// Humanitarians AI — Investment Landscape 01 / 02

Private funding tripled in two years —
but government still anchors the base

view: Absolute
Humanitarian AI Funding by Source, 2020–2024 — Stacked Bar Chart Five vertical stacked bars, one per year from 2020 to 2024. Each bar is divided into four segments: Government (walnut), Private Foundation (blood-red), NGO/Multilateral (dim-gray), and Research/Academic (mist). Total investment grew from $93M in 2020 to $286M in 2024. The Private segment grew fastest.
click legend to highlight · hover segment for detail
// Learn — Stacked Bar Chart 02 / 02

Two chart types, one encoding system — choosing between them

What this chart is

A stacked bar chart divides each bar into sequential segments representing subcategories of a total. Each bar's full height encodes the total value; each segment's height encodes its contribution. Unlike a grouped bar chart — which places subcategory bars side-by-side — stacking preserves the total while showing composition.

The two variants serve different analytical goals. The absolute stacked chart (default) answers "how much, and how is it made up?" The 100% stacked chart (toggle above) answers "what is the proportional composition, independent of size?" They use identical data but answer different questions.

The baseline problem

The stacked chart's structural flaw is that only the bottom segment sits on a common zero baseline. Every other segment has a different starting point in every bar, making direct comparison of non-bottom segments perceptually unreliable. The eye cannot accurately judge if the "Private Foundation" segment in 2022 is larger or smaller than in 2023 without reading the tooltip.

This is why segment order matters: put the category most important to compare at the bottom. Here, Government anchors the base — it's the largest and most stable, so its trend is easy to read. Private Foundation is second because its dramatic growth is the key story.

When to use absolute vs 100%

Use the absolute view when the total matters — when the viewer needs to know that overall investment tripled, not just that composition shifted. The total bar height carries real information that the 100% view discards.

Use the 100% view when you want to isolate compositional change from volume change. In the 100% view, Government's declining share is visible even though its absolute dollar value grew every year. Both facts are true; the chart choice determines which fact is foregrounded.

When to use alternatives

If comparing subcategory values across bars is the primary task, a stacked bar chart will always underperform. The correct alternatives: a grouped bar chart (each subcategory gets its own bar, enabling direct comparison), a small multiples line chart (one panel per category showing its trend), or a slope chart (comparing start and end points without intermediate clutter).

The stacked bar earns its place specifically when both the total and the composition matter simultaneously — which is exactly the case here, where the tripling of total investment and the structural shift in who provides it are both the story.

// Framework Reference

FT Visual Vocabulary: Part-to-whole (for 100% mode) and Change over time — Magnitude (for absolute mode). Abela quadrant: Composition (how a whole breaks into parts, tracked over time). The one design decision worth knowing: segments are ordered bottom-to-top as Government → Private → NGO → Research, matching descending average value. This ordering is data-driven, not alphabetical — it ensures the largest segment anchors the baseline where perceptual accuracy is highest, and the smallest segment floats at the top where comparison is least critical.