Value of Segmentation

Posted 03-Feb-2004 08:45 AM
Posted for Muriel Wahl [by Teri Robinson]

Hello,

The company I'm working for (industry: consumer business, BtoB business) is starting a CRM process. We are starting to segment our customer database.
Looking at the available information in our database, it seems to me that it will be impossible to know the cost of a customer (even in terms of sales costs); therefore making it impossible to distinguish profitable customers from others.
I may know quantities bought (with history data), turnover, number of orders...
QUESTION: do you think that not knowing how profitable a customer is will penalized the results for an accurate segmentation?
Thanks

Teri Robinson
Managing Editor

Reply With QuoteEdit or Delete MessageReport This Post
Vladimir Dimitroff
Member

Posted 05-Feb-2004 10:06 AM
When differentiating customers by valueyou certainly mean profitability. 'Gross' value (revenue) is an unreliable indicator of profitability, assuming uniform distribution of costs.

Many companies start with revenue-based value models, which is better than no segmentation at all—but the right thing is to seek to evolve to a customer profitability model (some call it contribution).

Individual customer costs are often a challenge to track and calculate and may lead to adopting ABC or similar approaches. Where this is too difficult some companies have adopted applying aggregated per-segment costs to all customers in a segment (but different from segment to segment—where original segments have been revenue-based).

Another aspect of value segmentation is the predictiveness of the model: the simplest segmentations are based on historic value, while more advanced ones seek non-linear predictiveness (and add to it the discounting component, i.e. net present value).

Whatever your current capabilities and resources, it makes sense to start simple and constantly try to improve value modelling to make it more meaningful and actionable.

Patrick Maxwell
Member

Posted 05-Feb-2004 10:18 AM
Teri,

Profitability of customer relationships is only one of many independent variables used in the successful implementation of customer segmentations. Up until the mid 1990's, the RFM (Recency, Frequency, Monetary) method was the most widely accepted and used form of segmentation among marketers. It was easy to calculate, easy to implement, and, for most industries, accurate in its measurement of customer behaviors. And, it did not rely at all on profitability within its calculation. The monetary nature of the segementation was based entirely on order revenue, regardless of the COGS related to those orders. Even now, I would estimate that more than half of the companies that segment their customers are using one variation or another of this principal method.

Don't get me wrong, having profitability as an available variable would be helpful in your desire to segment your customer-base as far as their relative value to the company, but it's certainly not the only variable to concentrate on, nor critical to your success. In addition, it is a variable that can be estimated with the help of other variables, and one that can be ultimately surrogated within your analysis through the incorporation of a few simple techniques:

1. Every order requires your company to provide a service, and there are distinct costs associated with those services. Therefore, the number of orders placed directly relates to the number of services rendered, and the associated costs of those services. Use the number of orders placed by a customer to estimate this cost. Keep in mind, you will want to limit the scope of these orders to a small enough timeframe to be relevent, e.g. 12 monts.
2. You already have the key driver of profitability, revenue. Use revenue as a variable within your segmentation process, again, limiting the scope of revenues to the aforementioned timeframe. 3. If your company has a standard operating margin, add in an operational cost component for those orders as well (revenue * (1—margin) ).

With any luck, you will find that revenue and the number of orders placed by a customer are plenty to understand their relative contribution to the company. Even in the best situation, the best you can hope for from a profitability measure is relative weighting. In today's world, with pricing discounts, shipping discounts, wage differences, net terms, etc., it's not reasonable to expect perfect profit calculations. So don't feel bad that your unsure of the validity of yours, you have a lot of company.

Sincerely,

Patrick Maxwell
Director, Consulting Services
The Wyse Group

Mukund CRM mukund_ks34@hotmail.com
Member

Posted 08-Feb-2004 02:06 AM
Hi

According to 3C Methodology Customer behaviour can be categorized into various sectors.

Revenue : Either through Monetary methodology of volume of products or services over a period of time.

