Observability
SLOs before dashboards
Most observability spend is decided before anyone agrees what good looks like. That ordering is why the bill grows and the signal does not.
6 min read
Practice 06
Observability Assessment
See your systems clearly without paying twice for telemetry you never query.
Engagement shapes
Cost and coverage review
2 weeks
Full assessment
4 to 6 weeks
Migration delivery
Priced per scope
SLO workshop
3 days
Who this is for
- Platform and SRE leaders facing a renewal with a larger number on it
- Teams carrying an alert volume nobody can act on
- Organisations locked into one vendor with no portability plan
The problem
Telemetry volume grew every year. Signal did not. The bill tracks ingestion, and ingestion tracks whatever the last team instrumented by default.
Alert fatigue is the visible symptom. The underlying cause is usually that alerts were written against symptoms rather than against a service objective anyone agreed to.
By renewal time there is no portability story, so the negotiation happens with no leverage.
What we do
The workstreams in an engagement
01
Telemetry inventory
Coverage and gaps mapped by service and by signal, including what is collected and never queried.
02
Signal strategy
Logs, metrics, traces, and profiles assigned to purposes and retention tiers rather than collected uniformly.
03
OpenTelemetry migration path
Sequenced by service, with vendor portability assessed as an explicit outcome.
04
SLO design
Objectives and error-budget policy for the services that matter, written with the people who carry the pager.
05
Alert quality review
Noise, actionability, ownership, and routing, with a before-and-after volume estimate.
06
Incident and on-call review
Response process, escalation, and measured toil.
07
Cost and cardinality analysis
The specific drivers named, per service and per label, not a blended average.
08
Tooling evaluation
Consolidation options with the trade-offs stated, including the cost of staying put.
09
AI and LLM observability
Token cost, latency, quality signals, and trace linkage for model-backed services.
Deliverables
What lands on your desk
You keep all of it, including the method behind it, so the work can be repeated without us.
- Coverage matrix by service and signal
- SLO catalogue and error-budget policy
- Alert rationalisation plan with a before-and-after volume estimate
- OpenTelemetry migration plan with sequencing
- Cost model with modelled savings and the assumptions behind them
- Tooling recommendation with the trade-offs stated
Standards we work against
- OpenTelemetry specification
- Google SRE workbook SLO practice
- Prometheus and OpenMetrics conventions
What we do not do
- We do not recommend a platform migration where tuning the current one is cheaper.
- We do not take commission from an observability vendor.
Engagement shapes
Three sizes, not one package
Durations are indicative and depend on estate size. Scope and price are fixed in writing before the engagement starts.
| Shape | Duration | What you get |
|---|---|---|
| Cost and coverage review | 2 weeks | Cost drivers, coverage gaps, and quick wins |
| Full assessment | 4 to 6 weeks | Coverage matrix, SLO catalogue, alert plan, and cost model |
| Migration delivery | Priced per scope | OpenTelemetry pipeline in production |
| SLO workshop | 3 days | A first SLO catalogue written by your own teams |
Related insights
Our thinking on this
Next step
Start with a briefing, not a proposal.
Thirty minutes on observability assessment. We will tell you whether we are the right firm for the problem, and who to talk to if we are not.