Use this when: You are starting from zero (or close to it) and need to get data flowing from your product into a place where people can actually look at it.
You're done when: You have five core events flowing through a CDI tool into a warehouse, connected to a BI tool, with a live dashboard the team can access.
The Sequence
Template
Buy the Boring Stuff
There is a strong temptation to build custom data infrastructure. Engineers love building tools. The problem is that building a custom event pipeline, a custom warehouse, or a custom BI layer is a full-time job, and you already have a full-time job building your product.
Monica Rogati's hierarchy of data needs makes this clear. The bottom layers of the pyramid, collection, storage, and movement, are infrastructure. They should be solved problems. Segment exists so you do not have to build an event router. BigQuery exists so you do not have to manage a data warehouse. Metabase exists so you do not have to build a charting tool. Buy the boring stuff. Use managed services for every layer of infrastructure. Save your engineering time for the top of the pyramid: the questions that are unique to your business. The analysis, the experiments, the insights that no off-the-shelf tool can generate for you.
The one exception: if your data has strict compliance requirements (healthcare, finance), you may need more control over infrastructure. In that case, RudderStack or Snowplow give you self-hosted options. But even then, you are buying a tool, not building one from scratch.
Example
A B2B project management tool had Google Analytics on the marketing site and nothing on the product. They picked RudderStack (free tier), BigQuery (free tier for their volume), and Metabase (open source, self-hosted). Total cost: $0. Setup took four days. Their first five events: account_created, project_created (activation), task_completed (core action), subscription_started (payment), and session_started with a 7-day lookback (retention). The first dashboard showed daily signups, signup-to-first-project conversion (their activation rate was 22%, which they had never measured before), daily task completions, and a simple week-1 retention curve. Within a week of launching the dashboard, the team noticed that users who created a project within the first 30 minutes had 3x higher retention. That insight shaped the next two months of onboarding improvements. None of it required a custom pipeline. It required four days and three free tools.
Written with ❤️ by a human (still)