Editor’s note: Michael S. Garver is a professor of marketing at Central Michigan University, Mount Pleasant, Mich.

The author was hired by a transportation company to assess how market research could better support its current and future strategy, given that the company had undertaken a relatively new strategic direction. While this is the story of one company, many of the themes discussed within are common to many companies in a range of industries.

The client is one of the top transportation companies in the industry. In the past, most transportation companies have obtained success through operational excellence. But in today’s business environment, a better understanding of the marketplace is also key to success.

For the client’s customers, the process of selecting a transportation vendor is complicated, with a large number of variables factoring into the choice. From experience and prior research, it is believed that customers do not evaluate all vendors. Rather, in this fast-paced environment, customers have their go-to vendors with which they have a relationship. In addition, customers usually have certain levels of performance that must be satisfied to win the order.

The company believes that a market segmentation strategy is the road to excellence. Its belief is that not all customers are created equal and that different types of customers have different needs and require different service levels. As a result, segments need to be served in different ways. Currently, the company is segmenting the market by revenue (i.e., bronze, silver and gold), so that it can focus on those customers with the most financial benefits. The company wants to build its organization around these market segments and have total company alignment, where all functional areas are focusing on the same market segments. Currently, this is not occurring.

Monitoring the marketplace

From a market research standpoint, the company is monitoring the marketplace through a number of different customer listening tools. For example, the company gathers quarterly customer satisfaction data, focusing on attribute satisfaction relative to the competition, the importance of different transportation attributes and which transportation attributes need to be prioritized for improvement.

Additionally, the company purchases an industry-wide annual competitive benchmark customer satisfaction study that surveys the customers of the industry’s top 12 companies. In addition to examining attribute importance, this study shows the satisfaction levels for each key competitor and highlights their strengths and weaknesses. While other ad hoc research studies are conducted, these are the major research studies conducted annually.

The questions facing the author were: Given the company’s strategic direction, how well does its current market research support and direct its strategic direction? And, what market research recommendations could be provided to better support and drive the client’s strategy?

Prior to attempting to answer these questions, key market research gaps needed to be addressed, including the lack of customer acquisition market research and issues with market segmentation.

Lack of customer acquisition market research

A customer retention strategy has become universally recognized as mission-critical over the last 15 years. The quality movement, along with the emergence of the balanced scorecard, has highlighted the importance of customer retention as a key driver of future growth and strong financial performance. This company has adopted this philosophy and thus places a premium on obtaining customer satisfaction data to support this strategy.

Market research needs to support both customer acquisition and retention strategies. The pressure of continual business growth also places a premium on effective acquisition strategies. Customer satisfaction research is an excellent source of information to support a retention strategy but is not well-suited to support a customer acquisition strategy. Almost all of this company’s market research dollars are spent on customer retention research. In contrast, customer acquisition research is non-existent.

Choice modeling (i.e., conjoint analysis, choice-based conjoint analysis, discrete choice) is an excellent research approach to support a customer acquisition strategy. While customer satisfaction data focus on attribute importance, choice models focus on attribute importance and various levels of performance for each attribute. For example, customers of this company consider transit time to be a very important attribute. Yet this is only part of the story. What level of transit time is needed and what level of performance are they willing to pay for? Would a three-day transit be acceptable or does the customer require two-day? If a two-day transit is needed, is the customer willing to pay for it? Or would this customer prefer a three-day transit for a lower cost of service? These subtle but key issues are ignored in traditional customer satisfaction research yet are incorporated into choice modeling.

Issues with market segmentation

This company makes a common market segmentation mistake as well: creating market segments with either demographic or financial variables and not creating market segments by customer needs. Best (2011) describes this mistake as the “demographic trap.” Best (2011) argues that many companies fall into the demographic trap, suggesting that they segment markets first and foremost using demographic or financial variables. Best (2011) offers that companies should first form need-based segments and only then should these segments be described with the use of demographic or financial variables.

This company segmented the market based on the customer’s revenue with it rather than on customer needs. In other words, based on the customer’s importance to the company and NOT on what was important to the customer. In fact, later research showed that the needs of the different revenue segments were virtually the same - defeating the purpose of segmenting the market in the first place.

