Digital Analytics Scholarship @ CXL Institute — 8th Week Review

This is the 8th week of my time in the program and this time, I’m in the class of Tim Wilson. This class was on Data Storytelling and Visualization and wasn’t available in the beginning. I think the course is growing with me and CXL is doing a great job in delivering the most updated course on the relevant topics.

Let’s find out what Tim Wilson has to offer.

9- Data presentation and visualization by Tim Wison

• I remember that I first got introduced to Tim Wilson by watching a recording of Super Week presented by Yehoshua Coren. Back then I didn’t know who I was getting the chance to know. But after a while, I figured out that this guy is a master that you can learn a lot from.
• Now I have the chance to learn from him using CXL Institute. How beautiful is this? I can’t believe that by using CXL, I have the chance to participate in classes taught by big names in the industry.
• Tim starts the course by introducing himself and the parts he will be teaching. It is apparent that he won’t be wasting our time by talking about a bunch of useless things. He says that you need to present data in the best way in order to complete the work you have started. I can’t agree more on this.
• How nice Tim puts it: data visualization is not about making your data prettier. It is about making it more understandable. Your data needs to be understood clearly and only then you can claim that you have done your job.
• For achieving this purpose, you have to understand how your audience understands your data. This is directly related to understanding how the human brain works. Tim has promised that he will teach us how to present the data to make it more understandable to flawed human brains.
• And he talked briefly about the curse of knowledge. That once you know about something, you simply think that everybody else must know about it, and because of this false belief, you shape all your thoughts around a set of wrong judgments.
• Vow! I am really enjoying this course only after watching two lessons from it. It is full of theoretical stuff that I need to know in order to be a better data visualizer.
• Tim Wilson says that we have three types of memory: Iconic, Short-term, and Long-term. Iconic one is that kind which our attention is drawn to instantaneously. And after that, short- and long-term memory comes in.
• Our job is to channel the attention of the user in order to make the data understandable and memorable. But attention is hard to earn. In order to earn it, we have to use Gestalt laws.
• We have to always pay attention to cognitive load. To not put a heavy load of data on our users and then expect them to understand them in an instant. We have to consider that our brains are limited in processing unfamiliar data.
• Our brains are very weak in processing areas and differentiating them is a hard job for a brain. But on the other side, comprehending length is a simple task, and the brain does it very well.
• Tim is talking about neuroscience in this course and emphasizing the fact that a marketer should know about this field as well. He is talking about the facts that I just learned in the Google Data Studio course a couple of weeks before. How odd is this?
• We have always heard about the ineffectiveness of pie charts. Some people claim that donut charts have come to compensate for what pie charts lack. But experts still believe that pie charts are truly nonsense. Why? Tim Wilson is trying to answer.
• He says that by using a pie chart, your eyes should constantly move on the legend and the chart itself. This imposes an extra cognitive load on the brain which is absolutely unnecessary. Even when you add data labels on the pie chart itself, the unnecessary eye movements are still in place. Why bother yourself and your audience and not use a line chart instead?
• There is something else that I’ve previously heard from Michelle Kiss as well and it is called data pixel ratio. It basically means that you have to try to use every pixel for presenting the real data. All of those decorations and fancy stuff are violations of the data pixel ratio rule and should be avoided.
• Tim said that this law was first introduced back in the time of printers. In those days, using as much low ink as possible was crucial and thus, the law was introduced as a data-ink ratio. But when the world was transformed by digital, when having modified the law and called it data pixel ratio.
• It is clear that we don’t need to save in the pixels. They are there to be used. But what about the focus you want to have on the viewer’s side? When you are adding extra, not important stuff to your visualization or slide, think if it distracts the viewers’ attention or not? If it is important or not? If it distracts attention and isn’t essential, you would better to ignore using it.
• The next lesson was about using color. We have a lot of colors at our disposal, but does that mean that we are allowed to use all of them at once? I would answer this question with a bold NO!
• Tim says that you only need to use color when you want to draw users’ attention. This color is called an accent color and is used alongside a neutral color. This way, you have paid attention to cognitive load rules as well.
• Each table, chart, and visualization, in general, should have a neutral color as its base color and then use the accent color whenever it is needed.
• It is better to not use colors in legends. If you have the option, remove the legend completely and instead, try to label the charts on themselves. This way if a chart is printed by a black and white printer, your charts are still readable and meaningful.
