Marketing Segmentation

Market segmentation is the process of dividing up a market into more-or-less homogenous subsets for which it is possible to create different value propositions. At the end of the process the company can decide which segment(s) it wants to serve. If it chooses, each segment can be served with a different value proposition and managed in a different way. Market segmentation processes can be used during CPM for two main purposes. They can be used to segment potential markets to identify which customers to acquire, and to cluster current customers with a view to offering differentiated value propositions supported by different relationship management strategies. In this discussion we’ll focus on the application of market segmentation processes to identify which customers to acquire. What distinguishes market segmentation for this CRM purpose is its very clear focus on customer value. The outcome of the process should be the identification of the value potential of each identified segment. Companies will want to identify and target customers that can generate profit in the future: these will be those customers that the company and its network are better placed to serve and satisfy than their competitors. Market segmentation in many companies is highly intuitive. The marketing team will develop profiles of customer groups based upon their insight and experience. This is then used to guide the development of marketing strategies across the segments. In a CRM context, market segmentation is highly data dependent. The data might be generated internally or sourced externally. Internal data from marketing, sales and finance records are often enhanced with additional data from external sources such as marketing research companies, partner organizations in the company’s network and data specialists (see Figure 5.2 ).

Don’t waste time Get a verified expert to help you with Essay

The market segmentation process can be broken down into a number of steps:

1. identify the business you are in
2. identify relevant segmentation variables
3. analyse the market using these variables
4. assess the value of the market segments
5. select target market(s) to serve.

Sales forecasting: Slide #6 (p. 136-8)
The second discipline that can be used for CPM is sales forecasting. One major issue commonly facing companies that conduct CPM is that the data available for clustering customers takes a historical or, at best, present day view. The data identifies those customers who have been, or presently are, important for sales, profit or other strategic reasons. If management believes the future will be the same as the past, this presents no problem. However, if the business environment is changeable, this does present a problem. Because CPMs goal is to identify those customers that will be strategically important in the future, sales forecasting can be a useful discipline. Sales forecasting, some pessimists argue, is a waste of time, because the business environment is rapidly changing and unpredictable. Major world events such as terrorist attacks, war, drought and market-based changes, such as new products from competitors or high visibility promotional campaigns, can make any sales forecasts invalid. There are a number of sales forecasting techniques that can be applied, providing useful information for CPM. These techniques, which fall into three major groups, are appropriate for different circumstances.

● qualitative methods:

customer surveys
sales team estimates
● time-series methods:
moving average
exponential smoothing
time-series decomposition
● causal methods:
leading indicators
regression models.
Qualitative methods are probably the most widely used forecasting methods. Customer surveys ask consumers or purchasing officers to give an opinion on what they are likely to buy in the forecasting period. This makes sense when customers forward-plan their purchasing. Data can be obtained by inserting a question into a customer satisfaction survey. For example, ‘In the next six months are you likely to buy more, the same or less from us than in the current period? ’ And, ‘If more, or less, what volume do you expect to buy from us? ’ Sometimes, third party organizations such as industry associations or trans-industry groups such as the Chamber of Commerce or the Institute of Directors collect data that indicate future buying intentions or proxies for intention, such as business confidence. Sales team estimates can be useful when salespeople have built close relationships with their customers. A key account management team might be well placed to generate several individual forecasts from the team membership. These can be averaged or weighted in some way that reflects the estimator’s closeness to the customer.

Account managers for Dyno Nobel, a supplier of commercial explosives for the mining and quarrying industries, are so close to their customers that they are able to forecast sales two to three years ahead. Operational CRM systems support the qualitative sales forecasting methods, in particular sales team estimates. The CRM system takes into account the value of the sale, the probability of closing the sale and the anticipated period to closure. Many CRM systems also allow management to adjust the estimates of their sales team members, to allow for overly optimistic or pessimistic salespeople. Time-series approaches take historical data and extrapolate them forward in a linear or curvilinear trend. This approach makes sense when there are historical sales data, and the assumption can be safely made that the future will reflect the past. The moving average method is the simplest of these. This takes sales in a number of previous periods and averages them. The averaging process reduces or eliminates random variation. The moving average is computed on successive periods of data, moving on one period at a time, as in Figure 5.10 . Moving averages based on different periods can be calculated on historic data to generate an accurate method. A variation is to weight the more recent periods more heavily. The rationale is that more recent periods are better predictors. In producing an estimate for year 2009 in Figure 5.10 , one could weight the previous four years’ sales performance by 0.4, 0.3, 0.2, and 0.1, respectively, to reach an estimate. This would generate a forecast of 5461.

