Yesterday, while writing about the difference between cloud cost optimization (observability) and cost cutting (accounting) on Twitter, I mentioned the importance of tagging as a key element of a spending observability strategy. Let’s dive into this using a real-world example.
Angry Chocolates International (ACI) [real company, actual name obscured] has a deep investment in Azure Databricks analytics – anecdotally, (based on my experience) one of Azure’s most popular offerings. ACI uses Databricks to statistically determine how their consumer products are doing and where to divert investment.
The monthly runtime cost of the production Databricks environment was 500k & climbing. This northward moving cost trend raised Finance team eyebrows but beyond that simple metric, critical info was missing: what is the relationship between this spending and revenue/budget targets of lines of business using Databricks?
The way to obtain this insight – which takes an org into the area of unit economics (closer, at least) is via tagging. Consider the visualization below, which shows how I approached this project from a business insight POV:
Out of habit, IT applied technical tags to the Databricks environment (PROD, DEV, etc.) – This provided no information to Finance about the value of the investment relative to organizational goals. I introduced a business-focused tagging strategy using this workflow:
The difference between the types of data visualizations that are possible using technical vs. business tags is shown in this graphic (a modified version of what I shared with the client):
One of the most intriguing results of this work was how it enabled the integration of Azure assets with data from Angry Chocolate’s business performance management platform – usually quite separate pools of data that can be pivoted via shared tags: