AI for Best Product Selection
Insurance
Industry
4
Insurance Products
90,000
Prospects per Month
An insurance company running direct-sales acquisition campaigns wanted to improve a conversion rate of just 0.25% by identifying which product each prospect was most likely to purchase.
Predictive Product Fit
Managing large acquisition campaigns requires more than distributing offers across broad prospect segments.
Teams need to understand individual purchase propensity, identify the most relevant product for each prospect and use those predictions to make acquisition activities more targeted.
Propensity
Estimate each prospect’s likelihood of purchasing a specific product;
Product Fit
Compare purchase probabilities across products to identify the strongest match;
Targeted Selection
Select the product with the highest predicted purchase probability for each prospect;
Sales Impact
Use predictive product selection to improve the likelihood of purchase across acquisition campaigns.
Predictive Modeling at Scale
Using Elai, the company trained four predictive models, one for each insurance product, using historical customer and sales data to identify the characteristics associated with previous purchases.
In just one month, the models were ready for validation and could identify the product most likely to be purchased by each prospect with over 94% accuracy.
From Prospect Data to Product Decision
01 – Analyze Historical Purchases
Historical acquisition data was used to understand which customer characteristics were associated with the purchase of each insurance product.
02 – Predict Purchase Propensity
Each predictive model estimated a prospect’s probability of purchasing its corresponding product.
03 – Select the Best Product
The scores generated by the four models were compared to identify the product with the highest predicted probability of purchase for each prospect.
04 – Automate Predictions
The models were applied to recurring monthly prospect lists, allowing product recommendations to be automatically updated as new prospects entered the acquisition process.
Best Product Strategy
To evaluate the impact of predictive product selection, the company tested the new approach against its existing acquisition strategy over two months.
One group of prospects received the product identified by Elai as the best match, while a control group continued to receive products through the company’s traditional random-assignment approach.
Before
Random Product Assignment
Prospects were divided into groups and assigned one of the available insurance products without considering their individual purchase propensity.
After
Best Product Selection
Predicted purchase probabilities were compared across products to select the highest-scoring option for each prospect.
Scaled to Production
Following the results of the Best Product strategy, the company expanded its adoption of Elai, reaching 8 predictive machine-learning models implemented in production.