This preview shows page 1. Sign up to view the full content.
Unformatted text preview: r unprofitable customers. They
used this model not only for customer retention but to identify customers who were not yet
profitable but might become so in the future. 5 Know Service then built a model to predict who among their profitable customers would leave.
As in most data mining problems, determining what data to use and how to combine existing data
is where much of the challenge lies in model development. For example, they needed to look at
time-series data such as the monthly usage. Rather than use the raw time-series data, they
smoothed it by taking rolling three-month averages. They also calculated the change in the threemonth average and tried that as a predictor. Some of the factors that were good predictors, such as
declining usage, were symptoms rather than causes that could be directly addressed. Other
predictors, such as the average number of service calls and the change in the average number of
service calls, were indicative of customer satisfaction problems worth investigating.
Predicting who would churn, however, wasn’t enough. Based on the results of their modeling,
they identified some potential programs and offers that they believed would entice people to stay.
For example, some churners were exceeding even the largest amount of usage available for a
fixed fee and were paying substantial incremental usage fees. They tried offering these users a
higher fee service that included more bundled time. Some users were offered as more free disk
space for personal web pages. They then built models that would predict which would be the
effective offer for a particular user.
To summarize, the churn project made use of all three models. One model identified likely
churners, the next model picked out the profitable ones worth keeping, and the third model
matched the potential churners with the most appropriate offer. The net result was a reduction in
their churn rate from 8% to 7.5%, for a savings in customer acquisition costs of $1,000,000 per
Know Service found that their investment in data mining paid off by improving their customer
relationships and dramatically increasing their profitability.
Applying Data Mining to CRM
In order to build good models for your CRM system, there are a number of steps you must follow.
The Two Crows data mining process model described below is similar to other process models
such as the CRISP-DM model, differing mostly in the emphasis it places on the different steps.
Keep in mind that while the steps appear in a list, the data mining process is not linear — you will
inevitably need to loop back to previous steps. For example, what you learn in the “explore data”
step may require you to add new data to the data mining database. The initial models you build
may provide insights that lead you to create new variables.
The basic steps of data mining for effective CRM are:
1. Define business problem
2. Build marketing database
3. Explore data
4. Prepare data for modeling
5. Build model
6. Evaluate model
7. Deploy model and results
Let’s go through these steps to bette...
View Full Document
This note was uploaded on 11/25/2010 for the course CENG ceng taught by Professor Ceng during the Spring '10 term at Universidad Europea de Madrid.
- Spring '10