Our client is a major global player in generics and biosimilars. The firm has a portfolio of about 1,000 molecules covering all major therapeutic areas and generated $10.1 billion in sales in 2015.

 

Their project: Predict the churn of top affiliate pharmacies, and the positive evolution of smaller ones.

 

Their Challenges:

The marketing team at our client’s was struggling to find leading indicators that could help prevent churn while also highlighti
ng future customer growth.

The company’s pharmacy customers are divided into two groups: those that earn at least 65% of their total revenue from the sale of the client’s products, and all the others. The former group is particularly important in terms of business strategy, as these customers represent both considerable growth opportunities and significant financial risks in the event of a decline in sales. Thus, our client needed a clearer picture of their top customers’ behavior, so that they could predict their sales and anticipate actions that would help these crucial commercial relationships continue to thrive.

At the same time, they also needed a way to anticipate which pharmacies in the latter group might experience rapid sales growth in the coming months—turning them into top-earning customers.

Manually calculating the classification of pharmacy customers and predicting their inherent churn risks was already a huge drain on the client team’s time and efficiency. Attempting to make their models actionable would only add an extra layer of difficulty.

Yet without a handle on these insights, they could miss out on important sales growth opportunities and the chance to maintain stronger client relationships.

 

 

Prevision.io’s response: 

To automate computational tasks and get access to crucial data much more efficiently, our client’s marketing teams began working with Prevision.io.

Using Prevision.io’s enterprise AI platform, the team developed a model that:

  • Effectively targets the most strategic customers—the top 10% of pharmacies for the client interms of generated sales
  • Identifies pharmacies that are likely to have asignificant decrease in their revenue generated with the client’s products in the next five months
  • Highlights the pharmacies with the highes tprobability of increasing their sales

 

 

Benefits to the client:

The churn objective is computed from a combination of probability to churn and value ( marketshare ) of the churner in order to tune the model for maximising expected gain on a 5 months basis.
The Acquisition targeting model is compute the same way : a model is optimized to target drugstore retail with the greatest expected value instead of only the “most probable” or “most easy to acquire”


The two models runs for 2 years and were validated on the top100 french drugstore. It was estimated by the user of the platform that 1,3M€ were saved on one semester

In addition to offering targeted, automated predictions, the Prevision.io platform gives the client’s team access to monthly reports detailing these predictions and describing the risk level associated with each pharmacy targeted by the model. The reports also include a ranking of the targeted pharmacies by decreasing churn risk score, and with a recommendation threshold.

For maximum efficiency, the Prevision.io team backtests regularly and checks the model’s result levels with the client’s team. If the model is showing signs of degradation, or if the client’s team wants to evaluate new features, our platform allows them to retain the model in just a few clicks.