Creating the Narrative of a Data Story

When dealing with sensitive and difficult topics, a closely-related alternative is to use the decision tree structure.

The hero's journey

The hero's journey is the most famous narrative structure (story arc) used in storytelling. Think Luke Skywalker:

• Call to adventure: R2D2 plays Princess Leia's message
• Supernatural aide & Helper: Obi-Wan Kenobi
• Abyss: Almost squashed in the trash compactor
• Return: blows up the Death Star 

Pixar has its own variant of this, the Story Spine:

• Once upon a time...
• Every day...
• Until one day...
• Because of that...
• Because of that...
• Because of that...
• Until finally....
• And every since then...
• The moral of the story is...

It is quite common for people to use a hero's journey structure when presenting data, where the hero is the data analyst, overcoming challenges, and then, at the final stage, figuring out what the data means and coming up with recommendations. Such a structure involves presenting lots of facts and then at the end summarizing the key conclusions and recommendations. In the case of the margarine recommendation pyramid, for example:

Mousetrap

Such a story structure is a form of a mousetrap. As discussed in How to Communicate The Story in your Data, a requirement for an audience when looking at data is to know how the data is going to be used.

Consider the finding that "Only 10% of consumers indicated price was the 'most important consideration when buying. If somebody presented this result to you without the context of the recommendation, you might think "So what, I don't really care, and therefore can't be bothered to stress test it." Like the mouse, you eat the cheese, not aware that this is a trap. You silently leave the data unchallenged.

But, if you know that this fact is being used to support a recommendation to raise prices, it completely changes the way you look at it. If your intuition is that raising the price is a bad idea, you're going to attack this fact and make sure it's solid before relying on it in your planning.

Lead with the recommendation

The way to avoid creating the mousetrap is to lead with the recommendation and its logic so that the audience can truly stress test the data and buy into the recommendations, or, work out if the data and consequently the recommendations are not sound. This structure starts with the recommendations and then progressively adds detail.

Use decision tree argument structures for bad news

Sometimes data stories involve recommendations that are known to be unpalatable. In particular, recommending not to launch a product that has been tested often results in the messenger being shot. Consequently, while the following slide may be a paragon of virtue as a clear way of introducing the findings of a study, it may also be career limiting.

The fix for this problem is to use the decision tree argument structure. From an analysis perspective, we still start by creating the recommendations tree, as this forces the presentation to be useful. However, when we craft the narrative, we instead:

1. Create a decision tree.  An example is below.
2. Introduce the decision tree at the beginning of the report, explaining to the audience that it is the framework that is being used to evaluate the data. 
3. Walk the audience through the data, each time letting them work out, using the decision tree, the implications.

While from the perspective of clarity this is almost as good as starting with the recommendations, it has the virtue that it gives the power to work out what to do to the audience, thereby avoiding the potential for conflict.

Next

Techniques for Propelling a Story Forward