Digital Analytics
Data presentation and visualization | CXL review
Deliver effective reports that lead to action
“Data presentation and visualization” course is a part of the CXL Digital Analytics Minidegree Program. This four hours long course is made of 27 lessons. If the number of lessons looks scary, you will be relieved to know those videos are short (up to 15 minutes each) and easy to follow.
The instructor is Tim Wilson who is, among many other things, well-known for the podcast Digital Analytics Power Hour.
The “Data presentation and visualization” was the most interesting course so far in my journey to Digital Analytics minidegree. Wilson split the course into three major topics: data visualization, data storytelling, and effective communication. If you want to effectively deliver analytics information, those are the crucial skills you should have.
If you would rather listen to (and watch) Tim Willson, I recommend this presentation on Superweek Analytics Summit 2020:
Table of content:
Part 1 — Communication Overview
- The brain science of communication
Part 2 — Data Visualization
- Maximizing the Data-Pixel Ratio
- Data Visualization: The Importance of the Axes
Part 3 — Data Storytelling
Closing
Part 1 — Communication Overview
The curse of knowledge is a cognitive bias that occurs when an individual, communicating with other individuals, unknowingly assumes that the others have the background to understand.
The curse of knowledge is one of the main reasons why most tutorials on the internet are useless. Because authors assume their audience knows more than it does.
Also, it is one of the main reasons why most reports are useless. Yes, you got that right: because the authors assume their audience knows more than it does.
Think about Google Analytics reports, for example. You are no doubt aware of what Users or Session or Bounce Rate stands for. If you present a GA report to another skilled GA user, they won’t have any difficulty understanding what’s that all about. But, the chances you are going to present a GA report to another advanced GA user are close to none.
The audience to whom we are presenting our reports generally falls into two categories:
- They may not be familiar with the data itself
- Even if they are familiar with the data and the metrics, they certainly aren’t familiar with the analysis you are going to present.
When you are presenting, you are delivering information to someone else. If you do not deliver the results effectively and undoubtedly, then your audience is going to struggle to take action.
The brain science of communication
To better understand what does it mean to deliver “effective data presentation”, some “brain science of communication” needs to be explained.
Types of memory
There are three types of memory:
- Iconic,
- Short-term, and
- Long-term.
Short-term memory is where conscious thought occurs. Iconic memory, also called the visual sensory register, is where pre-cognition happens. Long-term memory stores information that will be retrieved an hour from now, tomorrow, next week, next month, or a year from now.
Miller’s Law
While the exact number of pieces of information that can be held in short-term memory simultaneously varies, it is limited. Psychologist George Miller’s 1956 study put the number at 7 ± 2. Some subsequent studies have indicated the actual number may be even smaller.
Cognitive load
Cognitive load refers to how much “work” the brain has to do to process information. Reducing cognitive load is an idea that lightening that effort will enable the brain to process and retain information more effectively.
Effective data visualization requires some level of experimentation: access the initial visualization of data and be willing to try alternative approaches to see if a better option can be constructed.
The Cognitive Load of Pie Charts
Pie charts require the brain to do a lot of unnecessary work and because of that, they are considered a poor choice of visualization.
Humans are much better at comparing lengths (e.g., bar charts) than areas, which is one of the reasons that pie charts are generally less effective than other visualization options.
Doughnut charts are no better than pie charts. The audience is simply asked to compare the lengths or areas of arcs rather than the areas of pie wedges.
Part 2 — Data Visualization
Maximizing the Data-Pixel Ratio
Maximizing the data-pixel ratio is an easy way to reduce cognitive load by removing clutter and unnecessary information. The technique can be applied to charts, tables, and also slides.
Maximizing the data-pixel ratio is simply the approach of having as many non-background pixels as possible representing data and information, which means minimizing the non-background pixels that are simply decorative or providing structure/organization on the screen. “Decluttering” is another way to think about maximizing the data-pixel ratio.
Data Visualization: The Importance of the Axes
Non-zero-based y-axes can make differences between different data points and/or trends in the data appear to be much more drastic than, in reality, they are. Be aware that many analytics/visualization platforms will rescale the y-axis to a non-zero axis, which can result in a misleading visualization that may be misinterpreted.
Scaling of the secondary axis on a dual-axis chart often causes the appearance of relationships or “key” points in the data that are, in reality, arbitrary.
Dual axes should generally be avoided. They add cognitive load as the brain has to “split” the data into two groups and retain which series are associated with which axes.
Part 3 — Data Storytelling
(…) when it comes to long-term memory, there are all these things our brain does to try to use as little information as possible to retain information in long-term memory. And it’s kind of hard to retain specific numbers in long-term memory, but when we’re retaining a story, it kind of taps into a couple of different areas of the brain and piece that together, which enables the recall. So it does help. Telling a story is something that helps with retention.
The Power of Narrative
How do we present data in a way that it can be understood? We tell a story because stories are more memorable than statistics.
Data stories include data, visuals, and narrative, while data visualizations are just the visualization of data.
Know Your Audience, Know Your Goal
The best way to identify the core audience for a data story is to identify who is equipped to take action based on the information being provided.
While there may be many people in the audience (or distribution list) for a data story, typically, there is only a single person for a very small group of people who have the responsibility and authority to take action based on the information that is being presented. That person (or small group) is who the data story should primarily target.
For someone to act on information that is presented to them, they have to understand it, care about it, and remember the essential takeaways of what is presented.
McKinsey Titles and Reinforcing Content
McKinsey titles are slide titles that are written in sentence form as a declarative statement of the main takeaway for the slide. The body of the slide is then built to reinforce that takeaway.
“Horizontal logic” addresses the idea that reading all of the headings (and only the slide headings) are read in order, the process will yield a complete and compelling narrative.
“Vertical logic” is the practice of ensuring that the body content on any given slide directly reinforces the takeaway stated in the title of the slide. It’s “vertical” because it is content that the audience consumes by reading “down” from the title of the slide.
Kill the Bullets and Limit the Text
“The audience will either read your slides or listen to you. They will not do both.”
An effective data story goes from one key point to another in a linear flow. An effective way to help the audience follow that flow is to reveal only a single key point at a time. Do that even if a slide has multiple elements that reinforce that point. This will also help you to reduce the amount of text on each slide.
Presentations vs. Documents
Data stories can be delivered as:
- in-person presentations,
- remote presentations,
- hybrid in-person/remote presentations (some attendees in person, some participating remotely),
- in 1:1 interactions, and
- via email.
As a presenter myself, I’m well-aware of the famous questions “Will your slides be available after the presentations?”. I was always eager to make them publicly available, but now I know better. This is why Lea Pica avoids emailing her presentation slides:
A well-designed live presentation deck needs you to guide the audience through your narrative. But the better you’ve designed your live slides, the less sense they will make to an offline audience.
On the flip side, an overcrowded, text-heavy presentation handout will lose your audience’s attention and sacrifice their live experience.
Closing
Data visualization, data storytelling, and effective communication are crucial skills you should have if you want to effectively deliver analytics information. For someone to act on information that is presented to them, they have to understand it, care about it, and remember the essential takeaways of what is presented.
Take as much time as needed to make your presentation serve its point: to make your audience act on it.
