How to write a data story

We’ve all heard someone say that a story needs a beginning, middle, and end. Maybe we cringed thinking about how simple it sounds. But stories DO need a beginning, middle, and end – just not in the chronological way we may think.

These three parts give stories a Three Act structure, a narrative model used by storytellers for centuries.

Act 1 (Agreement Act) sets the scene by introducing the character and providing some initial context.

Act 2 (Conflict Act) contrasts Act 1 and in doing so increases the tension. For the narrative to advance, you need contradiction.

Act 3 (Resolution Act) resolves the conflict between Act 1 and 2.

When writing stories for a business context, I prefer to use Randy Olson’s version of the Three Act narrative structure. He uses the words and, but, and therefore, to differentiate between acts. This structure is super flexible, conversational, but still advances the narrative. We can use this ABT (And, But, Therefore) Three Act structure when writing our data stories.

Always start with a high-level Three Act structure (ABT) as this will summarise the story and define its scope. If you have trouble framing your ABT in a few sentences, you may be trying to tell more than one data story at once. When you’ve settled on a high-level ABT you can begin to add further detail where required. The narrative structure remains the same, but the level of information gets deeper.

Most data stories are built around one metric. Other metrics will likely be included, but one metric will be the main focus of the story. This metric is compared either across time (in Event data stories) or across characters (in Character data stories). You can read more about finding the right metrics for Event or Character stories in the previous blog.

Event Data Stories

High-level story (ABT): This will give context to the focus metric and identify how the event has changed it. For example, below is a high-level story written to explain how climate change is affecting green sea turtles.

The sex of a baby green sea turtle is determined by the temperature of the nest it develops in. But a warming climate is altering this temperature-dependent sex determination process. Therefore, there is an immediate need for nest management strategies to avoid a green turtle population collapse.

Nested story (ABT): These expand on the overall story, while providing context and detail. For example, below is a nested narrative to further explain the first sentence of the high-level story above.

When a female green turtle reaches about 30 years old she will often return to the beach where she was born and lay her eggs. She’ll dig a pit in the sand with her flippers, fill it with around 100 eggs, then cover her nest and return to the sea. Her eggs will hatch after two months. But unlike most animals where sex is determined at fertilization, the sex of a baby turtle is determined by the temperature of the nest it develops in. Cooler nests produce more male baby turtles, and warmer nests more females. The temperature that produces an equal number of male and female turtles is known as the “pivotal temperature”. The pivotal temperature for green turtles is 29.3˚C. Therefore, green turtles have temperature-dependent sex determination or TSD.

This data story can then be visualised based on the above written narrative:

Character Data Stories

High-level story (ABT): This will give context to the focus metric and identify what segments to profile further. For example, below is a high-level story focused on the vaccine confidence metric for health professionals in New Zealand.

The New Zealand government funds vaccinations for all New Zealanders, and the majority of adults strongly agree vaccinations are safe. Most health professionals have more confidence in vaccines than the public. But the vaccination confidence of midwives and alternative medicine practitioners is low. Therefore, the type of health professional seen by someone could be an important influence on their own confidence in vaccines.

Segment story (ABT): These expand on the overall story, while providing context and detail for an individual segment. Metric comparisons to other segments can help make information relative but segment stories should focus only on the specific segment. Each segment story includes the focus metric.

For example, both midwives and alternative medicine practitioners were identified in the above high-level story. These segments can be expanded on in their own story, like the midwife story below. Comparisons have been made to nurse and GP segments to help make the training metric more relative.

Most women in New Zealand will choose a midwife for their maternity care. Midwives are qualified after three years training (by comparison, nurses also train for 3 years and GPs for 11 years). But one in three midwives doesn’t have a strong confidence in vaccines. Therefore, understanding the reasons for vaccine resistance of midwives, could help in changing the anti-vaccination attitudes of the public.

The above midwife data story could be supported with a graph like this:

Data visualisation is a popular way to tell a data story. But these visualisations should only be used if they aid in the communication of the data story. The graph you use to communicate the insight may be different to the graph used to help uncover it.

Videos to help explain these concepts can be found here.