Building a Data Stack: Beyond Series A

2 min read
Thumbnail for Blog Post - Building a Data Stack: Beyond Series A || blog/data-stack-series-b/datastack-series-b.jpeg

Scaling companies is hard. If you have raised series A a while back, and are now gearing up for your next round of funding, then a lot of things will be on your mind— not in the least how to manage all the challenges of your growing team!

Data might be less of a priority, but you are getting to a vital stage in growth where it can make or break your business…

This is our advice for what to think about to avoid some of the big pitfalls you will encounter. It is written for venture-backed startups between 20 and 50 people.

  • DO ✅: domain-driven data model with clear idea of business value that you can realise (same as around the series A but even stronger here — know how you measure success!)
  • DO ✅: clear tiering of data warehouse in staging/domain/presentation layers with sequencing of the different data models — good housekeeping in how you store data is essential in fast-growing (data) teams
  • DO ✅: good housekeeping as well in model engineering (regression/unit/integration tests, peer review in development, repository — use Elementary) and model performance (careful with those join assumptions!)
  • DO ✅: when building charts, use presentational tables rather than querying the raw data directly — minimise the modelling that happens in the viz tool
  • DO NOT ❌: At this stage you are going to have more and more stakeholders demand more and more reports. Do not rush things! Do not build charts or derive insight before you have explored the data first.
  • DO NOT ❌: overhire your data team, you will probably have started with an analyst and an analytics engineer and you should make sure they are generating value first before adding other team members. Remember that adding more engineers make late projects even more late :)
  • DO NOT ❌: think you can get by without proper documentation (“I’ll do it later” turns into never) — focus on it NOW! Documentation sprints (combined with code reviews) are a very efficient way of getting ahead.