Summary stats answer "how are we doing?" — but they hide when things went wrong. The new Deployment Stability heatmap gives you the shape of your reliability over time: one square per day, colored by failure rate, laid out like a GitHub contributions graph. A rough week jumps off the page; a steady run of calm color shows when things are healthy.

Each cell is colored on a continuous gradient scaled to that day's failure rate — cool, low-intensity tones for stable days and hot, high-intensity tones for days with the most failures. Hover any cell to see the exact number of runs, how many completed, how many failed, and the computed failure rate for that day. Days with no deployments render in the neutral baseline color so you can tell "quiet" apart from "broken."

On the Insights page the heatmap follows whatever time range you've picked for the rest of the dashboard — 7 days, 14, 30, 60, 90, 180, or a full year — and the card resizes itself to match: quarter-width for short ranges, half-width for a few months, and full-width once you zoom out to a year. Available ranges are gated by your plan's doraLookbackDays entitlement, so a year of history shows up automatically for teams on plans that include it.

The heatmap ships in two places. On the Insights page it aggregates every workflow run across every repository in the workspace, so you can see organization-wide stability trends. On the analytics tab of each repo, the same heatmap is scoped to that repo's runs — useful for isolating which repo is dragging the overall numbers down after a bad week.