Data storytelling is the analysis—or story—that you tell with your data. What is the data telling you and why? It is a method for presenting your data to make it easier to understand and interpret. Visualizations, such as charts and graphs, are used to draw attention to key observations.
Good data storytelling allows an end user to view data in a flow that not only gives them insight to what their data means, but allows them to critically think about how that data can affect their business decisions. We talked to Mico Yuk, CEO of BI Brainz, to gather some best practices and mistakes to avoid with data storytelling in the context of embedded BI.
What are the best practices for data storytelling in the context of embedded analytics?
Mico Yuk: When telling data stories, there are three key things to keep in mind:
- Focus on anecdotes, not personas – Focus on anecdotes and not personas, as my good friend Ryan Goodman points out, to ensure that you capture the most important stories within a broad audience.
- Forget metrics, think story parts – Good stories have four parts, a goal, a KPI/metric snapshot, trends, and most importantly actions. Just randomly placing cool charts on a screen to show a specific metric adds no value. Identify clear story parts in your analytics to provide a clear path for the readers.
- Forget fancy, instead KISS – Keep It Simple Superstar. When it comes to visualization use charts that are familiar and require no translation, such as bar and line charts. Fancy charts can lead to more confusion than conclusion.
How should application teams think about data story when they’re embedding dashboard reports as compared to a more standard BI Instance? Is it different?
Embedded analytics has the advantage of being available directly in the end-user workflow process as opposed to having users access a separate application to view their data. With that said, from a visual perspective, I believe the application team and BI teams alike should focus on the four-story parts I outline as they answer the four fundamental questions users have:
- So What?
- Now What?
Depending on where the analytics appear in the workflow, ensuring that your application answers at least two of those questions will make them useful. If you can get to all four then you have actionable analytics.
Are there common data storytelling mistakes you see application teams making, and what can they do to avoid this?
Most applications that I’ve seen fall within two extremes. Focus on simplicity—some applications don’t provide enough data while others provide too much. Embedded analytics has the advantage of being more dynamic than traditional BI reports. Instead of focusing on putting everything on one screen, focus on presenting just the data that is needed at a specific point in the workflow. In addition, I don’t see enough personalization. There are opportunities to use relatable indicators like emojis to communicate binary results like up vs. down, good vs. bad, and more. Keep your analytics simple and relevant to the exact step in the workflow. Last but not least, ensure that the analytics are consistent across the process in look/feel even though the data changes. If not, the users may get confused or spend extra time trying to figure out what has changed. That’s not useful—it’s distracting.