By: Eric Siegel, Founder, Predictive Analytics World

In anticipation of her upcoming conference presentation, Using Machine Learning to Move the Enterprise, at Predictive Analytics World for Business Las Vegas, June 3-7, 2018, we asked Theresa Kushner, SVP, Business Intelligence at Dell EMC, a few questions about her work in predictive analytics.

Q: In your work with predictive analytics, what behavior or outcome do your models predict?

A: Our models predict several things associated with sales.  We predict what the revenue for any given quarter will be. We predict which customers will be ready for the latest offering. We also predict which big orders will actually be signed in a given timeframe.

Q: How does predictive analytics deliver value at your organization – what is one specific way in which it actively drives decisions or operations?

A: First, every predictive model we create must be associated with a financial return to the business- either in sales generated or costs saved.  Second, we accumulate all the revenue/savings from our projects and set a goal for delivering that number for each fiscal year.  The groups we work for internally should agree that they received the benefit we product BEFORE we include it in our accumulated number. That keeps us honest.

Q: Can you describe a quantitative result, such as the predictive lift of your model or the ROI of an analytics initiative?

A: There is a lot of work that we have done that depicts qualitative results. However, I would like to mention one recent project. The goal of this project was to improve business predictability by forecasting revenue in presence of limited and volatile historical data. Historically, Dell has a good process to monitor pipeline sufficiency; however, there was scope for improvement. The Performance Analytics Group team at Dell EMC proposed a methodology that improved the accuracy of the revenue forecast. They used blended time series technique to ensure a consistent and an accurate prediction. This solution showed a huge improvement by reducing error in prediction. The quantitative results were as follows:

  • The Mean Absolute Prediction Error (MAPE) for Dell’s overall Global Commercial businesses was reduced by 6 times in the first 12 months (trial stage)
  • Similar improvement of reduction of error by 6 times was observed for the next consecutive year
  • Post getting scrutinized globally at various business levels, this methodology replaced the previously existing methodology within the company

Dell EMC has also received a patent for this technology

Q: What surprising discovery or insight have you unearthed in your data?

A: The two recent discoveries that come to my mind are:

  • The timing of registering a sales deal on our internal tool is important: We found out, that the conversion rate of a deal registered at the end of the quarter is higher than the deal registered at the beginning or middle of a quarter
  • Most searched word on the website: One would assume that the most searched word on the website is ‘laptops’ ‘servers’ or the name of one of our products. However, the most searched word (US region) on our website in the first half of last year was ‘google’. The ranking of this word eventually dropped down but it remains one of the top 10 searched words on our website today

Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.

A: The topic for my talk at PAW is: How are large enterprises implementing ‘machine learning’ to make their business predictable? It will cover tips for MNCs to apply predictive analytics in their business and how they can effectively partner with their IT organization to get the most out of machine learning.


Don’t miss Theresa’s conference presentation, Using Machine Learning to Move the Enterprise, on Tuesday, June 5, 2018 from 2:40 to 3:25 pm at Predictive Analytics World Las Vegas, 2018. Click here to register to attend.

By: Eric Siegel, Founder, Predictive Analytics World