MODEL CITIZN

View Original

Visual Influence and Storytelling

Lance Rubin

Introduction to the co-authors

Aaron Parry is a professional analytics instructor, coach, and mentor for aspiring analysts and is the lead Power BI instructor and Head of Customer Success at Maven Analytics. Maven Analytics is a leader in analytics education, whose mission is to empower everyday people to change the world with data. 

Jeff Robson is the Principal Business Analyst and Founder of Access Analytic. Access Analytic is a Perth-based consulting and training company that specialises in financial modelling, Power BI reporting and data analytics. 

Overview & problem definition

I’m sure you’ve heard the phrase “beauty is in the eye of the beholder” but what exactly does that mean, how much of this is subjective, and how does it relate to data visualization? 

As you’re reading this article, think about examples of great dashboards, just okay dashboards, and really terrible dashboards. What’s the difference between these, what makes one great and another terrible? Additionally, how does that impact the story the dashboard is trying to tell?

Of all the topics in this series, we know that creating a visually appealing and compelling story can be challenging. Following a standard and or other technical concepts is often easier when building financial and data models but it can be far more subjective when it comes to visualization and storytelling that has an impact.

Information on this topic often conflicts and experts can sometimes disagree on what technical components a great data visualization contains. To help address this, we’ve made this a 3-way article to get different perspectives. We tried to summarize our thoughts into no more than 3 sentences but that was too challenging for us 😂 

Instead, we’ll define 3 main factors to keep top of mind when designing great visuals and dashboards: 

Design with a purpose (exploratory vs. explanatory)

  1. Less is often more (eliminate clutter)

  2. Tell a clear story (reading order, preattentive attributes, Gestalt principles)Ignore these three factors and you’ll easily overwhelm and/or lose your audience. The only outcome is a whole bunch of hard work with little impact. 

Ignore these three factors and you’ll easily overwhelm and/or lose your audience. The only outcome is a whole bunch of hard work with little impact. 

There is a lot more you can read about the different types of charts, visuals, and rules of data viz but keeping these 3 main factors in mind will maximize your data visualization’s impact. Up next, let's dig into why we’re passionate about data visualization and storytelling.

Why are we passionate about data visualization & storytelling?

Simply put, data visualization brings data to life (keep this phrase in mind for later). There’s a lot of truth in this statement because it’s based on facts – which is exactly what proper data viz is all about! It’s not just about making beautiful charts, cool interactive visuals, stunning infographics, or engaging dashboards, it’s also about communicating insights clearly to help make data-driven decisions.

One of the most interesting things about data visualization is that it sits squarely in the middle of the intersection of art and science. Don’t believe me? Well, let’s think about it for a minute. The human brain isn’t built to interpret raw data, at least not quickly and accurately, and even seemingly simple tasks may require a high amount of brain processing power. Let’s look at an example: 

Can you find the word “MAVEN” in the grid below?

We’re sure you were eventually able to find it, but how long did it take and how much concentration did it require? What's happening here is as you were trying to make sense of this non-visual information, your brain was relying on its prefrontal cortex - the slow and conscious part of your brain that’s responsible for cognitive functioning and problem-solving. This means that no matter how many combinations of this word search puzzle I show you, it’ll always take time and effort to solve. See…it’s just science! 

But what if we change things up a bit and add some color to the mix?

This was a brand-new word search but we’re willing to bet it didn’t take you more than a few seconds to find “MAVEN”. Chances are you actually saw “MAVEN” before you even thought about it! That’s because in this scenario your brain used its visual cortex. This is the part of your brain that’s responsible for visual perception and understanding, which means it helps us make sense of colors, patterns, shapes, and sizes. It’s also instantaneous and subconscious. 

Okay, back to the question at hand - why are we passionate about data visualization?

Well, it’s because there is a scientific reason (reads data-driven reason) why data visualization is such an important skill to learn as an analyst. It allows you to combine the power of cognition and perception to understand large sets of complex data both quickly AND clearly. In other words, data visualization lets you bring your data to life (see I told you to remember that phrase!).

As we’ve just demonstrated, data visualization is rooted in science and is a critical part of any effective dashboard, report, or infographic design but it requires a bit more understanding of what makes a good visualization. A proper visualization. A visualization that doesn’t immediately lose its audience.

If you were teaching this topic to a class of school kids what key tips would you share?

We love thinking about teaching anything this way. When you boil down a complex topic to its basic composition you should be able to explain the topic to an audience of any age. Think about trying to explain how gravity works to a 6-year-old child, you probably aren't going to lead with “gravity is a constant that states free-falling objects, on Earth, accelerate at 9.8 m/s/s”. You’ve lost your audience in the first 4 words! Instead, you’d probably use an example to illustrate the effect of gravity so the child can understand the concept based on something they’ve seen - like a ball falling from their hand to the ground.  

While this example is a bit dramatic, it helps highlight the point that there are some very tangible and basic concepts that anyone designing a dashboard should keep in mind and they should be easily explainable. Keep in mind that no tool is going to automatically provide all of the correct, or proper, data visualization answers for you. Just because you can do something with a data visualization doesn’t mean you should.

