Building A Data Science Solution to Increase Conversions

Author: Joe Romani

At RevGen, we focus on building solutions that provide measurable value. Frequently, our clients are curious about areas like data science and machine learning, however they are skeptical about the return on investment and wary of hype. We understand this completely. Companies can’t simply hype their way into success. 

However, when our solution development expertise joins forces with our data science team, we can quickly provide meaningful results, as we recently did for a leading healthcare provider. 

Our client approached us with a challenge: could we find an innovative method to increase their closing rate and generate more revenue?  

This client typically converted about 25% of clients who booked an in-person consultation. Increasing that percentage even by a few points would result in significantly more revenue. The question was how?  

After looking at their data, RevGen proposed a data science solution to intelligently adjust pricing, increasing the closing rate while accounting for capacity constraints, business rules, and individual customers’ propensity to purchase. The solution needed to automatically retrain to adjust for data drift, always return a reasonable result for a discount, and integrate well with their Salesforce platform. It also needed to be available in real-time to provide quotes to prospective customers. 

Location-Specific Data Science Solution 

Because our client operates multiple offices, the solution had to provide discounts tailored to the conditions and capacity at each location. We wanted a solution that would adjust the discount for a location based on historical trends in location activity and currently booked appointments. During slower periods of time, when there were fewer appointments booked, the system would offer greater discounts to motivate customers. Customers who inquired during periods of higher demand might not receive a discount at all. 

To accomplish this, RevGen began by creating a forecast of business activity based on booked appointments for each location. Comparing that with each location’s capacity allowed us to create a recommended daily baseline discount. Since there was no deterministic way to calculate the discounts, RevGen used a method of artificial intelligence which optimized multiple parameters to generate the optimal discounts for each individual, given their credit and financial history.  

The daily baseline discount for each location was further fine-tuned and adjusted using location specific business rules. For example, some locations wanted to limit discounts. Others did not want to offer a discount at all if their booked capacity exceeded a certain percent.  These parameters were incorporated into the calculation of the baseline discount. 

Customer-Specific Data Science Solution 

Each customer is unique, and by quickly building an understanding of individual customers, we were able to further refine the recommendation. We built a model to predict the likelihood of purchasing based on gathered sales data, as well as the type of treatment, and their sensitivity to changes in pricing. These data points were fed into a classification model to infer likelihood to purchase as a function of price.  

Unfortunately, some crucial historical data was lacking that would have let us infer the likelihood of purchasing directly as a function of discount; discounts had only been recorded for sales and were not available for quotes, making this approach impossible. 

If a customer’s data suggested that they were insensitive to price, the daily baseline discount for that location would be reduced and the customer would see a price closer to the list price. On the other hand, if that customer’s data suggested a high degree of price sensitivity, the system would recommend a discount closer to the daily baseline discount.  

Customer Experience and User Interface

The pricing system was paired with Salesforce running on a tablet to allow each location’s sales team to quickly input information from consultations and receive immediate pricing recommendations. This required model results to be available near instantaneously, allowing the sales rep to immediately provide a quote to the customer, further increasing the likelihood of a sale. 

[Read More: How the AI Factory Cuts Through the Tech Hype]

Building a Reliable, Responsive, Accurate Solution

We knew that the solution would quickly become core to our client’s business; it had to be reliable, responsive, and accurate. In addition to validating the models, our team built and performed extensive testing to ensure reliability of the pricing system. We implemented a process of code reviews to examine changes as they made their way to production. The objective was to ensure that the software was never down or offline and never impeded a location’s ability to conduct their business. 

Summary

RevGen delivered a solution that provides location and customer specific discounts in real-time. It also captures pricing and discount data for quotes as well as sales, providing critical information for the business to continue to refine their pricing strategy.  

The client experienced significant success in the application of this solution: Close rates increased by over 10%, a 40% improvement, revenue grew by millions of dollars, and they achieved a positive return on investment within six months. 

This is just one example of RevGen’s work applying data science to pricing. We have done similar projects using data science to analyze wholesale price elasticity, enterprise level pricing strategy, AB testing of price changes, and others.  

To learn more about building a solution to optimize your pricing and grow revenue in your business, visit our Artificial Intelligence site or contact us to speak to an expert today. 

Joseph Romani is a Data Science Architect at RevGen, with over seven years of experience implementing data science solutions that provide tangible value for clients.