I mentioned earlier in my series on risk and relationship based pricing that client valuation was an important component of the retail strategy and that I would post my ideas on the subject.
Before writing this, I discussed with a number of banks and checked on the internet the current state of customer valuation policies and models. This very brief analysis confirmed what I believed was the current state of the market, which I can summarise as follows:
1. Client valuation is not a systematic process. This management information is sometimes totally absent, often based on weak models and rarely integrated in a customer centric value based management policy.
2. Models are nearly always based on a current product profit and rarely on risk adjusted value metrics like Life Time Value, although there are industries that are more advanced than others in this approach. The banking industry is definitely lagging!
3. The LTV models used are more often derived for physical product sales and service providers such as telco’s. Consequently they are not adapted to financial products that have very specific characteristics (see Risk and Relationship Based Pricing Part 1).
4. The difficulty of developing valuation models is less in its formulation that in the access to detailed behavioural client data, although this data is developed in many analytical CRM programmes.
5. Customer valuation can only be achieved if Marketing, Finance and Risk management silos are bridged. This is still an exception rather than a standard operating model in big or small banks. The integration is easier in the smaller institutions but they often lack the data… and expertise!
If you don’t work for a financial institution you will find many sources of quality information on the web on this customer value, Customer Life Time Value, Customer Equity… Even Harvard Business School proposes a model with a spreadsheet down load (http://hbsp.harvard.edu/multimedia/flashtools/cltv/ ). I also recommend Strategic Planning: What’s the Lifetime Value of Your Customers? by Erica Olsen from Strategic Planning Kit For Dummies, 2nd Edition. Also check the following blog which has a great description and discussion on the subject (http://www.kaushik.net/avinash/analytics-tip-calculate-ltv-customer-lifetime-value/?replytocom=491800#respond).
BUT NOTE none of these sites are discussing LTV for banking products. The only one that does refer to banks is Don Peppers (http://www.peppersandrogersgroup.com/blog/). But I believe the approach developed there has major model weakness because Don Peppers and friends are more marketing oriented than finance focused. That being said read their stuff on managing trust in banking, it is good!
My last comment in this introduction: Bank client value metrics have been developed primarily for retail markets not for wholesale. It goes without saying that the concepts of client value and life time value are also applicable in theory for wholesale and specifically for wholesale banking. Although the principles are similar the usefulness of a precise calculation is different because of the nature of the corporate relationships. What retail banks must develop through an integrated data centric system, the corporate banker will develop on a 1:1 basis on the basis on more specific and complex information. He will be doing this when developing the “relationship plan” and integrate that information in the corporate business model (usually a corporate finance model). This makes it sufficiently different to treat the approach differently to retail banking.
Customer value in Retail banking
When I mention customer value I’m not referring to a vague idea of positive, high, low or negative value. I’m referring to a precise financial calculation of a financial value based on corporate finance theory. At the highest level customer value is the present value of the profit cash flows the bank will generate from the existing portfolio of products sold to a customer, those with a contractual maturity and which generate profits until that time, plus the profit flows of products without contractual maturities, but that will probably be maintained over a foreseeable future, plus future sales of maturity and non-maturity financial products and services.
This is of course different from a simple formula that multiplies an average product profitability per sales, by the projected number of sales per year, minus the costs of sales, the whole being projected over a 1 to say 3 or 5 year horizon with appropriate variables to derive a statistical probability of achieving an “expected” profit of x over that period (with or without present valuing the projected profits). This is great if you want to project the profitability of selling a pre-paid mobile phone card to a customer or a theatre ticket, or a car… But it is inconclusive if you want to measure the profit flow of a 25 year variable or fixed rate mortgage loan or a credit card…
If what I say is true, why are the banks not at the forefront of LTV, and would this be of use in their management challenges?
To illustrate the importance of profit and value metrics, let me run through a case study based on true data and management strategies of a very large North American retail bank (name withheld for obvious reasons, although the information used is publically available in its annual reports and presentations). I will also use the results of a LTV demo calculation based on bank data to illustrate the benefits of LTV.
The case study will complete this first chapter of the Client Value series. In later posts I will describe in more details the model and client centric strategies that it allows.
Customer value case study
The bank initially used a very simple client profitability analysis, which was based on relative product average profitability. The profit margin (in %) was used to multiply the average outstanding of the products (loans, deposits…) and the turnover or usage of financial services used (payment services…). Adding all these profit amounts allowed the bank to have a good idea of the profit contribution of this client for all his product holdings. This also allowed the calculation of the profit contribution by household.
