Three linked reads of the same transition tree where the flow goes over real time, which destinations are worth the friction, and what a whole journey costs end to end. Every number traces back to the column map.
A Sankey laid over a real time axis. Horizontal position is cumulative durationDays from the root, so distance means years, not tree depth. Ribbon thickness is conserved flow; hatching is resistance; colour is the AI band of the destination.
The quadrant, rebuilt. Resistance against transition-weighted AI impact, bubble area = flow, rim arc = likelihood. Twenty-five names cannot be de-collided in this space, so bubbles are numbered and the names live in a key grouped by quadrant — which turns the chart into four decisions instead of a cloud of dots.
Every root-to-leaf journey as one decision. Compound likelihood is the product of the hop likelihoods; the gauntlet bar spends its width on durationDays per hop and its colour on that hop's band — so you can see when in a career the exposure lands.
One question: you need more Data Analysts. Do you hire them from the market, or convert people you already have? Enter your headcount in each feeder role and this works out who can realistically make the move inside your deadline, what you'd have to teach them, and what it costs either way.
| Feeder role | You have | Skills match |
Could convert | Take | Time to convert |
Cost each | Call |
|---|
The provenance table as a live inspector. The right column shows the value for the currently pinned role — — — so the derivation rule and the number it produced sit side by side.
| Column | Source | How | Live value |
|---|
Band thresholds follow the existing report model — <20 insulated,
20–49 shifting, ≥50 exposed — matching AI_DIST_BAND_LOW /
AI_DIST_BAND_HIGH in gen-report-model.js, so a role sits in the same colour
here as it does in the AI distribution chart.