Major Rankings Company

Scalable Data Cleaning and Visualization Process
performance_icon

Major Rankings Company | Case Study

Scalable Data Cleaning and Visualization Process

STARTING POSITION

The client fielded a 54 page survey to millions of customers over the course of six years yielding:

  • Hundreds of millions of unwieldy data records, slowing down reports. Workbooks built from the data could only contain a small portion of the data at the response level.
  • Inability to display important data points. The structure of the data didn’t allow the client to display open ended responses.
  • Displaying the data to customers was inefficient. The client would calculate results for each pitch deck and create a stylized report manually. This work was duplicative and time consuming.

PROCESS AND IMPACT

CDA’s work allowed the client to

  • Untangle unwieldy data. We used Tableau Prep to align naming conventions, join data keys, and transform the data for Tableau optimization. We applied level of detail calculations to aggregate the data to specific view levels. These changes turned unworkable outputs into quick performers. The client is now able to compare survey results across multiple years rather than one or two at a time. The client can also track initiative progress.
  • Scale to future surveys. We joined in data keys to standardize survey questions and responses. This keeps the client from recreating the wheel every time they field a survey. The data process allows the user to connect the new data, run the flow, and start analyzing in minutes rather than days.
  • Save $80,000-$180,000 per year automating visuals for an ROI of $15-30 per dollar of investment. The client saves between $80,000 to $180,000 per year from Tableau visualizations we built to replace manually built PowerPoint visuals. These visuals were point in time creations and require rebuilding for each customer. The client has over 200 customers and the visualizations would take approximately 2-5 hours to build.
  • Get more value. We used R to extract each word in open-ended responses. We transformed the data into a long format and cleaned the responses to remove valueless words (e.g. articles, pronouns, punctuation, etc.). We then aggregated the data into frequency tables and applied survey weights. The client is now able to get value from the sentiment of open-ended responses by viewing word clouds and frequency tables.

These guys can handle very tough/complex assignments and come up with innovative ideas and solutions. I was very pleased with the outcome of this engagement.

Eric Durdov, Managing Director of Utilities Intelligence