3 Best Practices for Data Visualizations in Manufacturing

Visualizations aren’t just pie graphs and histograms, they build trust with your audience and share valuable insights. These insights will be used to make decisions.  Here are three best practices that will help your analytics resonate with your audience.

Lights, Camera, Data Visualization

In theatre, the audience only sees the final performance.  They don’t see the hours spent in auditions, scripting, cleaning up the delivery and costume design.  In analytics, the audience only sees the visualizations.  They don’t see the hours spent in obtaining data, cleaning, transforming and preparing the coordinated visualizations.  In both, there is a perfectly curated story to tell.  The audience may not remember the exact line or data point, but they will take away the storyline/plot?.

Visualizations aren’t just pie graphs and histograms, they build trust with your audience and share valuable insights. These insights will be used to make decisions.  Here are three best practices that will help your analytics resonate with your audience.

 

Collaboration

Unlike theatre, your audience may know more about the data than you do.  It’s like performing to a group that already read the book.  Their “tribal knowledge” or “experienced intuition” has to be pieced together to obtain the true picture of the data.  There is character development that is essential to cleaning and transforming the raw values.  I always meet with the stakeholders closest to the business process at least once a week.  In these meetings I don’t make any assumptions about the data, just show them what I’m seeing and ask for feedback.

For example: Early in a project I shared a map of customer density by state.  The business immediately pointed out customers had been self-reporting this information and entering the first location in the dropdown instead of their actual location.  This isn’t something I could have known from reviewing the data and we decided to join an additional data source to bring in this information.

 

Documentation

When you walk into the theater you’re usually handed a playbill so the audience knows who is performing.  The first question I’m usually asked when showing a visualization is “where is this data from?”  When preparing a dashboard for a meeting, I include a slide that lays out 

  1. Data Source(s)
  2. Filtering and Cleaning 
  3. New column definitions (especially those created with formulas)

Additionally, on each visualization I add labels, titles and a detailed legend.  There may even be a note of something I found peculiar or interesting to get feedback from the audience.  These cues help me remember everything to review in a way I won’t “forget my lines” during the performance.  When the audience goes to rave about the performance later, they can easily reference the details they have forgotten.

For example: The following visual doesn’t provide any measurement for total on the axis or in the title.  The date is ambiguous, is it run date? Order date? Ship date? Invoice date? Requested delivery date?

bar graph

The following visual identifies the most common packaging type, but not the definition of that type.  Additionally, the business may identify some of these packaging types need to be grouped.

pie graph

 

Tie to Business Value

Every scene in a performance should tie to the storyline/plot.  Every visual you share should tie to a business question or provide an actionable insight.  By knowing your audience, you can prepare visualizations that clearly answer their questions.  They may ask you for information on sales, but they actually want to know how to make more sales.  You may show them information on their current customers, but they want to understand how to retain them.

For example: The business wants to understand why a part is rejected from the manufacturing line. The following graphs are the same data, just shown differently.  

This first graph first answers “What is the count of reject reasons by part?”  Count is on the Y-axis and Part is on the X-axis.  The colors represent different reject reasons (we have removed the reasons for confidentiality). Because there are so few defects on parts B and C it is hard to see the reject reasons.  

bar graph

 

This second graph answers “What is the most common reject reason by part?” The Y-axis is percentage of rejects and Part is on the X-axis.  The colors represent different reject reasons (we have removed the reasons for confidentiality). The audience can better understand the difference in reject reasons by part.

 

bar graph

 

Denouement

The denouement is the final scene of the play, when everything comes together and the situation is resolved.  The audience walks away feeling satisfied with the outcomes.  In the same way, you can empower your audience with visualizations that lead to ongoing insights and business decisions.  What information will they need to continue monitoring?  How can this data be displayed so it is easy to understand?  How can you use visuals to provide early warnings or alerts?

For example: Providing a way to compare sales by locations and plants allows the business to identify those that are doing above average.  Learning from these teams will provide value to the rest of the organization.

sales and plant graphs

Conclusion

By following the best practices of using visualizations, your audience will be begging for an encore!

Have your own example of a data encore? I’d love to hear it. You can contact me here: vmaus@excelion.io