• It is also noteworthy that you need to abide by the color palette of your brand. By using this palette in your data viz, you are using the best practices that are currently in place. If your company doesn’t have one, try to use online tools and color theory to develop one for your visualization purposes.
• The next lesson was about axes. Tim himself predicted that this might seem odd in the first place, but reiterated that it is very common to misuse them.
• Firstly, he said that all the y axis should start with zero. He called them zero-based y axes. If the axis starts with a number that is not zero, it might confuse the audience and lead them to wrong perceptions.
• And after that, you should always avoid two y axes. I mean such charts that have one y axis on the left and one on the right side. These kinds of tables seem to convey a bigger concept or trend in a very tight space, but in reality, it will only lead to higher cognitive load.
• Tim says this again that we have to think about maximizing data pixel ratio. If a pixel has lit up, it better be a pixel containing meaningful data in itself.
• Now we have officially started to analyze different chart types. Horizontal bar charts are very easy to comprehend and our brains can understand them without much cognitive load. Tim says that in western countries, people have used to read things from left to right and from top to bottom. So you’d better stick with what people prefer.
• But try not to use stacked bar charts as much as you can, because they don’t have a common zero base and thus impose cognitive load on the user.
• Try not to merge two or more horizontal bar charts together. Although this thing is somewhere inevitable to use, it has the potential to put too much load on the brain. I think Tim is seriously underestimating our brains! They are there to process things, don’t they?
• Line charts are specifically designed to show the trend of a metric over time. It is not designed to compare categories and in such cases, we’d better use bar charts instead.
• Not every data point should be labeled, because it will clutter the chart with numbers that are not necessarily important. Try to only label the points selectively.
• Using multiple line charts in one chart should also be avoided. You can obviously do this thing for sure. But it not only doesn’t add anything meaningful to the chart, but also will make the chart unreadable as well.
• We have sparklines and something rather new named small multiples. Sparklines are line charts that don’t have any axis or legend. They are only designed to show data in a small size. It is best to use them beside scorecards (google data studio naming) to make them more meaningful.
• But small multiples are totally new and at the same, very familiar. Think about small multiples as a data that has been broken down by dimensions and each dimension is shown by a small sparkline. It is portrayed in table fashion and you can compare data the way it is shown. Although small multiples are a bit hard to understand, they are very good at visualization.
• Text itself is a kind of data visualization. We can use bolder and bigger text to show the emphasis point and draw the eyes at certain points. Using colorful text and color accents makes the visualization more professional and easy to digest.
• Tim says that we have to avoid using pure black, because it is very harsh! I don’t get it at all. It is actually his own preference and doesn’t necessarily apply to others.
• Heatmaps show a metric across two dimensions. If you know about click maps or scroll maps, you know what I mean. They can be tricky to use, because a heatmap with bad color choices can increase the cognitive load and can result in misunderstandings.
• Heatmaps can be independently presented or come in tables (that are called heatmap tables.)
• I have to admit that prior to this course, I didn’t know that Scatterplots are designed to show the relationship between two metrics. Using this chart type, we can compare two metrics with each other and decide what to do with them.
• In scatterplots, you can find outliers which are the performing way below or above others. You can differentiate them using colors or annotate them so that the viewer can simply understand the reason behind them.
• In the next lesson, Tim talked about some non-standard charts. He said that these charts are not usually used and when you use them, you often lead to high cognitive load. But sometimes they are needed to be used and you have to leverage their existence.
• In such cases, you have access to Venn diagrams, Waterfall charts, Candle charts and lots more. Not that they are bad, but your audience isn’t necessarily familiar with them and might need some time and effort to understand them.
• Consider this: never use a chart just for the sake of adding variety to your reports. Sometimes repeating a chart multiple times is better than adding a rare chart that adds unwanted cognitive load.
• When we are trying to visualize the data using charts, we have to pay attention to the way that our brains and minds work. We are accustomed to reading texts from left to right (western countries) and top to bottom. Anything that tries to change that law is doomed to failure.
• Tim again says that our brains are terrible at understanding areas. We need bar charts and some context to understand what is going on. Don’t try to reinvent the wheel in this matter. Just stick to the rules that are in place and try to convey as much meaning as you can.
• Tim Wilson did something that taught me something, or at least I think I’ve learned something new. He divided dashboards into two types: Analytical Interface dashboards and Performance Measurement dashboards.
• Analytical dashboards are often prebuilt and don’t necessarily have the best data pixel ratio. They come with a lot of cognitive load and you need context in order to understand them. Something like Google Analytics or Adobe Analytics interfaces.
