How we increased the efficiency of our client’s salesforce by identifying the high potential deals, enabling them to prioritize resources better.

Increased conversion rates of sales opportunities

Through machine-learning models

Hundreds of better utilised working hours of the sales force

Through machine-learning model creation and deployment

THE SITUATION

In their equipment business, our client opens thousands of sales opportunities per year. Sales teams needed help in identifying the right priorities and chasing the deals with the highest potential.

A new CRM system additionally increased the workload of the sales organization by raising the reporting requirements. Starting to implement data-driven solutions, leveraging the reported data, and demonstrating the value in capturing and analysing this information for the sales organization was a logical next step for the client.

Technical Solution: Machine Learning Models deployed on Azure Cloud Environment for maximum scalability in real-time

THE SOLUTION

Amplify created and deployed а predictive machine learning model for calculating opportunity scores, which indicates from 1- Very Low to 5-Very High the potential chance of winning an opportunity. The score is (re-)calculated at two moments of the lifecycle of an opportunity – at its creation in the CRM and when the first offer to the customer is submitted. The calculated scores are visible in the CRM and allow the Sales Reps to evaluate their winning potential and estimate the effort necessary to close the deal. 

The model was trained on CRM information on the opportunity and its customer such as the equipment type to be sold, the value of the deal, the past relationship with the client, etc. The quality of the data in our client’s sales organization has improved as a result of the implementation of the CRM but still needed a lot of focused effort in order to bring it to the required level for a smooth-running AI application. Therefore, Amplify put significant effort in reviewing all available data, selecting the relevant one, and identifying specific data improvement activities which the client initiated still during the project.

THE METHODS USED

Machine learning models were used to achieve the right opportunity scoring. 

During the “Modelling“ stage two gradient boosting algorithms were developed to predict the probability of an Opportunity being won in the future.

The first model evaluates the opportunity’s potential at the time when it was just identified and created in the CRM. At this moment there is still little definite information on the opportunity and the model uses some basic information about the account and its history with our client. 

The second model evaluates the chance to win the opportunity after the first offer has been submitted to the customer. At this moment a lot more details are available about the specific deal which are used as predictors for the model.

ML MODEL DEPLOYMENT

The model is created and deployed in the Azure cloud environment using the following infrastructure: the creation of a new Opportunity entry in Microsoft Dynamics (MSD)the API sends a message to the Azure Service Bus (ASB).

The message waits in the ASB queue to be read by a Virtual Machine (VM) set up at Amplify’s Azure Active Directory. When the message is received the VM calls the Dynamics API to get all relevant data for the newly created record. The data then is cleaned and transformed into a scoring-ready format. The VM calls upon the algorithm to score the ready data. After the data has been scored a call is made back to the MSD API to populate the “Opportunity Score” field for the given record with the newly calculated score value.

THE SUCCESS

This case study shows how Amplify Analytix supported the client in prioritising sales opportunities by using machine learning models to predict which opportunities would be won or lost. ​

The opportunity scoring model led to an increase in efficiency for the salesforce by identifying the high potential deals, enabling them to prioritise resources better.​

​By being scalabe, the solution allows the entire sales organization to make use of the model results in their own CRM system.