A choice modeling approach

The recommendation was to use a choice modeling approach that addresses both issues previously discussed: the current overreliance on customer retention research as well as segmenting the market by revenue. Choice modeling will provide data for customer acquisition strategies as well as allowing researchers to first form need-based segments, which can then later be described with financial and other demographic variables.

While choice modeling is a sound approach, there are limitations associated with traditional choice modeling techniques (herein referred to as conjoint analysis) that are particularly troublesome to this company’s situation. Specifically: assumption of compensatory decision-making; irrelevant choice tasks; and large sample-size requirements.

Compensatory decision-making

Conjoint analysis studies assume a compensatory decision-making process. Researchers suggest that this assumption may be invalid for many different types of products and services. For example, Hauser and Wernerfelt (1990) suggest in one study that 90 percent of the customers use some sort of non-compensatory decision-making process. Non-compensatory choice models suggest that customers simplify their choices by narrowing down their purchase alternatives to a small, manageable number of choices (i.e., evoked set).

This limitation was particularly relevant for this firm. While customers have a large number (15) of potential transportation companies to choose from, they quickly narrowed down the choices to three or four.

Typically, customers use screening rules to determine their evoked set of acceptable products and services. These evoked sets are developed by must-have or must-avoid decision rules (Best 2011) - which were common with this company’s customers. For example, some customers may “have-to-have” a two-day transit time for a particular shipment and thus would only consider services that deliver a two-day transit time. Furthermore, customers may “have to avoid” companies with late pick up times.

In short, if customers use a non-compensatory decision-making process, then this is a major limitation of conjoint analysis studies. The research team felt confident that the majority of transportation customers were using screening rules to determine their evoked sets and simplify the choice.

Irrelevant choice tasks

Because choice-based conjoint analysis studies ignore the customer’s non-compensatory decision-making process, many of the choice tasks in a traditional conjoint analysis study are irrelevant to respondents (Orme 2009). For example, a customer in a conjoint analysis study who “must-have” a two-day lead-time will find all of the choices that ask about a three-day or four-day lead-time to be irrelevant. Given this problem, it is likely that only one of the choices is feasible. Not only is this inconsistent with reality but it may also skew the results significantly.
 
Large sample-size requirements

Finally, conjoint analysis data that is analyzed with hierarchical Bayes typically requires relatively large sample sizes. This can be problematic for studies conducted in a business-to-business environment, where large samples are difficult to obtain.

Overcomes some of the limitations

Adaptive choice-based conjoint analysis (ACBC) is a recent innovation in conjoint analysis and choice modeling, developed by Sawtooth Software (Johnson and Orme 2007), that overcomes some of the limitations associated with conjoint analysis. Most importantly, ACBC does not assume a compensatory decision-making process and thus is better aligned with how respondents actually make choice decisions (Johnson 2008). As a result, the choice tasks are significantly more relevant and the survey experience is more engaging and enjoyable to the respondent. Finally, sample-size requirements are much smaller than conjoint analysis studies (Johnson and Orme 2007), which make this technique more relevant for business-to-business customers.

ACBC builds on a number of choice modeling techniques and integrates them into an adaptive modeling framework. The ACBC survey has three sections, each of which will now be discussed.

Build-your-own concept

ACBC studies start with respondents completing a build-your-own exercise in which they design their ideal product or service (see Figure 1). This ideal product or service is the foundation on which the ACBC experimental design plan is built. Specifically, ACBC uses a “near-neighbor” methodology, which means that competitive product choices are created which are close to the ideal product. In a typical study, 24 near-neighbor product concepts are created on-the-fly.

At this stage, ACBC allows the researcher to specify different price points for different levels of performance. So, if your company wants to charge a premium price for a premium service level, as did this transportation company, this can directly be tested in ACBC. For example, the transportation company in this case study wanted to test whether the market was willing to pay for faster delivery times.

Possible buy/no-buy screening rules

In the second part of the process, respondents examine each of the near-neighbor concepts (Figure 2). For each one, the respondent makes a binary choice, stating either that it is a possibility (“possible buy”) or that it is not a possibility (“no-buy”). Based on the respondent’s decisions, possible screening rules are detected from their choices.

