Tier · T0
Ad-hoc
Telemetry exists because someone enabled an agent. Nobody can describe the data hierarchy. Pages mostly noise; MTTR unpredictable.
- Top-10 dashboards 18+ months old
- Bill grew 40%+ year-on-year
- Most alerts silenced or auto-resolved
Observability & SRE · Singapore
Observability is a definition problem,
not a tools problem.
A Singapore observability and SRE practice. We rebuild telemetry, SLOs, and incident response on systems that have outgrown the dashboards they started with.
01 Diagnosis
The maturity profile & roadmap
A 12-axis assessment that scores the seven structural dimensions of your observability stack on evidence. The panel on the right is the actual artefact from a recent anonymised engagement — read it the way a client reads it.
Dimension scores · 0-4 scale
- 01 Telemetry coverage 3/4
- 02 Alerting quality 1/4
- 03 Incident response 1/4
- 04 SLO maturity 1/4
- 05 Correlation & RCA 2/4
- 06 Tooling & IaC 2/4
- 07 Platform & culture 2/4
Tier verdict · worst-of axes
T0 · Ad-hoc
Bottlenecked by alerting quality, incident response, and SLO maturity · all scoring 1/4.
Maturity ladder
What the score actually means.
Each dimension above scores 0–4. Your tier is the worst-of axes — observability is bottlenecked by the weakest signal, not averaged across the group. The four states below are where systems land in practice.
Tier · T1
Reactive
Pages get answered. Dashboards exist. But the design is inherited rather than chosen, and cardinality grows in step with revenue.
- SLOs exist on infra metrics
- Cardinality controlled by exception
- Trace coverage spotty across boundaries
Tier · T2
Operational
Data hierarchy is intentional. SLOs reflect customer journeys. Cardinality has a budget. Cost-per-signal is a tracked metric.
- Journey-keyed SLOs in place
- Cardinality budget at the collector
- OTel-portable trace surface
Tier · T3
Proactive
Telemetry is treated as a first-class surface. Engineers check it before shipping, and the dashboard count trends down rather than up.
- Pre-deploy SLO impact reviews
- Cardinality regressions caught in CI
- Multi-vendor by choice, not lock-in
Most teams we audit start at T1. Foundational moves you to T2 in 12 weeks.
Selected engagements
Same standard. Very different starting points.
Foundational · 14 wk Foundational · 12 wk
Fintech · APAC · clearing engine
From 98.4% to five-nines on a Tier-1 clearing engine.
MTTR · min
−94%
before 252
after 14
Logistics · APAC · platform
Cut a $2.1M Datadog bill by two-thirds without losing fidelity.
Annual spend · $k
−65%
before 2,100
after 720
38 engagements · 2021–2026
Same standard. Six sectors. Thirty-six more case studies.
Fintech, logistics, gaming, healthcare, e-commerce, banking. Numbers redacted; outcomes are not.
All engagements arrow_forward
Names withheld under NDA · sectors and metrics anonymised at client request.
KT
From the founder
I'm Ken. I started Tracefox after enough years inside platform teams watching telemetry budgets compound while incidents stayed slow. Almost every observability problem I see is upstream of the tools — missing definitions, no shared vocabulary, dashboards nobody owns. That's what we fix first. Everything else is configuration.
House rule
Every service we manage emits all four Golden Signals. Every user-facing journey has at least one SLO. Every SLO has a burn-rate alert. Anything beyond that is improvement, not baseline.
Diagnostic · 2 wk · USD 18,000 fixed
Maturity profile, telemetry redesign, and a 90-day roadmap. Same price in Singapore, San Francisco, or Sydney.