SLOs that survive your next reorg
Reorgs shuffle names on call; SLOs should anchor on player journeys and services, not individuals.
SLOs tied to hero engineers rarely survive reorgs. Anchor measurements on journeys like login, match start, or inventory sync, then map owning services. When teams split, dashboards still make sense because the journey persists.
Burn-rate alerts need humane routing. If every warning pages on-call, engineers learn to ignore them. Delayed notifications and multi-window thresholds should be explicit decisions, not accidents left over from a template.
Training decks matter. New engineers should spend their first week understanding which SLOs tie to revenue or retention events. Without that context, dashboards become wallpaper.
Review SLOs quarterly with product in the room. Sometimes the right move is to loosen a target because the feature changed, not because infra failed. Document those shifts so future audits do not misread them as regressions.