If such screening rules are detected, the respondent is asked to confirm them (Figure 3). Once confirmed, respondents will see only concepts that are consistent with their screening rules. For most respondents, approximately half of the concepts will be removed in this section.

Tournament-style choice exercise

The final exercise requires participants to choose the best product concepts that survived the “possible buy/no-buy” exercise. From this pool of surviving concepts, respondents are asked to select the one they prefer the most (Figure 4). In a typical tournament style of play, three concepts are presented together and the exercise continues until a final winner is chosen, typically about five to six choice tasks.

Support its segmentation strategy

ACBC was the research method recommended to this transportation company. In short, the company needed to add choice modeling or acquisition research that would support its segmentation strategy. It was strongly believed that its customers used a non-compensatory decision-making process when selecting transportation services. Additionally, ACBC requires smaller sample sizes.

Finally, recall that the transportation company wanted to pursue a need-based market segmentation strategy. Researchers suggest that ACBC delivers more robust and stable individual-level results, making this approach much more appropriate for segmentation analysis (Johnson and Orme 2007).

As with all research methods, ACBC has limitations. The first and most relevant limitation is that these studies generally take 50 percent to 200 percent longer to complete than a comparable conjoint analysis study (Johnson and Orme 2007). Second, because of its adaptive properties, ACBC works best with five or more attributes. Third, while ACBC can test market reactions to charging premium prices for premium services, it is important that these price changes are relatively small compared to the overall price of the service. If not, price sensitivity analysis might be skewed as a result.

Adjust its offering

The transportation company learned a great deal about the marketplace following the suggested approach. By better understanding how its customers select transportation vendors (not just their satisfaction), it was able to adjust its offering to be aligned with customers’ desired performance levels. More importantly, identifying need-based segments was a critical link in truly gaining a deep understanding of customers. Undoubtedly, the “average” customer does not exist in this marketplace! Instead, market segments have very different needs and marketplace “averages” actually disguise these differences among customers.

Once recognized and profiled, the company prioritized and targeted segments in its acquisition strategy and developed service levels that would attract and retain these targeted segments. As a result, it was able to focus on the important customers, better utilizing scarce resources for an improved ROI.

Does not assume

ACBC is a new choice modeling approach that overcomes some of the drawbacks associated with traditional conjoint analysis. Most importantly, ACBC does not assume a compensatory decision-making process and thus is better aligned with how respondents actually make choice decisions. As a result, the choice tasks are significantly more relevant and sample size requirements are smaller. Finally, ACBC data are excellent for creating need-based segments.

Is ACBC right for you?

While ACBC is a robust approach for modeling customer choices, it is not a panacea for all choice modeling situations. To decide if ACBC is the right tool to examine how your customers make choices, ask yourself the following questions:

  • Do the majority of our customers use a non-compensatory decision-making process, whereby they use screening rules to simplify the choice?
  • Do we want to learn about screening rules that customers may use to make purchase decisions?
  • Do our customers have to make a complex decision that would typically involve five or more attributes?
  • Do we want to use this data to create need-based market segments?
  • Do we want to test the market reaction to charging premium prices for premium levels of performance?

References

Best, R. (2011). Market-Based Management: Strategies for Growing Customer Value and Profitability. Upper Saddle River, N.J. Fifth Edition, Prentice Hall.

Hauser, J.R. and Wernerfelt, B. (1990). “An evaluation cost model of consideration sets.” Journal of Consumer Research, Vol. 16, March, pp. 393-408.

Johnson, Rich (2008). “A New Approach to Adaptive CBC.” Sawtooth Software Research Paper Series, www.sawtoothsoftware.com.

Johnson and Orme (2007). “Perspective on Adaptive CBC (What Can We Expect from Respondents?)” Sawtooth Software Research Paper Series, www.sawtoothsoftware.com.

Orme, B.K. (2009). Getting Started with Conjoint Analysis: Strategies for Product Design and Pricing Research, 2nd Edition. Research Publications, LLC. Madison, Wis.