By David Ashirov | VP of Data, Freshly
In our recent venture into an implementation of the core principles outlined in Working Backwards (Bryar and Carr Working backwards: insights, stories, and secrets from inside Amazon) we needed to instrument visualization of our corporate business metrics. The visualizations prescribed by the authors are unusual. The main component is a line chart, but it shares an X-axis across two distinct time series aggregates: to the left of a visual gap we find six most recent calendar weeks and to the right we find twelve most recent months. The very nature of the effort required a platform for rapid deployment of metric calculations and their subsequent visualization within a seemingly endless dashboard.
In our case, we have started with what most of the readers will find in their possession — an incarnation of a modern data stack (Handy The Modern Data Stack: Past, Present, and Future). Ingestion was governed by Fivetran, which so happens to also house Fishtown Analytics’ dbt to facilitate Transformation. Warehousing and BI are handled by Snowflake and Google Looker, respectively. You may find your world to be ever-so-slightly different, but essentially we are running similar data operations. Another prerequisite we have already satisfied prior to committing to run the WBR marathon was an agreed-upon architecture for a business metric repository. The team was inspired by ideas put forward by Dieter Plaetinck (Plaetinck Metrics 2.0) at metrics20.org, at least the theory behind a concrete implementation he proposed. It is brilliant in its simplicity, extensible, and portable.
This article aims at providing a step-by-step instruction manual on setting the infrastructure necessary to expose business metrics to the enterprise following the directions outlined by the Working Backwards book authors.
Read the full article on Medium
