Claritas One
Consulting/Cloud/FinOps & Cost Optimisation

FinOps & Cost Opt.

A shared operating model between finance, engineering, and product. We cut the bill without cutting velocity — and leave behind a governance cadence that keeps spending honest long after we leave.

The cost waterfall

Where a 48% run-cost reduction typically comes from.

A representative first-year trajectory for a mid-sized enterprise estate. Mix shifts by workload, but the sequencing is consistent — easy wins first, commitment planning second, architectural change last.

Current run-rate
100%
$18.2M annualised
Rightsizing
-12%
Compute, DB, cache
Retire & delete
-9%
Unattached volumes, dev leaks
Schedule & scale
-7%
Non-prod off-hours, autoscaling
Commitments
-14%
Reserved + savings plans
Architecture
-6%
Serverless, spot, managed services
Target
52%
$9.5M annualised

Six levers

Six optimisation levers. Pulled in the right order.

Rightsizing

Match capacity to actual load — not architect comfort.

Compute, DB, cache, and storage scrubbed against real p95 and seasonality. No blanket 10% haircut — targeted downshifts with automated rollback.

Plays in this lever

  • CPU / memory rightsize by workload
  • DB family shift (e.g. Aurora → Graviton)
  • Storage tier migration (gp2 → gp3)
  • Per-service kill switches tested

Anomaly stream — 24h snapshot

Cost as a real-time signal,
not a month-end autopsy.

A composite 24-hour slice from our anomaly detection pipeline. Routed to the owning team with context — the signal beats the surprise every time.

02:14
aws.rds.prod-eu-w1
+$412/hr vs. 14-day median
payments-platform
auto-scaled write replicas
04:22
gcp.bigquery.analytics
scheduled query cost anomaly +18%
data-platform
investigating query plan
09:47
aws.eks.cluster-prod-2
unassigned node group · $2,180 wasted
platform-core
terminate queued
11:03
azure.blob.logs-archive
hot tier retained past 30d · $740/mo recoverable
observability
tier shift scheduled
13:30
aws.s3.export-staging
80TB egress to external · review required
etl-team
approved · billable
14:48
gcp.gke.ml-training
GPU utilisation <12% for 3h
ml-platform
scheduler tuned
6 anomalies · 24h · 4 auto-remediated · 2 queued streaming

The FinOps operating rhythm

A week in the life
of a programme that sticks.

Rituals are how the savings become structural. Miss one for two months and the unit cost graph will tell you. We install them, then hand them over.

MON
09:00

Weekly cost standup

30 minutes. Engineering, product, finance. Anomalies, wins, and the single biggest lever for next week.

WED
anytime

Anomaly response

Automated alerts on spend, unit cost, and budget burn. Routed to the owning team with a playbook — not a Slack message.

FRI
14:00

Engineering enablement

Dashboards, IDE cost hints, and policy-as-code that put the cost signal where engineers write the code.

MONTHLY
T-1

Business review

Cost trend vs. plan, unit economics, forecast confidence, and the top three optimisation bets — framed for the exec team.

MONTHLY
T-3

Unit-economics scorecard

Feature and product-level cost shared with product. Cost becomes an input to prioritisation, not a year-end surprise.

MONTHLY
T-5

Forecast accuracy review

Reconciliation of forecast vs. actual at service and account level. Tightens the forecast quarter over quarter.

// exec attendance > 85% after month 3 is our tripwire

Case snapshot

“We walked in owning $18M a year in cloud spend we couldn't fully explain. Nine months later, the bill is $9.5M, the forecast is within 3%, and engineering volunteered the rituals they used to resist.”

VP Platform Engineering
Enterprise SaaS, 1,400 employees
48%
Run cost eliminated in 9 months
3%
Forecast variance month-on-month
$8.7M
Annualised savings locked in
0
Performance regressions shipped

Two weeks from audit to a savings plan your CFO will bank on.

Give us billing export access for a recent quarter. We'll come back with a prioritised savings plan, a sequencing recommendation, and the FinOps ritual set to install.