Implementation Tactics For Segmentation

For many firms, market segmentation has failed, not so much because there was a failure to find appropriate segments, but more often because there was an inability to translate the segment definitions to customer and prospect databases. This paper describes the use of custom segmentation in comparison with a traditional approach.

For many firms a market segmentation strategy is essential for effective and efficient marketing management. Well-segmented markets allow the firm to craft specific product and value propositions to individual customers and prospects. However, for many firms, market segmentation has failed. Not so much because there was a failure to find appropriate segments, but more often because there was an inability to translate the segment definitions to customer and prospect databases, thus making it difficult to efficiently reach those segments and address their specific needs.

This white paper will demonstrate a method of leveraging widely used commercial segmentations, like PRIZM, Personicx and P$YCLE to develop customized segmentations. These customized segmentations make a direct linkage to customer and prospect databases extremely easy. In addition, having the custom segmentation based on widely used commercial segmentation allows companies to link diverse data sources to build very powerful marketing databases. This white paper will show how this technique was successfully employed in the context of a case history in the telecommunications industry. This approach works effectively for any B2C company using databases of customers and prospects to drive their marketing program. Other closely related industries that could easily apply these techniques include, telecommunication, cable, media, energy, real estate/housing, financial, hospitality/travel/leisure/recreational, computer, and healthcare insurance industries.

You just finished your segmentation and feel really good about the segment descriptions and color glossy pictures that go with them. Now comes the easy part, the implementation. To start, you first have to assign each customer and prospect to your newly formed segments. The information that you collected from the respondents told you so much about what they buy and why they buy, however, you do not have this same rich information on your customers and prospects. So now you need to build a model that takes the information you do have on your customers and prospects and score them into the new segment definitions. This isn’t looking so easy because the information that you do have on your customers and prospects is so limited and has so much missing data in it. Now you need to get the database people to clean up the databases to fill in missing data and to program in this new scoring model. Finally you need to have the sales group add any additional new fields to the new customer information form so all new customers will have the necessary data to score them. Clearly, effective implementation will not be easy. In addition, implementation might get held up due to a lot of database issues, and the segmentation might end up on your bookshelf and not the marketplace.

A large national telecommunications company fought this battle for many years and working with SDR Consulting finally came up with an approach that develops segments based on needs and behaviors and at the same time allowed them to accurately classify customers and prospects into segments by simply knowing their address. The key for this success was to develop segments by first aggregating respondents by PRIZM clusters and then segmenting on PRIZM clusters rather than segmenting directly on respondents. PRIZM is an industry standard commercial segmentation, developed by Claritas, that divides the US into 62 socio-economic clusters. Commercial segmentation like PRIZM allow a company to assign a customer or prospect to one of their pre-profiled clusters by simply knowing their address.

Traditional Segmentation Approach
“The information that you collected from the respondents told you so much about what they buy and why they buy, however, you do not have this same rich information on your customers and prospects.”

The essence of any segmentation is to allow a firm to craft specific product/service and value propositions to individual customers and prospects. There are numerous methodologies to develop a segmentation scheme. The most widely used approach is to execute a primary research study among a representative sample of customers and prospects. From this sample, needs, values and behaviors associated with the product/service category would be collected. The next step would be to group together respondents with similar motivations for purchase into segments. For example, segments might be developed around brand loyal, price sensitive or feature sensitive respondents.

After the segments have been developed they are described in detail by other information collected in the survey. These profiles help the marketing manager better understand what each segment values and the product/service benefits they are seeking, whether they exhibit different behavior in the category, and who they are demographically.

“So even though the segmentation is based on very specific needs and values, the assignment of customers and prospects to these newly formed segments is constrained by the accuracy and relevancy of information contained in company databases.”

In certain cases the segmentation ends at this point. The marketing manager knows the relative size and benefits sought by the different segments in the marketplace and can develop specific offerings, pricing strategies and communications messages. However, these very targeted strategies can only be communicated to the overall marketplace since the marketing manager does not know which customers and prospects belong to each segment. This increases marketing costs and limits the effectiveness of developing specific marketing strategies to target segments.

In order to target these offerings to the correct audiences, current and prospective customers need to be divided into these newly defined segments. Obviously, if the detailed survey information could be collected from all customers and prospects, then it would be simple to assign current and prospective customers to a segment. Clearly this would be cost prohibitive. Therefore, most companies try to build a model that forecasts which segment each current and prospective customer belongs to.

This is usually where a lot of segmentations fail, in that companies do not have enough predictive information about their customers and prospects in their existing databases to develop a meaningful predictor model. Even those companies who end up building models, often sacrifice a lot of accuracy in segment membership assignment since the information that they have about their customers and prospects is not as rich as the primary data collected in the segmentation survey. So even though the segmentation is based on very specific needs and values, the assignment of customers and prospects to these newly formed segments is constrained by the accuracy and relevancy of information contained in company databases.

In addition, most companies have very rich databases on a transactional basis. These databases contain critical measures on items like purchase volume, price elasticity, churn, channel preference, and adoption to new products. Linking this information with the needs, benefits and attitudes from the primary research study would be very insightful, not only for the development of the segmentation, but also on an on-going tracking basis. However, in most cases this information cannot be directly linked to the survey-based segmentation scheme. This is because most segmentation studies should not be just limited to a company’s customers, so these measures would only be available for some of the respondents. In addition, by the nature of the transactional data, a given customer will only have very limited information about certain items like price elasticity and churn. Utilizing these types of databases would be much more meaningful on an aggregate basis where measures can be computed across enough customers to make projections.

“The telecommunications company could not effectively respond to changes in the marketplace since they did not know what was driving their customers’ decisions.”

Due to the difficulties with implementing these value based segmentations and incorporating rich transactional data, many companies develop segmentations built around easily obtained customer and prospect information like usage and do not use these value-based segmentations. The problem with usage segments, is that more frequently than not, the benefits sought really do not vary by these usage segments. This significantly limits the understanding of target customers and prospects and the unique performance attributes they are willing to pay for.

This is an excerpt; read the full version in pdf form here.

This content was provided by SDR Consulting. Visit their website at www.sdr-consulting.com.