Skip to content
Saiku Cloud

Charts and visualisations

Once you have a query, click any of the chart icons on the toolbar to swap the table for a visualisation. The data underneath doesn’t change — the chart just renders the same numbers differently. Picking the right chart matters more than people give it credit for: a sunburst answers a different question than a bar chart.

What’s available

Saiku ships ten chart types plus several variants.

Bar

The default and the most generally useful. One bar per dimension value, height proportional to the measure. Use it when you want to compare values across categories — sales by region, headcount by department, error rate by service.

Multi-bar is the variant for two-dimensional comparisons — several bars per group, colour-coded by a second dimension. “Sales by region by quarter” with quarters as the inner colour grouping and regions on the x-axis.

Line

For trends over time. Drag a date level into Rows or Columns, measures into the other axis, switch to Line. Multiple measures or multiple dimension values render as multiple lines.

Use Line when the x-axis is ordered (time, days-since-event, revenue brackets) and the slope between adjacent points is meaningful. Don’t use Line when the x-axis is categorical (regions, products) — there’s no meaning to the slope between region A and region B.

Area

A line chart with the area below filled in. Same use cases as line, with the visual emphasis on cumulative totals. Best when you have one or two series; multiple stacked areas get hard to read.

Pie

The ratio chart. Slices of a whole, summing to 100%. Honest use case: showing how a single measure breaks down across one dimension when there are five or fewer categories. Beyond that the slices get hard to read and a bar chart is usually clearer.

Saiku also offers donut as a pie variant — the only difference is the hole in the middle, which gives you space for a centre label.

Sunburst

A pie chart that drills down through a hierarchy. The inner ring is the top level; each outer ring is the next level down. Use it for hierarchical breakdowns (“revenue by region → country → city”) when you want both the high-level shape AND the detail in one view.

Multiple sunburst renders several sunbursts side-by-side for small-multiples comparison.

Treemap

A rectangle-packing visualisation. Each rectangle is a leaf-level value, sized proportionally to the measure. Hierarchies nest as nested rectangles. Treemaps handle hundreds of leaf values in one view without the chart becoming illegible, which is where pie and sunburst break down.

Use treemaps for “what dominates?” questions over a large set — top customers by revenue across a thousand-customer base, top SKUs in a product catalog, etc.

Heatmap (heatgrid)

A grid coloured by intensity. Two dimensions on the axes (rows and columns of the grid), one measure determining cell colour. Use it for pattern-spotting in two-dimensional data: “sales by hour-of-day by day-of-week” surfaces patterns no other chart type does as clearly.

Waterfall

A bar chart where each bar starts where the previous one ended. Made for showing how a starting value reaches an ending value through a sequence of additions and subtractions. Classic use: revenue waterfall from gross to net through specific deductions.

Dot

A scatter-like rendering — one dot per category, position determined by the measure. Strips away the bars and gridlines for a sparser look. Useful when you want to emphasise the values themselves rather than the comparison.

Sparklines (spark bar, spark line)

Tiny inline charts designed to live alongside numbers in a table. Excellent for “trend” columns in dashboards — one tiny line per row showing the metric’s history.

How to pick

A rough decision tree:

  • Comparing across categories? Bar (or multi-bar for two dimensions).
  • Trend over time? Line (or area).
  • Composition / ratio? Pie if ≤5 slices; sunburst if there’s a hierarchy; treemap if many leaves.
  • Two-dimensional pattern? Heatmap.
  • Sequence of additions/subtractions? Waterfall.
  • Side-by-side comparison of many small metrics? Sparklines.

When in doubt, start with bar — it’s the most readable default for most analyses. Switching is a one-click affair, so trying two or three chart types on the same data is cheap.

Chart editor

Most charts have decent defaults out of the box. For fine control — colour palettes, axis labels, legend position, gridlines — the Chart editor button opens a panel with the full set of options. Saved as part of the workbook so the look persists.

Exporting a chart

Use Export to PDF from the toolbar to save the current chart as a PDF. The chart renders at print resolution. For inserting into a report or slide, this beats screenshotting.

For a chart-in-context — chart plus the table that produced it — Export to PDF captures both. The drillthrough data underneath isn’t included.

When the chart doesn’t work

A few cases where charts make things worse, not better:

  • Too many series on a line chart (10+ lines on one axis) — switch to small multiples (multiple sunburst, multi-bar) or pick the top N to plot.
  • Wildly different magnitudes on the same axis — one series in the millions, another in single digits. Pick a log scale in the chart editor or split into two charts.
  • Few data points (one or two values) — the chart is just decoration. The table is more honest.

Where to go next