Theme 072 · Data storytelling

Streamgraph River

The organic, river-like streamgraph made famous by the 2008 box-office graphic: stacked layers flowing around a wobbling centre line. Your metrics become one living current; Pulse is where the streams braid together, and the anomaly is the checkout layer visibly pinching and turning coral after 09:42.

readability ●●○trust ●●○effort med

1 · HOOK0:00–0:08

Layers flow in from the right edge like a river entering frame, grey on grey; a thin coral thread rides inside one layer and the italic Friday lands on the downbeat before the drop.

2 · REVEAL0:08 (drop)

On the drop the grey layers thin out left to right and a single teal stream swells from the middle; the Acme mark floats onto the stream on beat 2 and the wordmark rises.

3 · PROOF MOMENT0:12–0:32

The four named layers flow steadily left to right; the dashed 09:42 guide drops on the snare and the Checkout layer pinches and turns coral downstream, then the decision strip slides up.

4 · CTA0:36–0:45

A calm four-tone teal river rolls through the lower half; the headline lands on the downbeat and Start free fills on the last bar.

COMPONENT KIT

Stream layer — stacked organic band with paper-coloured seams
Silhouette baseline — centre-weighted wiggle so the river breathes
In-stream label — italic serif name set inside its own layer
Event guide — dashed vertical with mono time + deploy id
Pinch segment — the anomalous stretch of a layer recoloured coral
Decision strip — rule, −18%, context, Roll back / Acknowledge

MOTION · MUSIC

River scroll — whole stream translates left continuously, 1 frame-width per 4 bars (8 s at 120 BPM), linear
Swell / thin — layer thickness morphs with sine.inOut 1 s (two beats); the teal swell peaks exactly on the drop
Pinch — coral region wipes downstream from the 09:42 guide with power2.out 0.5 s on the snare

WHY / RISK · ADAPT IT

Streamgraphs feel alive and premium — data as a natural current — which suits an always-on watcher. Risk: exact values are hard to read, so the −18% card must carry the precision.

Adapt to your product: Stack your product's parallel volumes (traffic sources, job types, plan tiers, model calls) as layers; name them inside the stream and let the anomaly be the one layer that pinches after an event.

References: leebyron.com/streamgraph/ (Byron & Wattenberg's streamgraph paper, used by the NYT for 21 years of box office)