Cloudthread Y Combinator
December 7, 2023

Cost Optimization Workflows

This blog entry discusses the challenges of designing an effective cost optimization workflow, emphasizing the importance of Distributed Cloud Cost Ownership.

Designing a good cost optimization workflow isn’t easy. There are many things to think about: stakeholders, cadence, steps, outcomes, tooling, etc. Generally designing any good workflow requires time and effort, and even with the best intentions a lot of workflows still don’t achieve their desired results. At Cloudthread, we are proponents of Distributed Cloud Cost Ownership, and in this blog article we’re going to explore how to think about the cost optimization workflow within this framework, as well as some pitfalls and some low hanging fruit that will make any existing workflow more successful.


Considering who needs to be involved in the workflow seems like an obvious place to start, but skipping this step with the assumption that it’s clear can be a mistake. Too often companies get this wrong, and find that a key person wasn’t involved, wasting time and money. Generally this will depend on your organization or team structure, but we’ve found that to successfully implement Distributed Cloud Cost Ownership for cost optimization there are three main stakeholders: cloud cost owner, engineer, and organization leadership. These may collapse into less than three distinct people (and at a startup may even all be the same person!) but even in this case, separating them out mentally is valuable. Most commonly people neglect leadership buy-in, so make sure there’s a clear communication line to organizational leadership. The cloud bill owner is responsible for closing the loop between proposed and completed engineering action, the engineer is responsible for change evaluation and implementation, and the organization leadership is responsible for maintaining organizational buy-in and validation of efforts (note: this cannot be overlooked – good leaders should always explicitly validate positive outcomes and/or effort).


When we think about the cadence of the workflow, we want to align it with the overall engineering cadence already implemented. If you aren’t already organizing your engineering work into defined chunks, then we’d recommend an every two week cycle to start with. The goal with defining the right cadence primarily is to create regularity for the workflow that leads to efficiency and learned capabilities, while simultaneously not being burdensome. Not frequent enough and there’s time wasted getting back up to speed, not to mention cost reduction work languishing in backlogs tied to resources eating up budget. Too frequent and too much time is being spent for marginal benefits. If you’re the cloud bill owner, we recommend maintaining a prioritized backlog that is still visible to the engineers. We have found that engineers will sometimes grab unscheduled work from the backlog if they move faster than anticipated or see similar tasks (upside), but don’t view a backlog as burdensome (minimal downside).


Deciding the right set of steps for your cost optimization workflow is connected to cadence as the steps must be doable within the cadence of your workflow. A mistake we see companies make is not allowing for a communication buffer within the steps, which causes the process to break down as the discrete nature of the workflow cannot be maintained, and things start to overlap. For Distributed Cloud Cost Ownership there are least three steps: discover, deliver, and track. Discovering different ways to save comes first, delivering these opportunities (ideally prioritized along impact, difficulty, and execution time) to the engineering actors comes next, and then closing the loop and tracking overall performance comes last. Breaking these three steps into sub-steps is up to you, but we advise against too much back-and-forth within the deliver step in particular. A good process should make task completion easy once approval is in place – in other words: don’t snatch defeat from the jaws of victory.


On this topic in particular, Cloudthread has a slightly different approach than other companies. We believe that task completion (i.e. changes made to save money) is the key metric to be monitoring across the workflow. Generally speaking, it is the job of the cloud bill owner and/or engineer (depending on your particular steps and structure) to prioritize the work in a way that means optimal task completion leads to optimal savings. In other words, the leading indicator (work done) becomes a good predictor of the lagging indicator (money saved). A solid prioritization step in your workflow creates down-funnel efficiency by allowing actors to focus on making changes, not figuring out the best action to take. Focus makes the desired outcome more obvious, which means tracking it becomes useful and meaningful. If in doubt, tracking total savings is also an obvious choice, but we have found that an effective repeatable workflow benefits from effectively correlating leading and trailing indicators. To some extent, we think this is the definition of an effective workflow. Lastly, be sure to celebrate positive outcomes and discuss negative ones in a constructive way! Too often companies don’t spend even minimal time celebrating, which we think is a wasted chance to create motivation and have some fun.

Tooling… or How Cloudthread Can Help

Cloudthread is the first platform addressing these points explicitly, and it’s our mission to be the best platform for creating a continuous, stable, and effective workflow for cost optimization. We provide all the tools you need to perform each part of the process. Further, we’ve specialized in Distributed Cloud Cost Ownership, which we think is a powerful framework nearly all companies can benefit from.

Contact us today to learn more about how Cloudthread can transform your cloud cost optimization workflow.

Make cloud costs a first class metric for your engineering organization.
Copyright © 2024 CloudThread Inc.
All rights reserved.
Copyright © 2024 CloudThread Inc. All rights reserved