Storytelling with data: What a book!

Sam Campitiello
11 min readJul 16, 2022

--

About a month ago, I took a look at the book Storytelling with data by Cole Nussbaumer Knaflic, given that it is a must-read for people working in the data world, and in this article, I will talk about it by writing a sort of summary of the take-home message of this fantastic work. It will be a sort of concise guide to the main concepts widely explained and covered in the book and the idea is to give a taste of what the book is all about.

The book is meant to be a guide for people who are working (or are willing to work) in the world of data. More specifically, it is a guide on how to maximize the power of data visualization to tell stories. The author does not give only a few tips on how to present ideas, how to structure a story, and how to use specific “tricks” to enforce the main message, but also she presents very detailed instructions and useful tips to build the perfect data chart. After reading this book, everyone can have a general idea of how to present data through visualizations in the most efficient way and tell the story behind it. The main feature of this book is its simplicity: the author uses very simple language throughout the different chapters, making the book very easy to understand and follow. It is very challenging to find a section hard to understand because there is none! Let’s jump into some details.

  1. Understand the context & first steps

Before starting to create a report or a presentation, it is important to understand the who (i.e., to whom we want to talk) and the what (i.e., the things we want to communicate). Based on these two, the first structure of the report and, more importantly, the story to tell can be brought to life, having an idea of how to communicate the results, the tone and the language to use. In the same context, the what is crucial to understand how we want to communicate with data, despite the kind of analysis:

  • A live presentation: we want to communicate something using slides. In this case, the control over what we want to communicate is maximized because we can lead the audience where we want to, even without showing all the details (people tend to ask questions about what they see).
  • A written document (e.g., e-mail, report,…): we have less control over the audience, and in this case, giving details is crucial because the audience is free to read the document at their own pace, and they will take only the information they want.

Another important step before proceeding with the creation of a report is understanding the problem we want to communicate: we must have a clear view of the what and it is very useful to create a first story to summarize the important points. The author suggests two solutions:

  • Three-minute story: it is a simple and direct summary of all the work done to answer the important questions and to solve the main issues regarding a particular problem, and it is very useful to have a broad but precise idea of the what.
  • Big idea: it is one single sentence summarizing everything in the clearest and fastest way, concerning context, results, and final suggestions. This idea can be used at the beginning of a presentation (i.e., to jump directly to the main point of the work) or at the end of it.

Once we have a very clear idea about what we want to communicate, we can start creating the report. One suggestion is to do it on paper (e.g., a blank sheet, with post-its,…) because it is simpler to visualize the first structure and to modify it according to what we want to tell. Of course, we can also modify the report while we are doing it on our laptop but it is always better to have a clear idea before doing it to avoid continuous changes, steps back, and second thoughts.

2. Choose a visual & eliminate clutter

For a particular set of data, it is wise to prepare and try different kinds of charts to find the best one in terms of clarity and tidiness. The author shows that, at the end of the day, there is only a small number of the most effective charts that people can use for communicating with data.

  • Simple numbers: (i.e., quantities, percentages,…) to communicate one single and straight information.
  • Tables & Heatmaps: to display multiple information or results. Here, attention must be paid to avoid the audience to lose focus while reading a “heavy” table: for this reason, a suggestion would be highlighting only the cells worth noticing to capture people’s eye.
  • Scatterplots & Line plots: used for plotting one value against another to show a possible correlation or distribution.
  • Slope-graphs: used for showing changing of values in two distinct cases (e.g., on two different dates) easy to understand because it is fast to see whether a slope is decreasing or not.
  • Bar plots: used for categorical data or the combination of these latter with numerical data. In this case, always start with a common 0-baseline to avoid confusion of scales.

In all those charts (the authors also included the Waterfall and Square area charts), it is possible to use different colors, sizes, and thicknesses to highlight particular values (useful to communicate the main message of the report) or to “hide” those that are not important.

The author also suggests avoiding 3D charts, because they may give a false sense of importance due to the perspective; pie/donut charts, because they are not good if used alone given that it may be hard to compare slices and arches, especially when they are very similar; and secondary axes, given that they may be very hard to understand, and they may lead the audience to pay extra attention to interpret the chart. In all these cases, it may be wise to create a parallel chart, add labels or change the type of visual.

When creating a slide or a report, each chart used for telling an important message must be set with a tidy structure, easy to read, and without distracting and uninformative elements. This “clutter” can be avoided by following the so-called Gestalt principles of visual perception, i.e. some basic principles that lead our brain to identify patterns and relations between elements: two or more elements can be considered similar, when they are close together, have the same shape or color, when they are inside an enclosed region, or when they are simply connected by a line. Smart use of those principles can help clean charts from useless sections and avoid the addition of extra distracting elements.

3. Focus the attention where you want it

To maximize the understanding of a message or a result through charts, data and text must form a tidy structure: this means that each element of a chart must be harmoniously located between the others without creating messing regions or hard-to-understand visuals. Moreover, smart use of alignments, texts, and colors is crucial to connect similar information and give extra details and highlight particular data with respect to others.

