Problem

Sales of one of Oracle's database solutions in the Latin America region was significantly lower than projected targets for the quarter.
The team wanted to understand the reasons for the lack of sufficient sales and find ways to improve it.

Process

The first step was to gather the right data and prepare it for analysis. Oracle's sales and CRM data for the past 4 years were gathered through their various database systems to form an exhaustive sales and CRM table. Many duplicates and NULL values had to be cleaned or removed. Oracle's data tables had values of both their past successful and unsuccessful sales - which is the key ingredient in formulating an ML model that could accurately predict sales and demand.
While devising an ML solution to a business problem it is very important to have good knowledge of the business processes to remove variables that are irrelevant but show high predictability. These variables are hard to find and hence require detailed scrutiny of the business processes and the industry. After careful statistical analysis, 30 variables out of about a 100 were chosen for devising a prediction model.

Analysis showed that many of database products were sold only in combination with other database products. Godzin AI calculated the perfect product combinations that would more than triple sales win rates. Moreover, the model was able to identify customers who would be the best fit buyers of these product combos. Previously, the sales teams had a 1/10 client win rate and with our predictive technology they were able to increase that to 6/10. The AI engine identified cross-selling opportunities among customers that had never before been approached for this product.
Our AI engine recommended the right products and product combinations to sell, the right customers and leads to sell to and the strategies to use to win these customers and leads. The LATAM sales teams were able to increase their client win rates by 6x and reduce up to 90% of their time spent in sales research and analysis.