Like many other companies/ industries the bank looked at the profit contribution of each client and segmented them in 10 profitability deciles. Oh my God, guess what, the well-known rule that 80% of the profits were generated by 20% of the clients! So obviously they devised strategies to concentrate marketing and relationship management expenses to the high profit contributors. You can’t argue against this.
They could of course also analyse the relationship between profitability and other client variables such as age. The graph below clearly indicate that older clients have a higher profitability (investment portfolios, high current account balances, etc…), while the 0 to 24 year old where loss leaders
All very interesting and useful to develop the marketing strategy (and these are two simple examples of the analysis possible.
But… a few years later the bank realised that the profitability data was not correct or at least was oversimplified. The profit was using average revenues and average cost allocations which do not reflect the true nature of both because they are not customer specific. To get close to the reality the bank had to individualise all revenues and costs at a contract granularity. For example, the use of banking services can have different profit contributions depending how and where they are generated. A simplistic example being the cost of a bank transfer generated by internet or through the branch. The same is true for all products; hence the bank must recognise this by going granular. From granular profitability at product level they can then analyse profitability on all dimension, by client, household, branch, region, product, risk rating…
When this bank calculated the new profitability deciles it appeared that up to 75% of the customers moved from their original decile to another one! The consequence is that the bank was concentrating marketing spend to possible the low profit contribution client and vice versa.
After a large investment in technology and systems (Teradata.com products) the bank set up the new system and developed its marketing strategy. This is what they presented and approved (the actual figures have of course been changed and some of the information hidden nevertheless the table below is good summary of the official strategy of that large retail bank.
Impressive! See how they allow a very small acquisition of low value customers (you can’t turn them down automatically but you can discourage them to bank with you through uncompetitive pricing, limited product offerings…. At the same time the bank plans to increase acquisition of profitability deciles 1 to 3 and increase the share of the wallet with those clients. Of course the bank will try to increase attritions for deciles 6 to 10.
If this actually happens expect big bonuses for the retail bank leadership.
But… as before are the profitability calculations correct? Is the model appropriate?
I had the opportunity to present to this bank a demo of an alternative customer valuation model. The proposed model was a Life Time Value model based on some real bank data (not the North American Bank).
The first approach uses a profitability model equivalent to the bank discussed. The client segmentation was based on socio-economic characteristics. The results by client segment are indicated in the table below.
Nothing specific to note other than there are major differences of profitability by segment, but that was already known.
Next we tried to estimate profitability for each customer segment through simple multipliers. This showed some major differences but was not acceptable because it was generic estimations based on expert knowledge. Finally we applied the LTV model by separating Current Value, Future Value and Life Time value. Remember the current value is the present value of all profit cash flows (adjusted for risk…) of the current product holding of the customer, while the future value is the present value of the profit cash flows of future sales to the customer. LTV is the sum of both of these. Of course both these projections are adapted to attrition risks, prepayment risks… as well as “sales” propensities and probabilities of roll=overs etc.
The results are I believe interesting. I have highlighted some segments to underline the changes in the customer valuation. Check segments 1, 7 and 9 for example.
Not only does this indicate that some of the very low current profitability clients end up as good or very good customers (they have a high future value) and some customers don’t increase value in the future, they have a low future value. It is of course obvious what these two segments could be: the young university graduate versus the older retired people. If you focus only on those you will of course have serious growth problems and will probably see your share of the market shrink over time. That will have very negative impacts on the bank’s market capitalisation.
When I showed this to the North American banks, they actually confined to me that they had come to the same conclusion and were actively working on an LTV model. The strategy they had proposed was cancelled because of the huge business risks it contained.
Without stating the obvious, think of the impact of a valuation model that can quantify the value of the current and projected product sales, using client behavioural characteristics. Not only will you focus on what is important, total value, but you will also measure/ quantify market behavioural variables to manage and control the efficiency of marketing:
• Focused on budgeted expected total value creation and the variance around that expectation and develop appropriate strategies 9campaigns, pricing…).
• Because the value model is based on quantified variable (financial, risk, client behaviour…) you can focus of the true value drivers and reduce business risks;
• Measure the value of marketing campaigns in a rational way i.e. measure the impact of up-front marketing expenses with the present value of future sales, retention strategies etc…
• Develop adapted pricing strategies to financial and business risks
• Drill down in all the variables that constitute client value and manage the clients on a one-to-one basis.
• Like with and in combination with other marketing models (such as Event Based Marketing) you can substantially increase the efficiency and effectiveness of marketing.
Again this post is too long… sorry!
If you have comments, ideas and if you disagree please don’t hesitate to comment on the blog or send me an email at firstname.lastname@example.org
The following chapters will describe in a little more detail LTV as I used it in the case study.