This approach is called exponential smoothing. The decomposition method is applied when there is evidence of cyclical or seasonal patterns in the historical data. The method attempts to separate out four components of the time series: trend factor, cyclical factor, seasonal factor and random factor. The trend factor is the longterm direction of the trend after the other three elements are removed. The cyclical factor represents regular long-term recurrent influences on sales; seasonal influences generally occur within annual cycles. It is sometimes possible to predict sales using leading indicators. A leading indicator is some contemporary activity or event that indicates that another activity or event will happen in the future. At a macro level, for example, housing starts are good predictors of future sales of kitchen furniture. At a micro level, when a credit card customer calls into a contact centre to ask about the current rate of interest, this is a strong indicator that the customer will switch to another supplier in the future. Regression models work by employing data on a number of predictor variables to estimate future demand. The variable being predicted is called the dependent variable; the variables being used as predictors are called independent variables. For example, if you wanted to predict demand for cars (the dependent variable) you might use data on population size, average disposable income, average car price for the category being predicted and average fuel price (the independent variables). The regression equation can be tested and validated on historical data before being adopted. New predictor variables can be substituted or added to see if they improve the accuracy of the forecast. This can be a useful approach for predicting demand from a segment. Activity-Based Costing: Slide #7 (p. 138-40)

Customer Acquisition costs
Terms of Trade
Customer service costs
Working capital costs
Activity-based costing
The third discipline that is useful for CPM is activity-based costing. Many companies, particularly those in a B2B context, can trace revenues to customers. In a B2C environment, it is usually only possible to trace revenues to identifiable customers if the company operates a billing system requiring customer details, or a membership scheme such as a customer club, store-card or a loyalty programme. In a B2B context, revenues can be tracked in the sales and accounts databases. Costs are an entirely different matter.

Because the goal of CPM is to cluster customers according to their strategic value, it is desirable to be able to identify which customers are, or will be, profitable. Clearly, if a company is to understand customer profitability, it has to be able to trace costs, as well as revenues, to customers. Costs do vary from customer to customer. Some customers are very costly to acquire and serve, others are not. There can be considerable variance across the customer base within several categories of cost: ● customer acquisition costs : some customers require considerable sales effort to move them from prospect to fi rst-time customer status: more sales calls, visits to reference customer sites, free samples, engineering advice, guarantees that switching costs will be met by the vendor ● terms of trade : price discounts, advertising and promotion support, slotting allowances (cash paid to retailers for shelf space), extended invoice due dates ● customer service costs : handling queries, claims and complaints, demands on salespeople and contact centre, small order sizes, high order frequency, just-in-time delivery, part load shipments, breaking bulk for delivery to multiple sites ● working capital costs : carrying inventory for the customer, cost of credit. Traditional product-based or general ledger costing systems do not provide this type of detail, and do not enable companies to estimate customer profitability. Product costing systems track material, labour and energy costs to products, often comparing actual to standard costs. They do not, however, cover the customer-facing activities of marketing, sales and service. General ledger costing systems do track costs across all parts of the business, but are normally too highly aggregated to establish which customers or segments are responsible for generating those costs. Activity-based costing (ABC) is an approach to costing that splits costs into two groups: volume-based costs and order-related costs. Volume based (product-related) costs are variable against the size of the order, but fixed per unit for any order and any customer. Material and direct labour costs are examples.

