FinOps — Unit Economics
Context-driven cloud economics.
In the cloud cost = usage x rate. When we are looking to optimise cost we seek to decrease these inputs. Unit economics when applied to FinOps is a concept that helps contextualise, relativise and provide more meaning to costs. It boils down to cost / <something-useful> .
Example
Imagine that all-in it costs Apple something like 100 billion dollars to make iPhones in a year. That is a tremendous cost in isolation. But when you contextualise by adding that means producing something like 200 million devices then the unit cost per device is $500. So when they sell all of those devices for around $1,000 they end up with 200 billion in revenue and 100 billion in profit. Unit costs anchor to context and make numbers abstract numbers more useful.
My experience
For the past few years my work has been providing an internal self-service container hosting product as part of a cloud platform team. We build and enable and support users (who are internal application development teams) to deploy and manage their own instances of the platform.
To continually holistically develop, maintain and improve this we require a team to handle all of the different concerns where we have our own development and test infrastructure costs.
For simplicity imagine it cost us $100,000 a year in cloud resource cost to do this. If / when we had only 1 team using the product it doesn’t paint a great picture. The unit cost to produce is $100,000! But imagine you serve 50 teams. Suddenly the unit cost to produce becomes around $2,000 which starts to sound a lot more reasonable.
In our case we could consider other units. Maybe per cluster or per production cluster. Each team is a different size so maybe we could work out the number of developers we support by adding that up and dividng by that. It just needs to be useful and provide the right context for measurement over time.
In our case we find that our cost to produce is basically stable and we are currently able to support more teams without cost increasing so we can deliver more value by onboarding more teams. At a certain scale this might not remain true. But currently it’s our story and it’s really important that we ourselves understand this!
Story
Teams should be able to clearly tell a story around their costs in relation to their product objectives. If cost to produce is reducing whilst maintaining the same level of service and quality then that’s great and you should be able to point to positive actions you have taken. If cost to serve has decreased due to performance optimisation then yay! If cost to serve is decreasing due to decreased users then perhaps not so much.
If cost is increasing then you should also be able to point to positive actions you have taken in terms of intentional, conscious investment in the cloud. Your team might have grown so that you can do more requiring more infrastructure. You may be doing more to improve the product for users. You may be doing more to improve technical quality. You may simply be needing more compute and services to support user growth.
If cost is increasing but products are not improving that is something to be analysed. If users / usage is not increasing or something like revenue per user then that is also a problem to be solved.
Unit economics provide the right objective lens for viewing the real value return and progress of your initiatives.
Terms
Terms used throughout the post:
