The Impact of Data Visualization on Business

Sometimes, at data science conferences, I feel like a member of an Alcoholics Anonymous support group. I have this strange desire to start off my presentations with, “Hi, my name is Mikhail. I am a recovering Microsoft Excel user. It has been 42 days since I last used Excel to do analytics.”

Everyone claps, congratulates me on my achievement, and offers their support.

I used to love Excel. That was my first programming language (VBA), my first business intelligence tool, my first venture into the world of data. However, I eventually got away from Excel as analytics started becoming more and more prevalent, new tools emerged, and new needs arose in this area called data science.

Lately, I spend my time as a data visualization expert creating dashboards that help organizations visualize the data they have to promote faster insights, better adoption of new technology, and more interactivity for users.

The demand for an executive dashboard has only increased more and more as we tackle new ways to capture new forms of data that have never been accessible to us before.

This data is often complicated and cannot simply be thrown into a pie chart for an accurate understanding of what they’re telling us. Also, as the sheer volume of data increases, the need to accurately summarize that data in a way that tells a distinct message has become vital.

Otherwise, we are left with data overload.

As one of my favorite quotes puts it, “Creating too many reports is the most time consuming way to measure nothing.” So, data visualization is not just about design, as some are led to believe, although, hopefully, it will include great design.

Data visualization is not just making Excel reports into pretty charts or, worse, recreating those same reports into a new data visualization tool because it is en vogue. It is about promoting faster insight: “MIT neuroscientists found that visuals are processed 20x faster in the brain than text.”

As an example of this, I want to turn to a famous example of the power of visualization. The example is called Anscombe’s Quartet and is a set of four datasets of x,y coordinates. Each set has 11 records.

While the actual x,y values differ between the sets, they have the same mean for x and y, the same standard deviation for x and y, the same correlation between x and y, and the same formula for linear regression which is a way to predict future values.

In short, statistics says these four datasets are incredibly similar, but the moment we visualize them… we see that this is not true at all. Each dataset tells a radically different story. This is immediately clear to the viewer.

This is the power that visualization is having for business. It is communicating messages faster, cleaner, and clearer. Also, a good visualization tool will hand the reins over to the end user of the data to be able to search through their own data visually in a way that makes sense to them, without code and without expert knowledge of a tool.

It is about dropping the dependency on the consultant, the programmer, and the business intelligence specialist to allow the user to answer their own questions, to dive deeper in their exploration, and cut the lag time caused by natural work processes.

Those people are still needed in their various roles to help our business grow and develop, but they are not required for me, the executive, to answer many of the simple questions I have about my business.

This should be as great a relief to the team as it is to the executive. They will now have time to help drive the business and uncover insights instead of spending all their time pulling together monthly reports and endless ad hoc requests.

Can our current situation really be that bad? Is this really that big of an improvement on the current situation?

Let me illustrate with something I lovingly refer to as the “Cycle of Death.” You see, it used to be that an executive would have an important question about his business based on some new information that arrived at his desk.

Let’s say he received word from a store manager that sales will be lower this month. The store manager has not seen any report. He just knows his store and feels that things have been a bit slower than usual. Now, our executive wants to know if the drop is reflected companywide or is just at this specific location.

Sorry, we have not reached the end of the reporting period, so data is not in yet.

So, he sets up a meeting with a business intelligence or IT manager at the company and communicates his question. The manager does not go back to his desk and start churning out an answer. Instead, he sets up a meeting with his team, hoping that all their schedules will align perfectly for 30 minutes, to talk about how they can get an answer, where to find the answer, and how to frame that answer.

The team then works to put together an ad hoc analysis using programming skills, knowledge of some BI tool, understanding of the data, maybe some design, and hopefully, finds an answer. The team sets up a meeting with the manager to showcase their work. The manager notices something is off, communicates that to the team, and asks them to fact check their work.

When I relay this in person, by now, the entire audience is either annoyed, chuckling, or becoming bored, but they are all reacting to what I am saying because they have experienced this process hundreds of times before. Back to the Death Cycle.

The team cleans up the report and sends it back to the manager. Then, the manager approves and sets up a meeting with the executive to communicate the results of his team’s work. The executive receives his report. His answer: sales have dropped 5% in May in 3 out of 20 stores. It is a week and a half later, but he has the answer to his question.

However, with this answer, comes more questions: which product categories, why these categories, is the fact that all of the stores are in the same region part of the issue, is the drop in online sales or in-store sales, did we have enough product, were there any external influences, how did this drop affect overall profit, how can we better plan for the future, why didn’t we know this was coming?

You know the process. The exec meets with the manager, the manager with his team… and each time the director receives an answer, it promotes more questions.

Data Visualization is about empowering the executive (as well as all relevant levels of the company) to open up her or his tool, quickly SEE where things are happening, and have the ability to easily interact with the data to find these answers NOW — not next week, not even tomorrow. Adjustments can be made and the business can avoid potential issues and/or be a step ahead of their competition.

Now, let’s talk about another aspect of the current business atmosphere. I recently worked with an organization that had 30 institutionalized reports that were completed on a monthly basis.

They were NEARLY ALL TEXT, endless sheets of endless rows of data in Excel. Each one of these had pieces to the puzzle, but just pieces. They would have to sort through email after email or file after file to put together an understanding of how any one of their clients were performing.

Sound familiar?

I worked with them to build a one-stop analytics tool so that they could go to one place and find a holistic picture of the health of their clients’ business. This completely changed the conversations they were able to have with their clients and the level of insight they had into how they were doing.

If this process is done poorly, it can be a disaster. There are countless examples of terribly built visualizations or business dashboards that are full of gauges, 20-slice pie charts, and other poor ways of communicating a story.

It is important to understand the profile of who will be using the dashboard, what questions they will ask of it, how they will interact with it, and the best way to structure everything so it flows naturally. Then, the design must be crisp and adhere to proper principles so the message is communicated faster, cleaner, and clearer.

When done properly, mid-level users will have more insight into their business, uncover trends they previously were unaware of, and free up their business intelligence or IT team for more meaningful work.

Executives will be able to find the answers they are seeking without evoking the “Cycle of Death,” digest this new information quicker, and make timely changes that will significantly improve their business.