Order-related (customer-related) costs vary according to the product and process requirements of each particular customer. Imagine two retail customers, each purchasing the same volumes of product from a manufacturer. Customer 1 makes no product or process demands. The sales revenue is $5000; the gross margin for the vendor is $1000. Customer 2 is a different story: customized product, special overprinted outer packaging, just-in-time delivery to three sites, provision of point-of-sale material, sale or return conditions and discounted price. Not only that, but Customer 2 spends a lot of time agreeing these terms and conditions with a salesperson who has had to call three times before closing the sale. The sales revenue is $5000, but after accounting for product and process costs to meet the demands of this particular customer, the margin retained by the vendor is $250. Other things being equal, Customer 1 is four times as valuable as Customer 2. Whereas conventional cost accounting practices report what was spent, ABC reports what the money was spent doing. Whereas the conventional general ledger approach to costing identifies resource costs such as payroll, equipment and materials, the ABC approach shows what was being done when these costs were incurred. Figure 5.11 shows how an ABC view of costs in an insurance company’s claims processing department gives an entirely different picture to the traditional view.

ABC gives the manager of the claims-processing department a much clearer idea of which activities create cost. The next question from a CPM perspective is ‘ which customers create the activity? ’ Put another way, which customers are the cost drivers? If you were to examine the activity cost item ‘ Analyse claims: $121 000 ’ , and find that 80 per cent of the claims were made by drivers under the age of 20, you’d have a clear understanding of the customer group that was creating that activity cost for the business. CRM needs ABC because of its overriding goal of generating profitable relationships with customers. Unless there is a costing system in place to trace costs to customers, CRM will find it very difficult to deliver on a promise of improved customer profitability. Overall, ABC serves customer portfolio management in a number of ways: 1. when combined with revenue figures, it tells you the absolute and relative levels of profit generated by each customer, segment or cohort 2. it guides you towards actions that can be taken to return customers to profit 3. it helps prioritize and direct customer acquisition, retention and development strategies 4. it helps establish whether customization and other forms of value creation for customers pay off. ABC sometimes justifies management’s confidence in the Pareto principle, otherwise known as the 80:20 rule. This rule suggests that 80 per cent of profits come from 20 per cent of customers. ABC tells you which customers fall into the important 20 per cent. Research generally supports the 80: 0 rule. For example, one report from Coopers and Lybrand found that, in the retail industry, the top 4 per cent of customers account for 29 per cent of profits, the next 26 per cent of customers account for 55 per cent of profits and the remaining 70 per cent account for only 16 per cent of profits.

Lifetime Value Estimation: Slide# 8 (p. 141-2)
The fourth discipline that can be used for CPM is customer lifetime value (LTV) estimation, which was first introduced in Chapter 2. LTV is measured by computing the present day value of all net margins (gross margins less cost-to-serve) earned from a relationship with a customer, segment or cohort. LTV estimates provide important insights that guide companies in their customer management strategies. Clearly, companies want to protect and ring-fence their relationships with customers, segments or cohorts that will generate significant amounts of profit. Sunil Gupta and Donald Lehmann suggest that customer lifetime value can be computed as follows:

Application of this formula means that you do not have to estimate customer tenure. As customer retention rate rises there is an automatic lift in customer tenure, as shown in Table 2.2 in Chapter 2. This formula can be adjusted to consider change in both future margins and retention rates either up or down, as described in Gupta and Lehmann’s book Managing Customers as Investments. The table can be used to assess the impact of a number of customer management strategies: what would be the impact of reducing cost-toserve by shifting customers to low-cost self-serve channels? What would be the result of cross-selling higher margin products? What would be the outcome of a loyalty programme designed to increase retention rate from 80 to 82 per cent? An important additional benefit of this LTV calculation is that it enables you to estimate a company’s value. For example, it has been computed that the LTV of the average US-based American Airlines customer is $166.94. American Airlines has 43.7 million such customers, yielding an estimated company value of $7.3 billion. Roland Rust and his co-researchers noted that, given the absence of international passengers and freight considerations from this computation, it was remarkably close to the company’s market capitalization at the time their research was undertaken.

Clustering (144): slide #9
Clustering techniques are used to find naturally occurring groupings within a dataset. As applied to customer data, these techniques generally function as follows: 1. Each customer is allocated to just one group. The customer possesses attributes that are more closely associated with that group than any other group. 2. Each group is relatively homogenous.

