Building a Data-Driven Culture Without Dashboards

Our Head of Data, Brittany, gave a talk about building a data-driven culture without dashboards.

Building a Data-Driven Culture Without Dashboards
Photo by Alexander Rotker / Unsplash

A few weeks ago Brittany, our Head of Data, gave a talk at Brooklyn Bridge Ventures on how to be data-driven without dashboards.

Building a Data-Driven Culture Without Dashboards

"Dashboards are really really good at telling you what happened"

Most companies start with dashboards first. They’re excellent at showing what happened but don’t help as much when making decisions.

Some problems with dashboards:

  • Don’t provide context (what does it mean if a number moves up or down?)
  • Not suited for exploring data (have to constantly add new filters, update SQL, etc to explore)

Brittany suggested a couple of approaches to address these problems and help companies get more out of their data.

Live Analyses in Story-like format (Data story-telling)

A live analysis is basically a story told with data. It focuses on a specific question and includes context and data visualization to support a specific conclusion.

The idea is to narrow in on a specific problem and outcome instead of answering a lot of unfocused questions. Focusing on a goal and specific recommendations leads directly to making a useful decision

Activity Schema for Insight Discovery

Decision making is important, but so is answering ad-hoc questions. Companies also need a way to explore data quickly and work through hypotheses.

This requires a flexible data model that doesn’t require work to support new data questions. Brittany points out that a traditional star schema model can get a bit out of hand as a company grows. An activity schema is excellent as data grows and requires more concepts to be represented. Since it uses flexible building blocks to model data it grows more naturally, remaining manageable even as the company scales.

Ultimately Brittany believes that dashboards still have a place as long as they’re not used beyond their capabilities. Instead, think about the appropriate tool for sharing context or doing insight discovery.


Check us out on the Data Engineering Podcast

Find it on the podcast page or stream it below