Optimizing Lifetime Customer Value, or – Don’t Worry, Use Churn Models

2
438 views

Share on LinkedIn

Somehow, with all the prominent articles and conferences addressing customer experience, frequency marketing, customer clubs, cards, and other loyalty-related programs, identifying customer lifetime value hasn’t received the attention it deserves. After all, in any company’s allocation of scarce resources, shouldn’t a lot of planning and tools involve identifying and tracking customer spending, customer responsiveness (in terms of spending) to various programs and processes, plus issues involving customer attrition, customer loss and recovery.

There are industries in which estimating lifetime value has become extremely important in all customer programs, most notably financial services and telecom. However, research we conducted a few years back involving factors behind customer loss, and methods used to recover defected customers, showed that under half of the marketing managers calculate customers’ lifetime value. A bit more encouraging was the finding that two-thirds of the sales managers calculate lifetime value. Once they do establish lifetime value, most of the sales and marketing managers provide different levels of support and service, and differing communication and promotional programs, based on how customers are then segmented.

One technique for protecting and optimizing customer value which has become increasingly important is the churn model. Particularly in the telecom industry, where the rates of annual customer turnover in the United States and Europe can approach 30% and even 40% annually, these models have become increasingly useful and sophisticated in rating customers on defection risk. We’ve spoken with one consultancy based in France that has taken churn modeling to the next level: selecting the most appropriate communication program based on the customer’s value to the company, likelihood to defect, and probability of positive response to the program.



In the U.K. the Liverpool Victoria Friendly Society used a churn modeling approach to determine that 80 percent of those defecting came from half of the customer base. They further found that most of those dropping coverage came from a quarter of their customers, principally those who were younger, had never made a claim, or had relatively few Liverpool Victoria financial products. This very definitely has helped them target value enhancing programs.

Companies like Charles Schwab, Wells Fargo, Hilton and DirecTV in the U.S. use a specialized relationship management system, modeling software which enables them to study customer activity by segment, geographic location, revenue, demographic profile, and other variables. Using this software system, Schwab, for example, is able to access information in its massive customer account data warehouse. Users in their divisional marketing and corporate development groups can get information on spreadsheets, which enables them to make decisions regarding customers who might churn. Schwab can personalize approaches to customers at risk, enhancing the relationship and optimizing customer value.

American banks like PNC have surveyed their customers extensively to identify those at high risk of defection and also to determine where there are greater opportunities for lifetime value optimization. One bank has determined that, when there were such behaviors as a customer opening an account at another financial institution, this was correlated very highly with a deteriorating relationship. Another has found that, if there was a decline in the amount of checking account activity, if account balances declined or stopped, or if debit card usage declined, they could apply a model to establish the impact of each on possibly defection. The Royal Bank of Scotland has used similar research to uncover areas of softness in their relationships with customers, enabling them to target communication programs, both telephone and personal, with moderate and high volume customers.

This kind of research may be done as a stand-alone approach, or combined with the modeling techniques just discussed. As the marketing manager of a leading overnight package shipping company told me: “Why did they go away? You’ve got to do some critical analysis there. Is it cost? Is it competitiveness of the product? A superior product? A better supply? Better delivery terms? Better payment terms? You’ve got to find out why it is they were willing to change from you to some other supplier. They are making a change for some specific reason or reasons. You have to figure out what those are and then really do an internal assessment of whether those areas can be addressed with the customer.” Such is the value of good information and insight in optimizing customer value.

The challenges to identifying and segmenting customers based on real lifetime value is even greater when they are generated through the Internet. On-line marketing, advertising, communication, and customer loyalty programs are dramatically different than programs through any other medium. In, part, this is because the on-line prospect, user, and buyer has different needs and requirements than other customers.

Relationship time frames, methods of value and volume estimation, gathering customer information and use of databases, prospective cross-sale and referral projections, and even the way in which customers and segments are defined, have had to be rethought to accommodate the dynamics of on-line purchasing and web site usage. One of the key issues online marketers must deal with can best be defined as ‘velocity’, the rapidity with which online prospects can become customers, and current customers can become former customers. One disturbing statistic we’ve seen is that e-commerce web sites, especially new entrants, may turn over up to 60% of their customers within six weeks. This certainly makes getting to profitability, as well as generating customer stability and loyal behavior, a challenge.

Even though customer segmentation and value projection in traditional markets and marketing channels remains an open issue for many companies, the significant increase in importance of the Internet means that organizations need to be more action-oriented in learning about this. Companies are finding that, irrespective of marketing medium, an understanding of contribution over a customer’s life is essential to their success.



