Industry

Insurance Companies

Pain

Targeting the most likely audience to sell from the correct channel

Solution

Learning from past data and predicting customers’ likelihood to buy, to trigger correct actions

Enhencer has one of the widest applications in the insurance industry. One of our insurance clients was having problems with customer profiles buying only one product, and not more.

In this occasion, Enhencer directly connected to the database which contains customer information (e.g. Income, Education, Complaints, Offer details) and from 149 attributes in customer data, 21 of them came forward as most relevant for buyer behavior.

They have a total of 1.1 million customers, and if they called each client without modeling, 114 thousands would have the tendency to buy the additional product. Instead of calling 1.1 million clients and selling to 114 thousands of them, Enhencer told them who to call, and they sold the product in 7 out of 10 calls (70% Hit Rate).

Lift
%
Accuracy
$K
Profit/Year

Industry

IT Sector

Pain

Unexpected server breakdowns

Solution

Predicting the breakdown before it happens

A subsidiary of a giant conglomerate, operating in IT/Networking field, have reached us to seek a solution for their server downtime prediction, using previously recorded log data.

The log data was not prepared, and our team also helped them with wrangling the data before we could import the features to Enhencer. In the process, we used certain features from the log data, which showed us patterns 1 minute before the server goes down.

After modeling this data with Enhencer's algorithms in a few minutes, we have exported the result and implemented it to the server for real-time predictive maintenance.

Why wait until your devices go down? Contact us for more on Enhencer.

Lift
%
Accuracy
$K
Profit/Year

Industry

FMCG / Retail

Pain

Why? Where? When? and Which products maximize my profit?

Solution

Stories with and often unexpected relationships between different business variables are where you can generate the most value.

One of the top ten beverage companies in Europe by sales volume used our Predictive Storyteller for analyzing their sales based on store, salesperson, region, profit, and market share.

Their data was quite complex with tons of variables possibly hiding many stories about their "hows" and "whys". We used store based variables such as total investment and sales amount, competitor and market sales values, and other contributing variables were salespeople and locations.

They were able to identify how and why they had bad performing resellers in good performing classes (hidden outliers) and pinpoint who were those outliers.

Also, salesperson performance was measured effectively with those parameters. Field Sales Managers were the ones who received the most useful insights as they did not know many of those stories about the resellers before.

Lift
%
Error
$K
Profit/Year

Industry

Insurance Companies

Pain

Losing existing customers

Solution

Taking the action for the customers who are the highest likely ones.

A worldwide insurance company had a case where 29% of the customers would cancel a current policy within three months. Enhencer connected to the database to import the historical data, and then performed the analysis in a few minutes thanks to Automated Machine Learning Algorithms.

As the churn trends differ depending on Life and Non-Life Product group, two different prediction algorithms were implemented. Results revealed that the most valuable information sources were; the age of a customer, the value of insured assets, purchasing channel, and the number of other insurances.

In the application phase, whenever a new sale is made, the churn probability of the customer is instantaneously projected to call center’s screen.

Enhencer identified every 8 churn case out of 10 correctly (%20 False Positive Rate).

Instead of calling each new policy owner, insurance company contacted the 20% with the highest churn risk and reached 90% of customers who planned to churn. With tailor-made offers to that high-risk group, they converted 28% of losses to profits.

Lift
%
Accuracy
$K
Profit/Year

Industry

Insurance Companies

Pain

Fraudulent work contracts

Solution

Performing checks to the highest risk group in the data.

Fraud is a pain in the majority of organizations, whether they are aware of it or not. Analysis and prediction of patterns and behaviors leading to fraud are best done with Machine Learning.

As the churn trends differ depending on Life and Non-Life Product group, two different prediction algorithms were implemented. Results revealed that the most valuable information sources were; the age of a customer, the value of insured assets, purchasing channel, and the number of other insurances.

This time we analyzed past company data containing Fraud and Not Fraud flags, and then scored the new data with Enhencer. The organization was able to spot 2 cases of fraud after checking 1.000 members.

After Enhencer's analysis, spotted fraud cases went up to 120 in 1.000 checks. Lift provided here is more than 60 times. This number can give you a feeling for the scale of the saved working-time and the prevented large financial loss with Enhencer.

Lift
%
Accuracy
$M
Profit/Year