SLO Anti-Patterns: The Five Traps That Kill Programs

Most SLO programs that fail die from the same five reasons. The fixes are mechanical.

Trap 1: aspirational targets

The first trap is setting SLO targets that the system cannot meet. The number sounds aspirational; the team learns to ignore it within a quarter.

• Symptom. Target above measured capability; e.g. 5 nines on a system delivering 3.
• Effect. Always missed; team learns the SLO is decoration; the real number drives nothing.
• Fix. Set target slightly above measured baseline; raise as capacity grows.
• Honest baseline. Measure for 90 days before setting the target; the data tells you what is achievable.

Trap 2: no consequences

• No consequence to missing: feature work continues.
• SLO becomes decoration; nobody cares.
• Fix: defined consequences; documented; enforced.

Trap 3: gaming

Gaming the SLO is the slowest poison. The numbers stay green; reality drifts; trust dies quietly.

• Symptom. Excluding causes you control ('not our fault'); selectively redefining what counts as downtime.
• Effect. Cooked numbers toward zero risk; the SLO no longer reflects user experience.
• Fix. Include all user-facing impact; users do not care about your blame model.
• Audit. Quarterly review of exclusions; each one needs written justification.

Trap 4: scope creep

Scope creep turns a useful SLO programme into noise. More SLOs do not mean more reliability; they mean more dashboard space.

• Symptom. 50 SLOs per team; every metric promoted to SLO; nobody can name the important ones.
• Effect. Maintenance burden exceeds value; SLO updates lag reality; the team disengages.
• Fix. 3-5 SLOs per service maximum; quarterly review to prune.
• Hierarchy. Customer-facing SLOs at the top; internal SLOs as supporting indicators, not as headline targets.

Antipatterns

• Trap 5: dashboard fatigue. Engineers stop looking; recovery requires redesign.

What to do this week

Three moves. (1) Apply the pattern to your most-impactful service. (2) Measure adherence for 30 days. (3) Rewrite the policy or the SLO if the gap is durable.