Brent Dykes is the Evangelist for Customer Analytics at Adobe and is responsible for guiding and evangelizing the vision of Adobe’s analytics solutions. While at the Adobe Summit 2015 in London, I had the opportunity to interview him on his keynote presentation "Data Storytelling". I also interviewed him in 2014.
Nicolas Malo: What is exactly data storytelling? How is it different from data analysis?
Brent Dykes: As I mentioned in my keynote, the definition of data storytelling is "a structured approach for communicating data insights more effectively to an audience using narrative elements and data vizualisations".
I've read a lot about this topic, but never really found an indepth definition and this the reason why I came with my own defintion.
When we explore the data, we have to craft a story like a director and make tough decisions about what to include and to exclude in order to make a point. Like in a movie, you need to choose an angle. Everything is chosen for a specific purpose in order to contribute to the story.
Nicolas Malo: What are the essential components of a data story?
Brent Dykes: From my experience, these are the essentials of a data story:
1. Main point: a data story must help the audience to understand a central insight or idea. It should have an intended end point or destination that drives discussion and action.
2. Explanatory focus: an analyst provides relevant context and useful commentary on the data points. These details can accompany the data as annotations (text) or speaking points (audio).
3. Linear sequence: a data story is arranged in a linear structure where supporting points build upon each other. The layout or presentation flow should provide a clear path for the audience to follow.
4. Narrative elements: a data story borrows various techniques from literature and film. Setting, characters, plot and conflict help to engage the audience more deeply on an emotional level.
5. Visuals: charts, diagrams and other imagery are used to simplify and clarify complex ideas. Data visualizations complement the narrative and contribute both to the comprehension and memorability.
Nicolas Malo: What is exactly the "data storytelling arc"?
Brent Dykes: Basically, it is taking the story arc that it is used in literature, movies, television series, etc... I just took that and tailored it for a data story.
Every story starts with setting. That's where you are establishing the status quo, the background. And then, you have an inciting incident, where something changes in the environment (for instance: revenue goes up or down). From there, we share rising insights where we share "oh this happened, and why did this happened ?". We are like peeling an onion removing the layers and going to the big "ah ah moment".
In a novel, there will a rising action and it gets to the "climax", for instance the big battle. In our case, this is the big insight. If you can monetize, it adds an impact. If we don't fix our conversion rate, this will cost us XYZ millions dollars next year.
After then, the last step is "next steps". You just can't leave it at the climax, otherwise executives won't know what actions to take or what options they have. Our analysis needs to look at the options the business can take.
I've customized it for the data storytelling.
Nicolas Malo: What are the main pitfalls with data storytelling?
Brent Dykes: You need to know your audience. First you need to know what are the main business objectives of your audience. Next, you want to know the key questions or needs that they have. And third, you want to know how familiar is the audience with your topic. If they are less familiar, you need to provide more background and use less jargon. How are senior the people you are talking to? Last thing, am I delivering this presentation in person and am I sharing it by Email or through a portal?
How can you engage the executives with your presentation? Rather than show one time period compared to another, ask for questions and then show the results. They are actively engaged with the data. It's an opportunity to wake up and engage the audience. You can also ask them to analyze the data and then you bring in your annotations and highlights.
Exectutive are often impatient and won't give you time to tell the whole data story. So you need to provide a compelling summary of the story, almost like a teaser. Then, they may ask "Tell me more". Then, you are getting permissions to tell this data story. With an executive, you may have 4/5 stories prepared, but they won't give you time to tell every story. So, you have a summary of each story and their curiosity will decide which one you share.
Nicolas Malo: What are the main tips for data storytelling?
Brent Dykes: You don't want to overload your audience with data. You need to stage your content and not give everything at once. We need to remove the noise. As analysts we tend to include too much information on the same screen. It's too much information for people to follow. You need to redesign the standard charts in Excel. Also, you need to engage the audience with your analysis.
You need also to make the data relatable to people, for instance comparing something so that they can releate to. For instance, you could hire two full-time persons for the same cost by better managing your serach tearms.
Nicolas Malo: Thank you very much Brent for this interview!
