In his paper, “The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information” (1956), George Miller detailed his law of human cognition and information processing and his supposition that humans can effectively process no more than seven units, or chunks, of information, plus or minus two, at any given time. The premise is that when using short-term memory, humans struggle to retain and recall more than 7 key points from a story. This means we need to consider the priority messages we want our audience to focus on and not overload the audience with all the detail.

As Sinead Jefferies, SVP Customer Expertise, Zappi, explains, clarity leads to impact and unlocks value for the audience:

“You must boil it down to a couple of really salient points and ensure that people can see the connection and why that's important. It's the classic thing - and we've been talking about this for my whole 20 years in the industry - if you go in with a 100-slide deck, nobody's going to be interested. If you go in with two or three really compelling points that people can see are relevant, they are much more likely to do something. People don't know what to do when you give them too much information.”

5 steps to ruthless data distillation

1 – Analyse the data for themes

Whilst this might seem obvious, often proper analysis of data is sacrificed in the rush to report the data. If less effort is spent on analysing for insights than on charting, the output ends up being a neat and tidy data dump! So, ditch the PowerPoint, PowerBI dashboard or the one-page summary template for a moment, and get your hands dirty with the data.

Invest time in a rigorous analysis process that allows you to:

• Immerse yourself in the data and actively read through the detail looking for initial points of interest or curiosities

• Generate a list of ideas about what is in your data describing any obvious patterns, trends, comparisons

• Sort the ideas into potential themes – looking across all the data sets to spot connections or contradictions. Use your list of ideas to create a visual map and go through each idea and decide where it belongs in the map to isolate the core themes

2 – Recode your themes using a consistent logic

Recoding is a key component of how audiences process information – it naturally occurs, as the brain re-organises information into fewer units to help overcome the cognitive limitations of the seven-item processing limit. This could be sorting information chronologically or grouping according to noticeable patterns. This natural distillation process is something you can duplicate when structuring your data story to reduce the cognitive effort required by the audience to process. This will also help ensure a consistent take out of the key messages, rather than individuals recoding in their own way and getting to different versions of the ‘truth’.

Questions you might ask yourself to help recode your key themes are:

• Which themes relate to the killer question we are looking to answer, and which are insightful but superfluous/ irrelevant/ repetitive for this story?

• Why do my priority themes matter more than other data observations I have discovered?

• Are the themes distinguishable from one another or can I combine?

• Is there enough evidence to support the theme?

3 – Translate your recoded themes into compelling points of view

Once you have distilled your analysis into 3-4 themes then you need to craft these into solid points of view. Spending time crafting your points of view is critical as these are the key messages you want the audience to retain and retell from your data story.

Below are some best practice principles to ensure you craft great points of view:

• Each point of view can be related to one another but should be a distinctive point in its own right – it can be ordered chronologically (first, second, third), comparatively (best or most important to worst or least important) or structurally (a,b,c)

• A strong point of view avoids truisms and generalisation to provide a unique perspective with specific references

• A point of view requires a short and succinct statement that incorporates verbs (doing words, such as optimise, inject, implement etc.) and conjunctions (therefore, as such etc.) to draw out the connections between the ‘what?’ and the ‘so what?’ and ‘now what?’.

Below are some examples of good points of view:

• Someone’s ability to quickly assess the quality of news underpins everything and therefore, if we want to reduce the risk of spreading fake news, we need to consistently urge platforms to promote and highlight verified and fact checked news content

• To drive subscriptions, we need to provide more space for our star content producers who rank highest on page views and reduce articles of less than 750 words that do little to drive engagement

• Considerable value is placed by customers in the support and stories shared by their peers, so Brand X can leverage this by either partnering with advocacy groups or signposting them to support resources that speak directly to the patient’s experience

4 – Ladder up to answer the killer question

You need to be able to ladder up from your 3-4 key points of view to provide an overall answer to your killer question. This is your story resolution and the main message that you want the audience to take from your overall data story. This answer will help you resolve some, or all, of the conflicts, challenges and tension in your story, and will help the business achieve their objectives and desired outcomes.

Questions you might ask yourselves when forming this judgment are:

• How does this answer impact our current plans? Does it mean we go, stop or change?

• What is the benefit of this answer, and which specific tension will it help us overcome?

• What are the Now, Next and Never actions?

A great answer requires two components:

• It actually gives an answer to the question – no sitting on the fence!

• It encapsulates all of the points of view developed from your analysis into one clear message

5 – Edit

Once you have identified the key messages it now time to edit to avoid overload. This is an essential step to maintain the quality of the narrative. Use your killer question as an anchor to make the decision about what stays in and what is left out of the data story. And when you think you are done, have one final check to make sure the data story is as fine-tuned as it possibly can be.

Editing takes practice – it takes us out of our comfort zone. But most of your audience will be used to consuming content that has been managed for the different levels of time and attention they have. And they will expect nothing else from your data story. So, seek inspiration from the news media, journalists, curators and content producers who excel at this skill. Try practicing editing your data story to create:

• a one-page summary for a five-minute read

• one or two paragraphs for a two-minute read

• a 30 second soundbite

• a 3 word slogan

Ready to transform your data storytelling?

If you are looking for more hints and tips on data storytelling, click here to get a 20% discount on ‘Data Storytelling in Marketing: How to tell persuasive stories through data’ Add DSM20 at checkout.