Building a Multi-Panel Analytics Dashboard: Coordinating Bar, Scatter, and Bubble Charts in One View

Why one chart is never enough for a siting decision

An investment or siting decision — where to expand a network of physical assets — usually depends on at least three independent variables: expected return, growth headroom, and how saturated the target market already is. Any single chart type can show one of those cleanly, but none can show all three without becoming unreadable. The solution used by most serious analytics dashboards is not one clever chart but a coordinated grid of several simple ones, each answering a different sub-question, laid out so the eye can cross-reference them.

A concrete four-panel layout that comes up repeatedly in site-selection work: a ranked horizontal bar chart of return-on-investment by candidate location, a bubble scatter of growth potential against current scale, a bar chart of market saturation, and a bar chart of projected profit. None of these panels is complicated on its own — the value is entirely in choosing which four questions to ask side by side and keeping their encodings consistent with each other.

Ranked horizontal bars for the headline metric

When the goal is "which options are best," a horizontal bar chart sorted from highest to lowest value beats a vertical bar chart or a table, for a simple ergonomic reason: labels (district names, product names, whatever the categorical axis is) read naturally left-to-right on a horizontal bar, whereas a vertical bar chart forces rotated or truncated labels once there are more than five or six categories. Sorting by the metric itself — rather than alphabetically — turns the chart into a ranking at a glance; the reader doesn't have to scan for the tallest bar, it's already at the top.

Colour-scaling the bars by their own value (a light-to-dark gradient tied to the same metric being plotted) adds a second, redundant encoding of the same information. That redundancy isn't wasted — it's what lets the chart still communicate correctly even if the reader is scanning quickly or the chart is shown small in a report thumbnail.

Bubble scatter: encoding a third and fourth variable without adding an axis

A standard scatter plot has two dimensions: x and y. A siting decision usually needs at least four: growth potential, current scale, revenue per unit, and total revenue. Rather than building four separate 2D charts, a bubble scatter reuses the same x/y plane and adds two more channels — marker size and marker colour — to carry the extra variables without adding another axis.

In the four-panel layout, growth potential sits on the y-axis and current number of installed units sits on the x-axis, immediately separating candidates into quadrants: high growth with few existing units (the most attractive, underexploited candidates) versus high growth with many units already in place (already-captured opportunity). Bubble size then encodes total revenue at that location, and bubble colour encodes revenue per unit — so a large, dark-coloured bubble in the top-left quadrant is doing all the work a paragraph of prose would otherwise need to say: "this location has strong per-unit economics, meaningful scale already, and more room to grow."

Keeping the panels visually and numerically consistent

The grid only works as a coordinated view — rather than four unrelated charts glued together — if a few consistency rules are enforced across panels. The same subset of candidates (e.g. the current top ten by the headline metric) should appear across all four panels, so a reader can trace one specific location's row in the bar chart to its bubble in the scatter plot without the two charts silently showing different populations. Colour scales should be used purposefully and not repeated for unrelated variables — reusing a green-to-red scale for both "return" and "saturation" in different panels, where green means something different in each, actively misleads a reader skimming quickly.

Axis titles, units, and number formatting (percentages with one decimal, currency with thousands separators) should match across panels exactly, because dashboards get read fast, and any inconsistency here is what erodes trust in the whole thing even when the underlying numbers are correct.

From dashboard to decision

The point of the exercise isn't the chart grid itself — it's compressing a decision that would otherwise require reading four separate tables into something scannable in under a minute. A well-built version of this layout should let a reader answer, without touching a mouse: which three locations have the best return, which locations combine strong growth with low current saturation, and whether the highest-ROI location and the highest-growth-potential location are even the same place (they usually aren't, which is itself the useful insight — it's the tension between "safe now" and "best long-term bet" that a single-metric ranking would have hidden).