3. The groups collectively are very different from each other. In other words, clustering techniques generally try to maximize both within-group homogeneity and between-group heterogeneity. There are a number of clustering techniques, including CART (classification and regression trees) and CHAID (chi-square automatic interaction detection).7 Once statistically homogenous clusters have been formed they need to be interpreted. CRM strategists are often interested in the future behaviours of a customer: segment, cohort or individual. Customers ’ potential value is determined by their propensity to buy products in the future. Data miners can build predictive models by examining patterns and relationships within historic data. Predictive models can be generated to identify: 1. Which customer, segment or cohort is most likely to buy a given product? 2. Which customers are likely to default on payment?

3. Which customers are most likely to defect (churn)?
Data analysts scour historic data looking for predictor and outcome variables. Then a model is built and validated on these historic data. When the model seems to work well on the historic data, it is run on contemporary data, where the predictor data are known but the outcome data are not. This is known as ‘ scoring ’ . Scores are answers to questions such as the propensity-to-buy, default and churn questions listed above. Predictive modelling is based on three assumptions, each of which may be true to a greater or lesser extent: 1. The past is a good predictor of the future … BUT this may not be true. Sales of many products are cyclical or seasonal.

Others have fashion or fad lifecycles. 2. The data are available … BUT this may not be true. Data used to train the model may no longer be collected. Data may be too costly to collect, or may be in the wrong format. 3. Customer-related databases contain what you want to predict … BUT this may not be true. The data may not be available. If you want to predict which customers are most likely to buy mortgage protection insurance, and you only have data on life policies, you will not be able to answer the question. Two tools that are used for predicting future behaviours are decision trees and neural networks. Decision trees (145): slide #9

Decision trees are so called because the graphical model output has the appearance of a branch structure. Decision trees work by analyzing a dataset to find the independent variable that, when used to split the population, results in nodes that are most different from each other with respect to the variable you are tying to predict. Figure 5.12 contains a set of data about five customers and their credit risk profile.

We want to use the data in four of the fi ve columns to predict the risk rating in the fifth column. A decision tree can be constructed for this purpose. In decision tree analysis, Risk is in the ‘ dependent ’ column. This is also known as the target variable. The other four columns are independent columns. It is unlikely that the customer’s name is a predictor of Risk, so we will use the three other pieces of data as independent variables: debt, income and marital status. In the example, each of these is a simple categorical item, each of which only has two possible values (high or low; yes or no). The data from Figure 5.12 are represented in a different form in Figure 5.13 , in a way which lets you see which independent variable is best at predicting risk. As you examine the data, you will see that the best split is income (four instances highlighted in bold on the diagonal: two high income/good risk plus two low income/poor risk). Debt and marital status each score three on their diagonals. Once a node is split, the same process is performed on each successive node, either until no further splits are possible or until you have reached a managerially useful model.

The graphical output of this decision tree analysis is shown in Figure 5.14 .
Each box is a node. Nodes are linked by branches. The top node is the root node. The data from the root node is split into two groups based on income. The right-hand, low income box, does not split any further because both low income customers are classified as poor credit risks. The left-hand, high-income box does split further, into married and not married customers. Neither of these split further because the one unmarried customer is a poor credit risk and the two remaining married customers are good credit risks.

As a result of this process the company knows that customers who have the lowest credit risk will be high income and married. They will also note that debt, one of the variables inserted into the training model, did not perform well. It is not a predictor of creditworthiness. Decision trees that work with categorical data such as these are known as classification trees. When decision trees are applied to continuous data they are known as regression trees.

Neural Networks (147): slide #9
Neural networks are another way of fitting a model to existing data for prediction purposes. The expression ‘ neural network ’ has its origins in the work of machine learning and artificial intelligence. Researchers in this field have tried to learn from the natural neural networks of living creatures. Neural networks can produce excellent predictions from large and complex datasets containing hundreds of interactive predictor variables, but the neural networks are neither easy to understand nor straightforward to use. Neural networks represent complex mathematical equations, with many summations, exponential functions and parameters. Like decision trees and clustering techniques, neural networks need to be trained to recognize patterns on sample datasets. Once trained, they can be used to predict customer behaviour from new data. They work well when there are many potential predictor variables, some of which are redundant.