For example: the main element at the base of our report can be highlighted with dark colors while the other elements can be pushed in the background by using light colors; an important result inside a table can be highlighted by a dark region and bright bold text to make the audience focus on that point straight away when looking at the table; a word written in bold or with a larger size with respect to the others can be used to drive the audience’s attention straight to the main point of a text.

4. Visual hierarchy

In general, it is important to create a “visual hierarchy”, i.e. a visible guideline, to help the audience: using the Gestalt principles, it is possible to set a sort of order for the audience to follow and it can be done for texts and charts.

For example: in a text or a list, big bold font can be used for the title (this will make people look at that first because it will be very evident), then a simple bold text can be used for highlighting small titles (e.g., in bullet points) or important words (in this way, people will know that those words are important), and finally the rest of the text can be written with a simple font (if people want, they can go into the details by reading it). In this example, the hierarchy is defined by the smart use of size and boldness of the text.

Another example of this hierarchy can be applied to charts: a big title can be put at the top of the chart to tell the audience the meaning of the representation (it may contain keywords to go straight to the point of the visual), particular results or data can be drawn with thick lines and/or dark colors to drive people’s attention directly to the main message contained in the chart, and finally, labels can be added close to the main elements to give extra information, maybe using the same color to connect them. In this example, the path that has been set goes from the title to the main elements of the chart, and in the end, to the explanatory labels, creating a sort of guide for the audience’s eye.

5. Tell a story

As seen before, the importance of the who, the what, and the how in the process of preparing a report is vital for creating a three-minute story and a big idea story. These two can help to have a general and broad idea about a problem and about the work done to solve it. When communicating an important result using data and charts, some very simple suggestions and guidelines can be followed to create a real story that can grasp the audience’s attention and make everything more interesting.

First of all, in general, very simple language, without the use of acronyms or very hard terminology (and this is valid also for text in charts and visuals) is preferable to establish an informal and relaxed connection with the audience. However, it is important to notice that this is context-dependent and who-dependent (is it an informal meeting or a communication for top-level people?). Therefore, the speech depends on the specific circumstance.

To grab the audience’s attention, some tricks can be used for achieving that:

  • The story must be something that sounds like yourself, prepared for the audience (and not yourself), and it must not sound like a robotic presentation.
  • The use of examples is a very good practice to show details and concepts that can be hard to understand, especially for the part of the audience that is not completely familiar with the topic of your work. In this context, the use of comparisons can be of great help (e.g., show and tell what will happen if no action is taken to solve a particular problem).
  • To make the most important concept clear and part of the take-away messages, the use of repetitions is a very clever way to assure that: the smart use of the same word or sentence or even symbols (also highlighted in different slides) tells the audience that it is something important and therefore, it must be remembered (this can happen also subconsciously).
  • Sometimes it is preferable to go straight to the main point of the presentation without adding uninformative details. However, it is always good to create the context (e.g., give information about the problem, the reason behind the analysis, etc.) before anything else, to avoid confusion.

6. Structure of a story

An interesting story is something that catches people’s attention straight away, so a good start is crucial. In the beginning, the main “characters” of the story and the context must be declared in the easiest and clearest way: for example, these characters can be the problem to be solved, the elements and people involved with the problem, and the damage the problem is causing.

In this part, a first “tension” is created by introducing the problem and its effects. The tension is something that can be used up to the first presentation of the possibile solution to solve the problem: for example, smart use of questions (e.g., “is the company going to fail?” “is this solution enough to solve the problem?”) may strengthen the initial tension and it may lead the audience to wonder about the possible outcomes. Finally, in the last part of the story, the resolution of the problem will release the tension grown in the previous steps: here, data and charts can be used for supporting the solution to the problem and the decisions that must be taken.

Despite the structure described above, a story can be set in two different ways, depending on how you want to start it.

  • The most common choice is to tell the audience the same chronological process you followed to achieve the final results. For example: start by creating the context related to the problem to be solved, move to the possible solutions that can be adopted, then to the data supporting some of those solutions, and finally to the summary and conclusions.
  • Another strategy is to start from the end: the first thing to tell the audience is the main results of your work or what is necessary to do to solve a problem. After that, it is possible to move on to the important details and processes supporting the story you just told. However, this strategy is recommended when the audience is known and there is already an established trust (so people want to know the “so what” directly).

Summary

The reasons I enjoyed reading Storytelling with data by Cole Nussbaumer Knaflic are two: its simplicity and clarity. It is not an exhausting book and the way each chapter is presented drove me to read it as if it were a real tale, made of a beginning (i.e., what is the most efficient way to communicate a story with data?), a middle process (i.e., find the best way to visualize data using charts and tables), and an ending (i.e., how to set up a story to communicate something clearly). Each chapter leads to the next in a very clear and organized way. No more, no less.

Want to learn how to present something with data efficiently?

This book is for you.

--

--

Sam Campitiello

I am a Data Analyst with a Ph.D. in Astrophysics who follows his passions, from science to sport, up to the Ancient Egyptian culture and the Data Science world!