Estimated reading time: 5 minutes
Prerequisites: Basic Charts in Define, specifically the Data Labels section
What you'll learn
By the end of this article, you'll understand:
- The core rule that determines how percentages are calculated in Basic Charts, Pie Charts and Stacked Bar Charts.
- How chart configuration with dimensions, pivoted dimensions, and multiple metrics affects the calculation.
- Which actions cause percentages to recalculate, and when that can produce misleading results.
The core principle
When you display percentages in a Basic Chart, One Model calculates each bar's share of a total. The key question is: what counts as the total?
The answer depends on your chart configuration:
| Chart configuration | Percentages calculated across |
| Pivoted Dimension or Pivoted Column present | The pivoted dimension — summed per row |
| No pivoted dimension (Dimensions only) | Each column independently. Stacked bar charts and pie charts always calculate percentages across Dimensions. |
| Multiple metrics with a Pivoted Dimension | Each metric by its pivoted dimension — summed per row |
| Multiple metrics with Dimensions only | Each metric independently — summed per column |
The sections below walk through each configuration with examples.
How configuration affects the calculation
Pivoted Dimension only
When your chart contains a Pivoted Dimension and no other dimensions, percentages are calculated across the pivoted dimension - that is, each bar shows its share of the row total.
A Pivoted Column behaves the same way in charts, even though its tabular format defaults to columns. Each bar's percentage reflects its share of the Pivoted Column total.
Dimensions only (single column result)
When your chart contains only non-pivoted Dimensions - resulting in a single column - percentages are calculated per column. Each bar shows its share of the column total.
Stacked bar charts always show percentages per Dimension. This can also be represented as a stacked bar chart.
The same applies for percentages in pie charts. Pie charts are limited to one metric and Dimensions only.
Dimensions and Pivoted Dimensions together
When your chart combines Dimensions with a Pivoted Dimension, percentages are calculated across the Pivoted Dimension. The Dimension defines the rows; the Pivoted Dimension defines what the percentage is calculated against.
For example, if Organisational Unit is your dimension and Gender is your Pivoted Dimension, each bar shows Gender's share within that Organisational Unit - not its share of the overall total.
Another way to represent the same data but with a different visual is using a stacked bar chart. In a stacked bar chart, percentages are calculated per Dimension. Each column represents the full total per stacked bar.
Chart with Dimensions and Pivoted Columns
Percentages are summed per pivoted columns which in this case is also Gender (grouped columns)
Multiple metrics with a Pivoted Dimension
When your chart has multiple metrics alongside a Pivoted Dimension, percentages are summed up per metric across the pivoted dimension (row). Each metric's bars sum to 100% independently of the other metrics.
Multiple metrics with Dimensions only
When your chart has multiple metrics and only Dimensions, percentages are calculated per metric per column. Each metric sums to 100% independently. The same is true on stacked bar charts. Pie charts only allow one metric per chart.
Things that affect your percentages
Percentages in One Model are always calculated from the data visible in the chart at the time of viewing. Several things can change what data is visible and therefore change how percentages are calculated on basic charts (including stacked bar charts) and pie charts.
Hierarchy levels
If your chart includes more than one level of a hierarchy, the percentage calculation includes data from all displayed levels. This can produce results that are incorrect.
For example, if Level 1 contains a parent node that rolls up all Level 2 nodes, the parent will appear to represent a large share of the total because it is being counted alongside its own children.
To display meaningful percentages in a hierarchical chart, include only one level at a time.
Drilling into hierarchies
When you drill into a hierarchy, the chart updates to show only the data at the drilled level. Percentages recalculate based on the new visible dataset, not the original full dataset.
This is expected behaviour — the percentages reflect what is shown, not what exists above or below the current view.
Filters
When a filter is applied, percentages recalculate based on the filtered dataset. In most cases this is the intended result, but it is worth being deliberate about what your percentage represents after filtering.
For example, applying a Top 5 filter means percentages reflect each item's share of the top 5 results — not its share of the full dataset. Depending on your use case, this may or may not be what you want to communicate.
Exception — legend deselection: Clicking a series in the chart legend hides it visually, but the data point remains part of the underlying dataset. Percentages continue to be calculated from the original dataset, including the hidden series.
Exports
| Export format | Percentage labels included |
| Image | Yes, if Data Labels are enabled |
| PowerPoint | Yes, if Data Labels are enabled |
| CSV | No |
Decimal places
Percentages inherit the decimal formatting of the underlying metric. If your metric displays two decimal places, your percentage will also display two decimal places.
Quick reference
| Scenario | Watch out for |
| Multi-level hierarchy in one chart | Parent nodes inflate percentages — use one level at a time |
| Top N filter applied | Percentages reflect the filtered subset, not the full dataset |
| Drilling into a hierarchy | Percentages recalculate at each drill level |
| Legend deselection | Visual only — percentages still include hidden series |
| Exporting to CSV | Percentage labels are not included in the export |
Related articles
- Basic Charts in Define - how to enable Data Labels and choose a format
- Guide to Formatting Data Labels - token reference for customising label display
- How-to Create a Pie or Donut Chart in Explore
Comments
0 comments
Please sign in to leave a comment.