This last week my work has been very focused on the importance of measuring customer value. I spent the early part of the week helping a European utility define the future direction of its CRM project, and customer value looms large in those discussions. Later in the week, the importance of customer value came up in discussions with a major CRM industry analyst when reviewing the process of developing customer strategies.
I haven't covered the topic of customer value in this series of editorials so far, so I'm going to give you the benefit of those thoughts today. Before I start you need to understand that I am very focused on service industries, where the profitability of a customer relationship is highly dependent on how the customer uses the services (e.g. retail financial services, utilities, airlines, hotels, etc) of the organisation. This article will not apply so readily to product companies such as car manufacturers or retailers where the profit is primarily delivered from the customers' purchases.
It is quite clear from working with many major organisations that very few of them have a clear view of the profitability of each customer. That has to raise the question as to whether it is important. The best example I know of how customer profitability can be critical to success is a case study of the US credit-card industry ("Towards the information bank", The Council for Financial Competition, Washington). This extensive report shows how new entrants in the US credit-card industry such as Capital One and Signet were able to rapidly gain market share at double industry-standard RoAs through the use of customer profitability data at four key points of the credit-card lifecycle: customer acquisition, card usage stimulation, customer retention, and collections triaging.
Of course, an example from the credit-card industry may not convince the board of your company of the relevance of customer profitability data to them. We have found the most useful tool for this is often to undertake a customer profitability analysis, using a very simple algorithm, and prepare a chart showing an analysis of customer profitability by decile. (You'll find an example of such a chart in the presentation Customer Retention and CRM Techniques.
When the board realises the consequences of the strong variation in customer value, and that you know which customers are in which value decile, it suddenly starts to get very interested indeed. The chart usually shows that Pareto's rule applies (and usually in spades). For example, the presentation shows that for that particular bank, 15% of the customers contributed more that 215% of the profits, with later deciles contributing significant losses. This sort of analysis usually leads to a variety of calls to action.
As well as understanding current profitability, there's an additional need to understand the future value of the customer. The CRM theoreticians will usually talk about the life-time value of the customers as the combination of the current and future profit streams, but I would argue strongly against combining these two values, for a number of reasons. Firstly, if we keep the two figures separate, we can use them, as I show below, to start on developing a customer strategy which can affect our communications with each customer. Secondly, the future life-time value of a customer is difficult to predict with any accuracy, particularly in the fast-changing environment of today. Perhaps we need something a little more tactical than an unpredictable future life-time value.
I hope we've demonstrated that it's a good idea to have an understanding of customer profitability and future value (or potential as I prefer to call it). There are, however, a number of issues about implementing any of these measures, which I would like to address. Some of the major issues are: the difficulty of coming up with an agreed algorithm for calculating customer profitability, similar difficulties with calculating future value, and thirdly, how to use that understanding of customer value to develop a customer strategy which can turn that knowledge into practical actions with each customer to improve their specific value. Let's try and outline ways forward in each of these areas.
Although most major companies don't have a good understanding of customer profitability, this is not usually because they don't realise the importance of customer profitability. It is the difficulties of implementing an agreed customer profitability algorithm. A couple of the issues which frequently come up are: 1) how should I spread fixed costs across the individual customers, and 2) I find it difficult to develop a consolidated view which incorporates all customer products or services into the algorithm. These issues make it difficult to come up with an accurate measure of customer profitability.
My answer to these and similar issues is quite simple. We have to realise that we don't actually need a very accurate measure of customer profitability. We are not approaching this from an accountant's perspective but from a businessman's perspective. If we can allocate customers on a five point scale from highly profitable to highly unprofitable, we will have something we can use to treat customers differently. Relatively crude and simplistic algorithms, which need not cover all products or services (only those which are the major profit generators for the business) can provide a first-cut definition of customer profitability. Of course that first-cut definition will not be accurate, and we may even treat some customers inappropriately, but as long as our algorithm is significantly better than random in its allocation of measures of profitability, it will provide us with the ability to leverage customer profitability. This crude, inaccurate approach may strike horror into the heart of many IT professionals, accountants and others, but there are two points I'd like to make. Firstly, "in the kingdom of the blind, the one-eyed man is king" - in other words, if no-one is using customer profitability information because they can't do it accurately, we gain advantage over our competitors if we implement even a crude measure. The second point is that the very process of implementing and using a crude measure will encourage the evolution of more sophisticated and accurate measures. The very process of using the algorithm will lead to demands for its enhancement.
If we can't measure current profitability, what chance do we have to measure future profitability? Well again, I believe that relatively crude measures can suffice, and the approach I'll propose is outlined in the presentation mentioned above: Customer Retention and CRM Techniques.
We have long advocated that companies should develop propensity models that measure the likelihood and value of a customer undertaking a specific action (take out a mortgage, renew a credit-card, payback a loan, etc). If these models include value measures as well likelihood (i.e. not just the likelihood of taking out a mortgage, but the likely size, and hence value of that mortgage), then the propensity score represents a measure of 'future value'. If we develop propensity models for the major profit opportunities we have with customers, then the sum of the scores (provided they're to a common base) will be a rough measure of future value in the short-term of each customer. Some may argue that this really is too short-term, and in such cases it may be necessary to complement this view of future value from a more generic algorithm. In many industries, the future value of a customer is likely to be proportional to the economic strength of the individual, and this can be estimated by understanding relatively few variables. It is quite depressing to realise that for most of us perhaps age, sex, occupation and geographic location define quite closely our future economic power, almost certainly well enough to implement the rough measure of future value we need initially.
I hope I have now convinced you that it shouldn't be impossible to develop crude measures of current and future customer values. They won't be 100% accurate, but the understanding they give should enable us to use it to start improving customer value. How? Well I refer you to a third time to the presentation focused on Customer Retention:
There you'll find a slide showing how having measures of Customer Potential (future value) and Customer Contribution (current value) will allow us to start developing strategies for dealing with customers in different sectors of the quadrant:
- Customers with negative current value and negative future value - we probably need to manage them "Up or Out". We could migrate them to cheaper products and / or channels, or charge them more for the services they already use, for example.
- What do we do with those customers who are currently unprofitable but have future value? Well it should be obvious that we need to cross-sell them those products and services that will turn them into profitable customers.
- Customers who are currently profitable, but have no future opportunities?? Well, we need to retain those customers, and minimise marketing and sales expense against them.
- And finally, those customers who are making us lots of money, and there's also lots of future value. Well, we better roll-out a 'favoured customer' programme which rewards them for their current business and the future opportunity.
There is at least one further variable which we should be incorporating into this approach towards a customer strategy, and that is the strength of the relationship that we have with that customer. However this editorial is focused on the use of customer value, so we'll return to that topic at some future date.