Case 5.2 Customer portfolio management at Tesco
Tesco, the largest and most successful supermarket chain in the UK, has developed a CRM strategy that is the envy of many of its competitors. Principally a food retailer in a mature market that has grown little in the last 20 years, Tesco realized that the only route to growth was taking market share from competitors. Consequently, the development of a CRM strategy was seen as imperative.

In developing its CRM strategy, Tesco first analysed its customer base. It found that the top 100 customers were worth the same as the bottom 4000. It also found that the bottom 25 per cent of customers represented only 2 per cent of sales, and that the top 5 per cent of customers were responsible for 20 per cent of sales.

The results of this analysis were used to segment Tesco’s customers and to develop its successful loyalty programmes.

SWOT and PESTE (p. 154-5): slide# 10

SWOT is an acronym for strengths, weaknesses, opportunities and threats. SWOT analysis explores the internal environment (S and W) and the external environment (O and T) of a strategic business unit. The internal (SW) audit looks for strengths and weaknesses in the business functions of sales, marketing, manufacturing or operations, finance and people management. It then looks cross-functionally for strengths and weaknesses in, for example, cross-functional processes (such as new product development) and organizational culture. The external (OT) audit analyses the macro- and micro-environments in which the customer operates. The macro-environment includes a number of broad conditions that might impact on a company. These conditions are identified by a PESTE analysis. PESTE is an acronym for political, economic, social, technological and environmental conditions.

An analysis would try to pick out major conditions that impact on a business, as illustrated below: political environment : demand for international air travel contracted as worldwide political stability was reduced after September 11, 2001 economic environment : demand for mortgages falls when the economy enters recession. social environment : as a population ages, demand for healthcare and residential homes increase technological environment : as more households become owners of computers, demand for Internet banking increases environmental conditions : as customers become more concerned about environmental quality, demand for more energy efficient products increases. The micro environmental part of the external (OT) audit examines relationships between a company and its immediate external stakeholders: customers, suppliers, business partners and investors. A CRM-oriented SWOT analysis would be searching for customers or potential customers that emerge well from the analysis. These would be customers that: 1. possess relevant strengths to exploit the opportunities open to them 2. are overcoming weaknesses by partnering with other organizations to take advantage of opportunities 3. are investing in turning around the company to exploit the opportunities 4. are responding to external threats in their current markets by exploiting their strengths for diversification.

Five forces
The five-forces analysis was developed by Michael Porter. 17 He claimed that the profitability of an industry, as measured by its return on capital employed relative to its cost of capital, was determined by five sources of competitive pressure. These five sources include three horizontal and two vertical conditions.

The horizontal conditions are:

competition within the established businesses in the market
competition from potential new entrants
competition from potential substitutes.

The vertical conditions reflect supply and demand chain considerations: the bargaining power of buyers the bargaining power of suppliers. Porter’s basic premise is that competitors in an industry will be more profitable if these five conditions are benign. For example, if buyers are very powerful, they can demand high levels of service and low prices, thus negatively influencing the profitability of the supplier. However, if barriers to entry are high, say because of large capital requirements or dominance of the market by very powerful brands, then current players will be relatively immune from new entrants and enjoy the possibility of better profits. Why would a CRM-strategist be interested in a five-forces evaluation of customers? Fundamentally, a financially healthy customer offers better potential for a supplier than a customer in financial distress. The analysis points to different CRM solutions: 1. Customers in a profitable industry are more likely to be stable for the near-term, and are better placed to invest in opportunities for the future. They therefore have stronger value potential. These are customers with whom a supplier would want to build an exclusive and well-protected relationship. 2. Customers in a stressed industry might be looking for reduced cost inputs from its suppliers, or for other ways that they can add value to their offer to their own customers. A CRM-oriented supplier would be trying to find ways to serve this customer more effectively, perhaps by stripping out elements of the value proposition that are not critical, or by adding elements that enable the customer to compete more strongly.

