A score out of 100, by application and business unit
FinOps scoring turns FinOps maturity into a simple number. For each application, we review the best practices that apply to its architecture. Each has a weight from 1 to 3 based on its impact on the bill, and a status: to do (0), ongoing (25%), done (100%) or not applicable (excluded). The score is the weighted average, scaled to 100.
Aggregated by business unit, it becomes a KPI tracked at the SteerCo, next to the savings achieved: it shows whether best practices are taking hold, not just whether the bill is going down.
A grid per cloud service
The grid covers governance and every service in the architecture. A few examples of items assessed:
| Domain | Examples of best practices |
|---|---|
| Governance | Budget alert per scope, application tag linked to the CMDB, unnecessary non-prod environments deleted, logs challenged |
| VM and EC2 | Auto Stop + Start on demand, rightsizing, sizes covered by reservations, Azure Hybrid Benefit, Spot in non-prod |
| Disks and storage | Right disk tier, unattached disks, lifecycle management, S3 Intelligent-Tiering |
| App Service, Beanstalk, ECS | Plan tier, Linux by default, empty plans deleted, autoscaling in prod, Fargate for short batches |
| Databases | Rightsizing, serverless, stopping non-prod databases, geo-replication and multi-AZ challenged, provisioned IOPS justified |
| Data | Need for Databricks challenged, job clusters, auto-termination, Data Factory pipeline frequency |
| Logs | Enabled logs justified, retention, commitment tier, daily cap |
The scoring session, before go-live
The best time to score an application is just before it goes into production. We bring together the project team and the team that will run it for a one- to two-hour workshop:
- review of the architecture's components and their cost risks (a log option, geo-replication, a massive data transfer);
- creation of budget alerts and who receives them;
- a starting score and a prioritized action list, which the run team takes over.
We then rescore at every major change. Without a scoring session, bad surprises arrive with the first bill: see our real FinOps incident cases.
A cookbook to make teams autonomous
Each item of the grid points to a page of our FinOps cookbook, for Azure and AWS: why this component costs money, how to detect waste, how to fix it, with the useful queries and settings. Application teams remain responsible for implementation; we provide the tooling and support them.
Take the self-assessment
Here is a short version of our grid. Tell us where you stand on each item: the score is calculated in your browser, nothing is sent.
Reading the score
| Score | Level | What it means |
|---|---|---|
| 0 to 40 | Crawl | Fundamentals are missing: no alerts, no cost allocation, non-prod always on. Quick wins are plentiful. |
| 40 to 70 | Walk | The basics are in place, optimization is still occasional. A tracked remediation plan is needed. |
| 70 to 90 | Run | Best practices are applied. The next step is re-architecting the most expensive services. |
| 90 to 100 | FinOps by design | Standards and building blocks do the work: every new project arrives already optimized. |