budget, a positive variance, a negative one.
Condense
Reduce visual noise. Gridlines, drop shadows, 3D effects and redundant legends are cut unless they carry information the reader needs.
Check
Numbers and visuals must be correct and verifiable. Every figure on a dashboard should be traceable back to its source in the semantic layer.
Express
Use the chart type that matches the comparison being made. Evolution over time, ranking, correlation, or distribution each call for a different form.
Simplify
Remove any element that doesn't help the decision at hand, including the decorative parts that "look nice" but add nothing to comprehension.
Structure
Organize dashboards in a logical hierarchy: overview first, detail on demand, so a reader always knows where they are in the story.
Standardize
Apply one style guide across every dashboard in the organization: color palette, typography, number formats and sign conventions, without exception.
Principles in practice
The same data, two very different reports
A short illustration of what applying Condense, Unify and Express looks like on a typical variance chart.
- Seven unrelated colors compete for attention
- 3D shadow distorts the actual values
- Dense gridlines add clutter, not clarity
- No stated takeaway, just raw bars
- One neutral tone plus a single highlight color for what matters
- Sorted by rank; the headline states the message
- Direct labels replace a separate legend block
- Flat bars, real baseline, no distortion
Pillar 02
A purpose-first visual vocabulary
Before picking a chart, we classify the analytical question being asked. This chart-selection-by-purpose approach is widely recognized practice in data communication and business intelligence, and we apply it as our working method, not as licensed content from any single source.
The question: "How do categories rank against each other?"
Comparison
- Bar chart, sorted by rank
- Dot plot for many categories
- One consistent baseline
The question: "How is this changing over time?"
Evolution
- Line chart for trend and seasonality
- Column chart for discrete periods
- Never a dual axis
The question: "How spread out are the values, and where are the outliers?"
Distribution
- Dot plot or strip plot
- Box plot for quartile summaries
- Histogram for large samples
The question: "Is there a relationship between two variables?"
Correlation
- Scatter plot, capped series count
- Trend line only when statistically meaningful
- Never used to imply causation
The question: "How do the parts make up the whole?"
Composition
- 100% stacked bar for share of total
- Single pie only for 2–3 segments, if at all
- Avoided when parts change over time (use area instead)
The question: "How does volume move between states?"
Flow & connection
- Sankey diagram for multi-stage flows
- Funnel chart for sequential drop-off
- Network graph only when node count stays legible
Pillar 03
Semantic layer architecture
A semantic layer is the translation layer between your raw data model and the business metrics people actually consume: consistent, reusable and governed across every tool that touches your data.
We treat this as consolidated market knowledge, not a single vendor's product: the same architectural principles apply whether the metrics ultimately surface in a traditional BI tool, an ad-hoc exploration workspace, or an AI analytics agent.
Dimensional modeling
Facts and dimensions, star-schema design, as the foundation for consistent metric behavior.
Centralized metrics
One definition per metric, reused everywhere. No re-derived logic hides inside a single dashboard.
Governance
Versioning, data-quality tests and documented lineage for every metric definition.
Decoupled presentation
Dashboards consume validated metrics; they never recompute business logic in the visualization layer.
Want to see this applied to your own dashboards?
We can walk through a real report from your organization and show exactly where our design principles would change it.