Strategically Significant Customers (157) slide #11

The goal of this entire analytical process is to cluster customers into groups so that differentiated value propositions and relationship management strategies can be applied. One outcome will be the identification of customers that will be strategically significant for the company’s future. We call these strategically significant customers (SSCs). There are several classes of SSC, as follows: 1. High future lifetime value customers : these customers will contribute significantly to the company’s profitability in the future. 2. High volume customers : these customers might not generate much profit, but they are strategically significant because of their absorption of fixed costs, and the economies of scale they generate to keep unit costs low. 3. Benchmark customers : these are customers that other customers follow. For example, Nippon Conlux supplies the hardware and software for Coca Cola’s vending operation. While they might not make much margin from that relationship, it has allowed them to gain access to many other markets. ‘ If we are good enough for Coke, we are good enough for you ’ , is the implied promise. Some IT companies create ‘ reference sites ’ at some of their more demanding customers. 4. Inspirations : these are customers who bring about improvement in the supplier’s business. They may identify new applications for a product, product improvements, or opportunities for cost reductions. They may complain loudly and make unreasonable demands, but in doing so, force change for the better. 5. Door openers : these are customers that allow the supplier to gain access to a new market. This may be done for no initial profit, but with a view to proving credentials for further expansion. This may be particularly important if crossing cultural boundaries, say between west and east. One company, a Scandinavian processor of timber, has identified five major customer groups that are strategically significant, as in Figure 5.22 .

The Seven Core Customer Management Strategies (158-9) slide # 12

This sort of analysis pays off when it helps companies develop and implement differentiated CRM strategies for clusters of customers in the portfolio. There are several core customer management strategies: 1. Protect the relationship : this makes sense when the customer is strategically significant and attractive to competitors. We discuss the creation of exit barriers in our review of customer retention strategies in Chapter 9.

2. Re-engineer the relationship : in this case, the customer is currently unprofitable or less profitable than desired. However, the customer could be converted to profit if costs were trimmed from the relationship. This might mean reducing or automating service levels, or servicing customers through lower cost channels. In the banking industry, transaction processing costs, as a multiple of online processing costs are as follows. If Internet transaction processing has a unit cost of 1, an in-bank teller transaction costs 120 units, an ATM transaction costs 40, telephone costs 30 and PC banking costs 20. In other words, it is 120 times more expensive to conduct an in-bank transaction than the identical online transaction. Cost-reduction programmes have motivated banks to migrate their customers, or at least some segments of customers, to other lower cost channels. An Australian electricity company has found that its average annual margin per customer is $60. It costs $13 to serve a customer who pays by credit card, but only 64 cents to service a direct debit customer. Each customer moved to the lower cost channel therefore produces a transaction cost saving of more than $12, which increases the average customer value by 20 per cent.

Re-engineering a relationship requires a clear understanding of the activities that create costs in the relationship (see Case 5.3). 3. Enhance the relationship : like the strategy above, the goal is to migrate the customer up the value ladder. In this case it is done not by re-engineering the relationship, but by increasing your share of customer spend on the category, and by identifying up-selling and cross-selling opportunities. 4. Harvest the relationship : when your share of wallet is stable, and you do not want to invest more resources in customer development, you may feel that the customer has reached maximum value. Under these conditions you may wish to harvest, that is, optimize cash flow from the customer with a view to using the cash generated to develop other customers. This may be particularly appealing if the customer is in a declining market, has a high cost-to-serve or has a high propensity-to-switch to competitors. 5. End the relationship : sacking customers is generally anathema to sales and marketing people. However, when the customer shows no sign of making a significant contribution in the future it may be the best option.You can read about strategies for sacking customers in Chapter 9. 6. Win back the customer : sometimes customers take some or all of their business to other suppliers. If they are not strategically signifi cant, it may make sense to let them go. However, when the customer is important, you may need to develop and implement win back strategies. The starting point must be to understand why they took their business away. 7. Start a relationship : you’ve identified a prospect as having potential strategic significance for the future. You need to develop an acquisition plan to recruit the customer onto the value ladder. You can read about customer acquisition strategies in Chapter 8.

Written by Essay Examples

Dependence of man on the environment