• But Performance dashboards are concise and to the point. They have tried to give you the most information even at a glance. You look at them and find out what is exactly going on. This is what this type of dashboard promises and delivers on.
• This is the end to the first part of the course, which was data visualization. The second part is about data storytelling and Tim says it is about how to stitch all of the data and visualization together to form a believable story.
• In the beginning of the storytelling part, Tim says that only stories will stick in the audience’s mind. We love to hear stories and tend to memorize them unconsciously. This is our innate nature and we don’t control it.
• Every report needs some data at its core, some visualizations to make the report understandable and of course appealing. But no data and visualization is complete without a narrative. Narrative is the story that you as the presenter want to tell and this is exactly the audience might crave for!
• Knowing your audience is a must, because you ultimately want to earn their actions. You are trying to convince them to do what you want.
• The audience is supposed to take actions based on your reports. Tim says that don’t think you are reporting one the data just for the sake of reporting. You are not there to tell people about the current situation, but instead are trying to give them hints on what can be done. Be specific and tell them exactly what you think.
• When you are presenting data, try to act based on the behavior of your audience. Make the presentation more like a conversation rather than a monologue. You are in the business of getting people’s positive opinions.
• When you are delivering the presentation in person, you have access to lots of things that are unavailable when you are presenting via Skype or just sending the link of the presentation via email. Try to learn the rules of each and act based on them.
• When you are in charge of presenting data, try not to follow exactly what has been done before. Try to rationalize the data flow instead and make the most sense of it.
• Tim says it is best to start with the end and think about the end in the beginning. Think with yourself about this: what am I going to ask in the last slide or part of the presentation and then form the beginning and the middle part in that direction.
• And use storyboards to have a picture of the output before you even start to design the report or presentation. It is crucial to see it before it even exists. Try to draw whatever that is on your mind and check if it is what needs to be produced. It is a great way and cinematographers use it a lot, because it saves them a ton of money and time.
• I just learned a new concept and it is McKinsey titles. They are used in presentations and are sentences with clear takeaways. The user can understand what the slide is all about by reading these titles.
• After reading these titles, the reader should see a proof for the claim that has just been made. Try to use accent colors and clear charts that reinforce what has been said. This way, the user can truly understand what you want to tell them.
• And after that, we have a simple notion that every presenter must know. When you are creating slides, try to only introduce one thing per slide and declutter the whole presentation. Your audience should have enough time to digest information on each slide.
• And if you are using data that has been extracted from a specific source, you have to refer to it and specify the date range which the data has been extracted from. If any kind of extra information is needed, refer to them in the appendix.
• Tim apparently hates using bullets and he says he has proof for that hatred. He says that bullets are counterproductive and are against the idea of slides clarity. Each slide has to have only one point and bullet points contradict this law.
• And texts should be succinct and minimal, explaining the most with the least. Selective use of color will enrich your text and make it more understandable.
• In facing presentations, we have two approaches: one time you are there to present what you have prepared and another time, it is only being sent to others. These two are apparently different and need different kinds of presentations.
• If you are presenting the slides yourself, try to be succinct on the slides and add more explanation yourself. This way, your audience only needs to answer whatever you are saying.
• But if it is being sent out to others, extra things should be placed on it. This time, bullet points are allowed and items have to have explanations. You can use footnotes, appendix and other addable things.
• In an ideal world, you need to make two different versions (document and presentation) have you need. It is a time-consuming task for sure, but results in the best possible outcome.
• Images can reinforce the notion that you want to convey. You can use still images, diagrams and charts to make it stick to the viewers’ minds.
• Images can be found using stock images, royalty-free services or using google search. There are certain rules that all focus on not infringing the author’s rights.
• After you think you have done everything, try to read the presentation out loud or present it for yourself. This way you can spot the potential flaws in the narrative and find out how much it takes for you to present it.
• You are reading it for yourself and if anything seems not logical or irrelevant to your ears, you can simply cut or edit them.
• No matter how much you know about data visualization and storytelling, you always need to learn about it. It is an ongoing process and learning never ends. Even if you don’t find anything new to learn, try to focus on things that others have prepared and think about potential ways of making their work better. Maybe they have done a work that is fantastic and can teach you about new things.
• And this is it. The data visualization course has ended and the journey has just started. I learned a ton of new things and owe this to CXL Institute and Tim Wilson.