How we track customer costs in Mixpanel

Mixpanel undertook an initiative to better understand and attribute the costs associated with their customers' data usage, particularly focusing on data ingestion, storage, and compute resources. They realized that while data ingestion and storage costs are directly proportional to the amount of data sent by customers, compute costs are more complex due to the flexible nature of queries, which require varying compute resources. To address this, Mixpanel developed a method to attribute compute costs by dividing the day into two-minute buckets and identifying peak customers, allowing for a more precise cost allocation. This approach enabled them to visualize and share cost data across the company, providing several operational benefits. These include helping support teams quantify customer costs more accurately, allowing infrastructure teams to prioritize efficiency improvements, and enabling the business team to test pricing models that align with customer value. This democratization of data within Mixpanel has streamlined processes and empowered various teams to make informed decisions without relying solely on the infrastructure team.