With that, here are our expert tips to get you started:

Aaron’s Top 10

1. Give your dashboard a purpose (exploratory vs. explanatory)

  • Exploratory dashboards are set up to facilitate data exploration and ask questions

  • Explanatory dashboards are set up to answer a specific question

2. Understand the type of data you’re visualizing (time series, categorical, geospatial, etc.)

  • This helps you determine and choose the proper chart type to display the data

3. Know what you want to communicate (composition, comparison, distribution, etc.)

  • With the proper chat type selected, make sure the data story is properly communicated

4. Design for the end-user (analyst, manager, executive, general public, etc.)

  • Always keep in mind the consumer of your report and deliver at that level. An executive probably isn’t interested in the tiny details of specific marketing campaign

5. Focus on metrics that matter (aligned with the dashboard purpose)

  • A dashboard should be built to serve a single purpose and isn’t effective when it tries to be a “one size fits all” approach. Think about the purpose of your dashboard and the outcomes you are trying to impact

6. Use effective visuals (“10-second rule”) 

  • Within 10 seconds, you should be able to understand what a visual is “saying”. Don’t make the end user work harder than they need to

7. Eliminate clutter (less is sometimes more)

  • Eliminate noise and clutter to facilitate understanding. Just because you have access to 100 dimensions in a table doesn’t mean you need to visualize them all

8. Leverage layouts (reading order, preattentive attributes, Gestalt principles)

  • Think about where the most important metrics should live, how a user will interact with and read a report, and design best practices

9. Tell a clear story (be thoughtful with your layout)

  • Follow a top-down approach, use descriptive titles and data labels, emphasize key points, use shapes & text boxes to customize the format or layout

10. Practice, practice, practice!

  • I can’t emphasize enough how important practice and feedback are to improving your data visualization skillset

Jeff’s Top 10

  1. Before attempting any dashboard, ensure the underlying data is reliable, accurate, timely, complete, and relevant.

  2. Understand the data and the purpose of the dashboard: is it intended to help users achieve a goal/KPI, or is it just an informational report? If you understand the key drivers in a business, you’ll be able to add a lot of value by assisting with forming meaningful KPIs.

  3. Everything in the dashboard should contribute to achieving the purpose of the dashboard otherwise why is it there?

  4. Don’t spray the user with multiple screens, that all say much the same thing, hoping that some will be useful. Consider how the viewer would know what is a good result vs a bad result. What should the viewer do about this?

  5. Use consistent visual language with colour, lines, background colours, headings and icons.

  6. Reduce clutter by removing unnecessary chart components, changing tables to charts, making detail available (e.g. with popups, filters, drill-throughs or drill-downs) but not visible.

  7. Use the right visual for the right type of data e.g. lines for time series, columns for series with short names, bars for those with longer names.

  8. Each screen in a dashboard should have a focus and a logical grouping of visuals.

  9. Solicit feedback from users about your dashboard before it’s released so you can make it as relevant and useful as possible.

  10. Monitor how the dashboard is used following its release. What parts are users finding most useful? What isn’t being used at all? Iterate and improve.

What practical steps can people take now to learn more?

Generally speaking, the best approach to learning more is practice and feedback. Practice can come in a variety of ways, like taking courses, reading books or articles focused on visual design, by creating your own reports and dashboards, participating in data visualization challenges, etc. Here’s a list of some of our favorite learning resources:

Learn

Courses (https://www.mavenanalytics.io/course/excel-dashboard-design)

Books

  • Storytelling with Data (Cole Nussbaumer Knaflic)

  • The Big Book of Dashboards (Steve Wexler, Jeffery Shaffer, et al.)

  • The Big Picture (Steve Wexler)

Practice

Fun Stuff

Where are good places (links) to find out more on the topic

Dashboard Design Mistakes

Power BI Resources

How important is this skill in the context of learning financial and data modelling?

If your financial or data model doesn’t have an easily understood story to influence its users, why bother? 

This skill in visual influence is probably one of the most undervalued and underrated skills whilst at the same time one of the most complex to develop given its part art and science. 

Those that master these skills can certainly create significant value for the users and generate incredible insights and influence amazing decisions. 

Next to building the model, visualising the story is not far behind (if not parity with) in terms of its importance. 

How does all this disruption, AI and automation talk impact this topic?

If there is one area that AI and automation will struggle to entirely disrupt humans its in the creation of visualizations that have influenced other humans who are driven by emotions and beauty (which remember is in the eyes of the beholder, not the machine).

That’s not to say there won’t be an augmentation of AI-created visuals that humans decide are useful and insightful like the decomposition tree or waterfall generate analysis to explain the variance (both of these are already embedded in Power BI).

So whilst AI may speed up the process of creating some visuals, it won’t replace the need for humans to understand the business context and purpose for which the model was created in the first place. 

We are safe for years to come.

If you want to find out more and follow the rest of the article series be sure to download the Financial Modelling App

If you want to find more information on financial modelling and content visit the Model Citizn website or Access Analytic website.