Prescriptive Analytics for Sales: Combining Analytics with Business Understanding

BI and reports have been driving salesforce in an organization for some time now. But building reports manually on a monthly / weekly / regular basis is daunting, boring, and error prone. The good news is that, making reports for Sales target/ Quality/ retention of clients/number of clients based on predefined KPI’s for a long hierarchy in an organization can be automated through programming. It’s about to change everything about your regular report creation process.

Using Statistical analysis, Machine learning algorithm based on past data and business understanding one can can make rules for finding low performing areas and opportunities to improve. Statistical analysis includes percentile analysis, anomaly detection, relationship/correlation analysis. Also using Rule based engines and ML models on past data we can predict targets and opportunities for each level of sales hierarchy. Using some statistical transformation on predicted target we will can targets in such a way that target would be realistic.

After finding low/high performing levels and target for each level we can automate reports with appropriate insights, tables and graphs.

Prescriptive Analytics Example

For example, let us take a hierarchy with:

  • 10 Regions
  • 250 Branches
  • 50 sales manager
  • 700 Sales persons
  • 100 products to sale

Let us suppose we want to make a report for Sales manager XYZ works in region ABC & handles 2 salespeople. Suppose salesperson A has sold products of Rs. 10,000 and B has sold products of Rs. 50,000 with growth rate is 40% p.a. and 10% p.a. respectively. Let us say if ML model says that focusing on a product let’s say XXX can increase the growth rate of A to 70% and focusing on another product YYY leads to an increase in B’s growth rate to 40%. Here salesperson A can achieve higher growth rate than B, but absolute increase in sales is higher for B. Here is where business priority comes into play. Business rules must be applied on top of ML models to decide whether growth rate or low performing salespeople is a priority of overall sales numbers are a priority.

Once the business priorities and ML models are combined, the machine can generate actionable insights like (assuming overall sales is a priority overall growth of individual salespeople):

  • Sales Manager XYZ has to focus on product YYY for his region to improve overall sales
  • Focusing on YYY can lead to growth in B’s sales by 40% p.a.