# Down Index methodology

Every confirmed incident gets a score from 0 to 100, from five parts computed from the incident's own facts:

- **Impact (30)**
- **Breadth (20)**
- **Duration (15)**
- **Reach (20)**
- **Acknowledgement (15)**

- Impact: the worst impact the provider posted (major outage 30, partial outage 18, degraded 8).
- Breadth: the share of the provider's components affected, or the number of regions (5+ regions is full marks).
- Duration: log-scaled minutes from start to resolution; 12 hours is full marks.
- Reach: how many builds depend on the provider (weekly registry downloads of its official SDKs plus the catalog providers that run on it), relative to the most-depended-on provider.
- Acknowledgement: minutes between down.guide's first corroborated evidence and the provider's first post; 60 minutes is full marks.

Maintenance and narrow-scope incidents (one country, region, site, carrier or asset) are never scored. History imported from provider archives is scored the same way, with the acknowledgement part at 0 (it can't be measured after the fact). Weekly tables freeze when the ISO week ends (UTC). This week: https://down.guide/leaderboard/2026-W45

---
down.guide reads providers' official status pages every minute and adds its own evidence, always labeled. Free, no account. Agents: MCP at https://down.guide/mcp, REST at https://down.guide/api/v1, feeds at https://down.guide/feed. Built by Sazabi (https://sazabi.com/?utm_source=down-guide&utm_medium=markdown).