Positive Customer : Customer buys from you in the first place.

Customer Share : If it is on a symbiosis process what else would you benefit out of the Customer.

Cusotmer Lifetime : Average life span of the customer providing business to your organization.

Please refer to 3C documentation. I am sure that might help you.

Regards
Mukund KS
CRM System Specialist

Graham Hill
Guru
Member

Posted 11-Feb-2004 12:19 AM
Muriel

Mukund CRM makes an important observation that builds upon the previous postings, namely, that segmentation isn't just about the economic value of the customer to your company.

The purpose of segmentation is to identify groups of customers that have similar characteristics, but whose characteristics are different to those of other segments. Although the value of the customer is one component, it only takes a few moments thought to recognise that value by itself is a pretty poor segmentation criterion, especially in a B2B business build upon two-way buyer-seller relationships.

I suggest you think first about what you will use the segmentation for. If it is just for targetting customers for outbound marketing communications, then (depending upon your business) an RFM model or a p(Alive) model may be adequate. But if it is for organising customers into groups for different levels of key account management, then value by itself is but one factor, along with the customer's industry, their buying behaviour, their underlying needs and their affinity to your company, etc.

You should definately consider spending time developing an understanding of the economic value of your customers. It will not only allow you to value your customers, but it will also give you a much better understanding of how your company creates and destroys value itself in the process.

Interestingly, a recent article in the Financial Times by the British economist John Kay (see http://www.johnkay.com/strategy/317) suggests that the companies who have created the most shareholder value over time are NOT those companies who have focussed single mindedly on shareholder value creation as their raison d'etre, but rather those who have focussed on a broader value creation picture.

The process of working out the value of your customers and of developing a more encompassing segmentation approach should give your company much more of this broader picture than just a value-based segmentation by itself.

Think about it, you know it makes sense.

Graham Hill

Independent CRM Consultant
CRMGuru Customer Value Management Guru

Behram Hansotia
Member

Posted 12-Feb-2004 01:55 PM
I would not recommend using profits or LTV directly in the segmentation, since the objective of any segmentation is to organize similar customers into groups based on their inherent characteristics. Profits are a function of different costs incurred because of your company's policies(marketing,pricing,purchasing,inventory,etc.) and past decisions. Though these obviously can influence revenues as well, I believe you will have a cleaner picture by using revenues (along with other variables) instead of profits.

The biggest challenge in any segmentation is thinking through how you plan to use the segmentation and then deciding which variables you should use to drive the segmentation (basis variables.) This is not a trivial excercise and considerable brainstorming and discussions should occur with key users before you settle on the variables.

Note, I did not mean to imply that customer profitability analysis should not be done. It is a very important excercise, particularly if you want to track how the quality of your customer base is changing over time (It should, if your CRM initiative is successful) I would also therefore recomend that this excercise be done after the segmentation is completed so you can estimate the value of similar customers, by segment.

Tracking customer transitions over time, using a Markov Chain approach, can then help you estimate the life time value of customers in each segment.

A segmentation has many applications and is essentially a key building block for a variety of analyses and marketing programs your company may want to implement.

Once you have decided on the basis variables, deriving the segments also requires considerable sophistication with statistical methods like cluster analysis and latent class modeling. This is not something a casual user should attempt on his own.

Good luck with your project.

Graham Hill
Guru
Member

Posted 16-Feb-2004 03:12 AM
As Behram suggests...

If you want an outline description of how Markov chains can be used in segmentation and how they can be applied (with other mathematical techniques) to maximise long-term profitability, take a look at the following paper from the Universiteit Gent in Belgium

http://www.eur.nl/WebDOC/doc/econometrie/feweco20020517161025.pdf

It is part of a series of papers about how mathematical segmentation models can be used to inform decision making in everyday marketing.

Graham Hill
Independent CRM Consultant
CRMGuru Customer Value Management Guru