So, how do you identify customer lifetime value, or contribution? Determining both the lifetime value of each customer – their revenue minus their costs, and the ‘second lifetime’ value of a recovered customer, are perhaps the most important considerations in decisions regarding what resources to invest, and where.

Irrespective of industry, we believe that a company will first want to calculate:

– Cost to acquire a customer
– Average amount each customer spends per purchase
– Purchase frequency
– Average time with supplier
– Time remaining in the customer’s life cycle
– Costs to retain each customer, i.e. ongoing relationship expenses
– Profit from average customer (sales, minus fixed and variable costs)
– Referral and information value attributed to each customer

Now, armed with this information, the organization can identify value concentration, the overall portfolio value of their most contributory customer relationships. They will then have to determine which customers are solid, and which are at risk for defection, and meld that information with projected lifetime customer value. From that insight, the company can make decisions regarding program types and timing, levels of investment, and success criteria.

Sound complicated? It can be challenging to analyze profitability and value, but much less so if there is good information on each customer, acquisition and maintenance costs, and knowledge about projected life cycle (and not in the context of spending money on CRM, frequency purchase, and other loyalty programs undertaken with incomplete information). One of the complications created by such analysis is exactly what to do with customers found to be unprofitable.

Though sometimes controversial, firing unwanted customers is becoming almost as important as retaining or recovering desirable customers. If a group of customers just cannot be grown to a point of profitability, complain excessively, consume company resources at a higher rate than the value they contribute, or if they return products or demand credits beyond what is reasonable, the company should not hesitate to cut them loose, charge them at rates which will make them profitable, or step down the services they are provided.

Another factor increasingly contributing to customer termination is technological change. If a supplier, responding to the ordering needs of its best customers, elects to set up on-line Electronic Data Interchange, and some marginal customers still prefer manual ordering, they may have to be eliminated.



Companies can also take intermediate steps with otherwise marginally profitable customers. As noted with customers who might be forced to leave, these include reduced services, recommending competitors, selling customer bases and/or product lines, or outsourcing the servicing of these customers. In doing any of these, the company should consider how customers are likely to react. It’s not unknown, for instance, to see banished customers become very vocal – even complaining to other customers or the media – about how they have been treated.

Once a company has figured out customer lifetime value, and estimated net customer profit by segment – customers who are loyal and profitable, those who are borderline or unprofitable, and those desirable customers who are at some level of defection risk – the next step is to go further and estimate what has been termed ‘second lifetime value’, what the company can realize from a defected customer it has successfully recovered. That, however, is even less frequently done than determination of lifetime value of current customers.

2 COMMENTS

  1. So true – I am surprised how much the interest and use of good old Customer Value has declined within many companies in recent years. The justified attention to Customer Experience does not, in any way, eliminate the importance of understanding and managing Customer Value, and optimising resources by appropriate allocation.

    Similarly, Churn and its prediction are seen totally separately from Customer Lifetime Value and despite some advanced modelling, many fail to realise that churn propensity is a direct component of the lifetime value. Loyalty (or indirect proxies for it like recommendation scores) is also measured and managed separately, owned by different silos and disconnected from Customer Value (and often even from Churn management).

    Isn’t it blindingly clear that these concepts – and metrics – are inseparably connected and are best viewed in a single picture, ideally measured and managed together by closely coordinated (or even the same) teams?

    Ideally Churn analytics should be integrated with LTV (CLV) models, e.g. a Churn score can be a computed variable or vector as one of the inputs in the Value model. Similarly, any Recommendation score(s) (preferably based on observed behaviour rather than self-declared propensity) can also be input variables in the Value model. Lower churn risk increases projected lifetime, and recommendations (real not stated intent) add value. It looks so logical to me, yet I meet many accomplished business leaders who don’t seem to grasp such logic…

    Needless to say, any effort to measure or improve Customer Experience that does not account for the Value drivers buried within perceived experiences – is doomed to remain a nice icing on a missing cake. Lip service with no impact on Shareholder Value.

  2. Terrific observations. To understand total customer value across the overall base, progressive organizations will want to understand, and project, monetary worth at each customer life stage: suspect, prospect, new customer, retained/loyal customer, at-risk customer, lost customer, recovered/won-back customer. Few companies have the discipline or the systems in place to do that.

LEAVE A REPLY

Please enter your comment!
Please enter your name here