Step by step process to surface customer behavior predictors

One of the most important tasks in trying to model customer behavior is to precisely zero in on predictors which influence a behavioral outcome using advanced analytical models. For example, in the insurance industry, a company might want to identify the top five influence levers that determine whether or not a policyholder would cancel his policy. Some questions that arise are:

Have all the causal factors been figured into the analytical model for predicting a behavioral outcome (i.e., policy cancellations)?

How many predictors were obtained from field immersion vs. exit surveys vs. 90-day customer conversation mining?

What is the step-by-step methodology used to bubble up predictors?

What are the various best practices and techniques in place to harvest solid predictors that can then be statistically tested for significance?

This article outlines a step-by-step process of methodically harvesting the most influential set of predictors using six different methods.

http://www.b-eye-network.com/channels/5407/view/15002

Segmentation + Scoring Model + Text Mining + Social Network Analysis = Force Multiplier!

Article on 4 stages of maturity combining Segmentation + Scoring Model + Text Mining + Social Network Analysis at

http://www.b-eye-network.in/channels/5407/view/14279

3 Real world text mining applications

Have posted 3 real world applications of text mining at http://www.b-eye-network.in/channels/5407/view/12783/

1O Customer segmentation best practices

Have posted 10 best practices in customer behavior segmentation at http://www.b-eye-network.in/channels/5407/view/13090/

Modeling customer behavior for segmentation in banking

There are 5 interesting buckets to model a banking customer

1. Risk variables

2. 6 month behavior

3. Product holdings

4. Revenue generated from Interest vs Other service charges

5. Channel interaction behavior

An indicative attribute model from a business point of view is attached

Differentiated actions on customer behavior

One of the simplest steps most oth can do is to segment their customers behavior data ( purchase, payments, claims, complaints, clickstream behavior etc )to find out what natural groupings exist and how this can be leveraged to drive segment specific interventions.

In the context of customers behavior segmentation a lot has been discussed around

a) CUSTOMER DATA DIMENSION : Experiences in the quality of the input customer data
- missing zip codes
- wrong customer classifications
- match merge customer identifiers
- enrichening customer profiles with credit rating and demographic information from experian etc
- identify derived behavior metrics
- rfm etc

b) STATISTICAL PROCESS DIMENSION : If the data hygiene dimension is stabilized, then the focus shifts to the segmentation process -
- Do we use K means cluster vs SOM/Kohonen method vs Hierarchial clustering etc ?
- How many clusters are ideal ?
- How many variables should we use to segment ?
- Which variables should we use ?
- How do we characterise the clusters ?

Assuming we get the data dimension and the segmentation process right, there needs to be a lot more conversation around ACTIONS to undertaken once this customer behavior segments are created. There have been situations where the customer behavior data was segmented but on account of poor 'actionability' framework the whole exercise collapsed.

Increasingly we are seeing 'ACTION' post the segmentation process being the weak link in the whole process.

Here are some actions we have seen working from a customer segmentation intervention point of view ( It has a predominantly retail and travel flavor :-) ...
1) Bundling multiple products and offering a discount to selective segments exhibiting certain behavior
2) Building multiple cross sell models to increase wallet share from certain segments
3) Reallocating marketing $ to run different kinds of campaigns for each behavorial segments.
4) Decisions on what kind of promotional stimuli to use for each behavorial segment. Some may respond well to a temporary price reduction, others may respond well to a coupon, some segments may respond well to a gift off on a product while others maybe open to an exclusive 'in store' event.
5) Each of these behavorial segments could be treated by different call center agents . For example the most valuable customers can get routed to the call center agents who are ranked high on performance . The cross sell customers can get routed to the agents who have a good track record of cross sell conversions etc
6) Channel decisions: Low value customers can probably be moved to the internet channel and the high value customers can have a dedicated relationship manager to offer personalized experience
7) Personalized gift with name printed could be offered to selective segments.
8) Steeper discount to customers who have generated value in other lines of products but have not tried out a new line of product
9) Personalized online portal for customers exhibiting certain behavior with customized recommendations

Since ,Data + segmentation + Segment intervention = Successful segmentation exercise
Wanted to understand from others their experential inputs on what segmentation interventions have worked for them .
1) What differentiated actions on customer behavior segments have worked to impact business outcome? ( Retail or Finance or Travel or any other industry is fine )
2) Is their a generic "actionability framework" which can be created which gives a menu card of differentiated actions to undertake on various segments ?
3) Are some interventions for customer behavior segments more effective than others.Are their innovative or effective actions which have been undertaken on segments discovered by segmenting customer behavior whichresulted in lift in sales or some business outcome metric ?

Monetizing from customer analytics in Travel industry

Every time you go to a travel agent to book a ticket on a flight, there are 2 broad kinds of transactions which are generated.
- Search request and response transactions
- Booking transactions

While most travel organizations have mined their booking transactions data, not many insights have been juiced out of the search patterns for air booking transactions.
For example , If you are a price sensitive tourist looking for the cheapest tickets between Bangalore and Colombo in Nov on Economy class on a Friday evening. Or you could be a value conscious business traveler seeking Economy or Business class tickets at the last minute to ensure that you are on time for a crucial business meeting in New York.
All the search requests and responses are captured in search log files and flushed out at regular intervals. These search logs which were traditionally seen as occupying a lot of disk space is suddenly viewed as a gold mine of interesting information. For example some interesting
- Which are the heavily searched destinations from Bangalore on weekends / Holidays where an say Singapore airline has no service?
o An airline could use this information to expand its fleet of services to destinations which it currently does not serve and increase its share of market
Another scenario consists of segmenting agents based on price conscious search versus value conscious search behavior. Business users are typically convenience shoppers (correct timing and service excellence is important) whereas holiday shoppers typically are price conscious. (Getting the lowest price to Colombo is more important than catching the flight at a convenient time)