Visualization vs. Analytics Part II – Visualization Tools
The Workings of Visualization Tools
Visualization tools like Domo, Power BI, Qlik, and Tableau to name a very few, are extensions of Business Intelligence (BI) reporting. These tools use historical data to visualize trends in a multi-dimensional manner. Dashboards are of prime value to combine visual charts with tabular data of KPIs and key values for comparisons.
The image to the right is of Microsoft Power BI1, where data and images of trends can work together to offer a view to the past and present. Like a car’s dashboard, the numerical readings at the top tell key performance data needed to be known; e.g. if we’re running low on gas. A properly designed dashboard will have the charts coordinated to explain the top data so as to answer the What and Why of the key performance data.
But dashboards are not predictive, and views of past data can lead to false negatives or positives of the future. Look at the image on the left below from the visualization tool Datahero2. The historical trend is essentially up. So, what’s the next bar to follow? Up? Down? What decision would you make if you predicted up? What would happen if you guessed wrong? As seen on the chart to the right below, the next bar was substantially down.
Interestingly, as we shall learn in the next chapter, an analytics tool would have predicted this downturn through the Statistical Process Control Index formula.
Note, that dashboards can and often do contain forecasts, but these are most often data loaded from other sources and not analytically calculated; e.g. the sales forecast downloaded from the Salesforce.com system is the salesman’s biased view of what will close and when. Like the prediction of the next bar on the trend chart above, an unbiased statistical formula will often result in a more accurate forecast compared to a biased forecast. With no quota to meet or boss to please, mathematics calmly yields its quantitative analysis.
The good news about dashboards is the ability to bring numerous fields of data together across several dimensions to gain an understanding about an area of performance.
The image below is a Sales & Marketing dashboard from Grow3. It graphically depicts many fields of data in differentforms; e.g. the large bar graph in the upper left corner is Revenue by Product, and the geographic chart in the middleshows a “heat” map of US Site Visits. Gauges, charts, and tables disbursed through the screen add data points about sales and marketing activities to date.
A highlight from this dashboard is that while revenue has rapidly accelerated (Revenue by Product) the upper bar chartLead & Funnel Activity (with the purple bars) is rapidly declining. Therefore, revenue growth can’t be sustained if leads are falling.
This is the highest use of dashboards; i.e. to gain from history what cannot be seen with Excel and BI reports. But putting dashboards together takes technical and business knowledge.
Too often dashboards are the purview of IT or consultants to design and program, where their focus is on “pretty” ratherthan for “effective” decision making. The image below is a dashboard from Domo4 regarding Supply Chain. It has many metrics but . . . What decisions can be made? Is inventory optimized? Are the channels aligned? Are there any breakdowns in the supply chain? From this dashboard we simply cannot answer these questions. So, while the dashboard is pretty it is not materially effective to support performance optimization and growth.
Visualization vendors try to make their products more productive to use. However, a large usability gap exists for the business user. For example, The Finance Analytics Institute holds the Analytics Academy semi- annually, and over two days provides attendees the “How To” implement an analytics culture for data driven decisions. Two of the 19 sessions are dedicated to visualization and storytelling with visualization using Power BI. But, as attendees have learned, visualization even with Power BI that they thought would be intuitive, has a large learning curve.
As mentioned in the Overview, there are five questions to answer for data driven decisions of: What Happened, Where it Happened, Why it Happened, What Will Happen, and How To Make It Happen? Visualization tools can be used to answer the first two questions of What and Where, but only partially the third of Why.
Dashboards have good application in operational analysis (not analytics) and executive presentations. As seen in theSales & Marketing dashboard above, there is information that can be gleaned from the collection of charts, especially, when trends can be visualized. Dashboards also have appeal to executives who have limited time and attention. By crystallizing KPIs with a distillation of compelling charts, executives can quickly consume the pulse of the business.
Another high value use of dashboards is in storytelling, as a key component of a good and compelling story is visualization. The central tenants of a good story are clarity and brevity. The saying “a picture is worth a thousand words” is so true, and graphics can drive a compelling epiphany to decision makers and reduce a ream of reports to several figures that make the case.
This is often why visualization is used in conjunction with analytics, as the latter can be complex and hard to comprehend. The story that can be woven with graphics can be used to demystify the intensity of analytics that revealed the insights but could not be digested by executives.
Visualization is not analytics, so for the Why, What Will, and How To, dashboards can fall short or are just not capable. Note that visualization tools can answer the Why, so long as the Why doesn’t require analytics. As more companies rely on making decisions with analytics as seen on the figure below from D&B/Forbes5, the distinction between datavisualization and analytics has growing importance. Below are two examples of where the Why alludes visualization when analytics are needed:
- Marketing Intelligence: An auto manufacturer needed to learn which incentives were driving retail sales, as billions of dollars are spent on incentives; e.g. dealer cash incentive, customer cash, 0% interest rates, etc. This could not be answered by visualization tools as it requires multi-dimensional correlative statistics over a variable time horizon. Dashboards simply cannot answer Why some incentives work and others not. The solution was rendered by analytics that identified which incentives and where dimensionally and when in time these incentives are and are not effective. For example, Why, does dealer cash incentives have limited effectiveness. Analytics revealed that dealers were pocketing the dealer incentives instead of passing this along to customers in the first six months of the year. But, in the second half of the year, dealers use this incentive to incent customers to buy current year model inventory to make room for new year model inventory. These insights enabled the productive deployment of capital to generate retail sales.
- Sales Management: A Fortune 500 technology company needed to predict which individual sales prospects would close in a quarter. The current CRM system merely recorded the salesmen’s biased view as to what deals would book and when. Visualization (like the dashboard above) simply visualized trends. The answer was the application of predictive analytics with AI that could use a collection of data to find the patterns that were indicative of when a sale had a characteristic to close and when it did not. This enabled the better allocation of sales resources to optimize sales bookings.
Companies seeking to incorporate analytics should develop a matrix for defining their requirements for analytics, whichwill expose the more appropriate tools. For example, below is a matrix that PTC, Inc. (NASDAQ:PTC) a $1.1 billion global software company used to specify the analytics capabilities desired to advance the exploitation of their data. They already had visualization vendor Qlik in their Finance group and evaluated two other data visualization vendors. None of the vendors could deliver the end-user driven, dynamic, and predictive analytics capabilities they sought.
Even for Power BI, predictive analytics and AI was rated minimum to none, and the ability to build models and predictive models is significantly high (Note, Microsoft claims ML capabilities but refers to Azure Machine Learning Services6 that is in the Azure cloud platform). Power BI does have DAX (Data Analysis Expressions) that is useful for computations but it’s a programming language you’ll need to learn.
As such, data visualization has its value when it’s used like a dashboard in a car; i.e. to tell key operational metrics that need to be monitored. Value is also obtained from combining and visualizing trends and associated data to give light on the mechanics behind the key operational data. However, when analytics is needed then analytics tools should be applied, as use of data visualization for analytics would both fall short of the needs as well as can mislead decisions.
Read part I in this series for a refresher:
When you have finished part II find the final article in this series here:
The Workings of Visualization Tools
5. Dun & Bradstreet / Forbes Insights 2017 Enterprise Analytics Study
Not a member-scholar yet? Join our financial community here!
Identify your path to CFO success by taking our CFO Readiness Assessmentᵀᴹ.
For the most up to date and relevant accounting, finance, treasury and leadership headlines all in one place subscribe to The Balanced Digest.