Silja Märdla

What will really happen when you implement a semantic layer?

This is the story of a large organisation undertaking a significant challenge: solving the problem of inconsistency in analytics, reporting, data science, experimentation and AI. Bolt is the European super app that makes cities better for people by offering ride hailing, delivery and rental services to more than 200 million customers in 50+ countries. Like many other modern enterprises, we spend too much time trying to understand our own data, wasting resources and losing out on opportunities. If we don’t understand our own data, we shouldn’t expect AI to do so either. AI will only replicate our own mistakes. We need a long term sustainable solution, not a one-time action. Bolt’s plan involves defining our core business logic and metrics in the (dbt) semantic layer and developing automations to make use of that semantic layer across the different use cases of data. We have implemented a few thousands of metrics, generated pre-aggregated metric tables, synced the business logic to Looker and our in-house experimentation platform, while making sure these data products are trustworthy and maintainable. In this talk we’ll share our biggest learnings from the process. We’ll talk about some of the conceptual and fundamental questions, some of the technical details, but also about the organisational changes that have taken place. You will walk away with insight into what it really takes to sort out your data and what are the benefits you will feel in return.

Silja Märdla is a Staff Analytics Engineer at Bolt, the European super-app that offers shared mobility services like ride hailing, rentals and delivery to more than 200 million customers in 50+ countries. Over different positions as a hands on data practitioner or a manager, she has seen her fair share of data mesh and data mess. It is now her mission to figure out scalable and sustainable data practices for the hundreds of data users in Bolt. Being able to find and reuse high quality data products is what helps a data team and the company beat the competition. She is passionate about knowledge sharing, aspiring to push the whole data community forward, into a less messy future.