Optimal Pricing Strategies Through Conjoint Analysis

Mary Jane Tyner, Levi Strauss & Co.
Jonathan Weiner, MACRO

This paper focuses on the topic of obtaining better market intelligence and is concerned with optimal pricing strategies for a variety of jean styles.

Background

Levi Strauss & Company, who produce a wide variety of men's jeans styles, finishes, and price points, were interested in measuring the elasticity of their jeans. As a prelimanary step, MACRO conducted an analysis of panel data collected for Levi Strauss & Company. After extensive regression modeling, a preliminary finding was developed that illustrated price elasticity for the LEVI'S brand, among different customer segments.

Before any decisions were made, management was interested in consumer feedback on price changes. It was also interested in knowing how various competitor responses to a LEVI'S price increase would affect sales and marketing share.

Research Objectives

The objectives of this research were:

• To determine the relationship between price and market share.
• To develop strategic pricing guidelines for LEVI'S men's jeans.

Research Methodology

A total of 304 men, between the ages of 15 and 44, were interviewed in shopping malls in 8 markets across the U.S.

Attribute and Level Specifications

Three primary attributes have been identified that impact product preference. They are jean brand, style, and price. These attributes are ideal for use in testing, because when the actual jean is shown to a respondent, the full jean profile, including b rand and finish, is being presented to the consumer without any additional explanation.

Twelve brands and 5 styles were tested and all jeans were tested in one of five price categories. Relative to their average base price the jeans were priced at two levels above and below base as well as the average base price. The prices ranged from less than $15.00 to slightly over $60.00. The combination of all attributes and levels yielded a total of 300 combinations. Using a one-sixth fractional design reduced the number of combinations to be tested to 50.

Not all styles existed or were available for every brand, so the total number of jeans tested was 48.

Data Collection

The data collection was split into two cells. Each had a unique set of 24 jeans. All jeans were labeled by placing cards showing the jean brand, finish and test price in front of each jean.

Respondents were asked to browse through the layout of 24 jeans, just as they would if they were shopping for jeans in a store. They were not allowed to try on the jeans. They were then handed a set of 24 sort cards. These cards were identical to thoseused for identifying the jeans in the layout. Respondents were asked to separate the cards into piles; one for those jeans they liked and another for those jeans they disliked. Respondents were then asked to rank order the jeans in tearms of their puchase interest. The higher the purchase intrest, the higher the rank order.

The sample was selected so that the characteristics of those who rated the first set of 24 had similar demographic and jean preferences of both groups within market and within age group would be as homogeneous as possible.

Those Respondents who sorted the first 24 would be matched with the ratings of the respondents who sorted the second set of 24. Respondents were matched by market, age, and preferred jean brand.

Analytical Method

To estimate the price sensitivity for each jean, the rank-ordered responses from the sort exercise were processed using the Bretton-Clark conjoint utility estimation program.

A cluster analysis of respondent price utilities was performed to measure the impact of price changes within different levels of price-conscious consumers. Since the utilities are estimates of what attributes and features are important to customers, the segments were defined by using the relative importance of each feature to the overall purchase decision. The relative importance for a feature is defined by:

Using Bretton-Clark's Simgraf software, market simulations were generated to determine the price sensitivity of the LEVI'S jeans as price varied. These simulations were conducted for each price segment and then summed to measure the total market response to price changes.

Market simulations were estimated for each price segment and then aggregated into the final models.

A variety of choice probability models were used to estimate market share and market share shifts as a result of price changes. Current market share estimates were used as a base to estimate the price elasticity for each product.

Company wide profit simulations were conducted by estimating the impact of each jeans contribution to the company's gross profit margin. By applying the following formula, gross profit margin for each LEVI's garment was computed:

Findings: Consumer Segmentation

Based on the results of the trade-off analysis, consumers were segmented into five groups. They are described below. By conducting a series of cross-tabular analyses of the panel data, the size of each price segment was estimated.

 

Group 1. High Price Sensitivity: This group of respondents were the most price sensitive and preferred the more basic styles. The jeans they preferred were usually the least expensive.

Group 2. Low/Medium Price Sensitivity: This group of respondents tended to be less price sensitive than the
first group and, preferred styles that were slightly more expensive.

Group 3. Low Price Sensitivity: These respondents tended
to prefer the widest variety of jeans. More often than the first two segments, they preferred more expensive styles.

Group 4. Zero Price Sensitivity: These respondents
clearly preferred the more expensive styles, and had little concern about price.

Group 5. Jean Driven Zero Price Sensitivity: This group of respondents were driven almost completely by the product. The jeans they preferred were almost always the most expensive styles.

Figure 1 illustrates these results by showing the relative importance of each product attribute in the jeans purchasing decision(for respondents for each segment). Notice how as the importance of price sensitivity decreases, the importance of the jean brand in the decision process increases.

FIGURE 1

Relative Importance Of Product Attributes In Jeans Purchase Decision

Consumer Price segments

Market Simulations

A wide variety of market simulations were conducted to estimate the impact of changing prices of different combinations of LEVI's and key competitor's products.

Due to the confidentiality of the information being presented, brand names, dollar, and market share changes have been masked. Indicies have been created to indicate the magnitude of market share and profit changes due to varying pricing scenarios.

In table one below, the prices of Brand 1 garments were raised, while all competitive garments' prices were held constant. The result shows an increase in profit. This illustrates an interesting phenomenon about high-volume industries. In some cases, volume is so great that the increase in revenue offsets the decrease in units.

Notice that there is a certain amount of cannibalization within brand. While certain styles of Brand 1 lose share, other Brand 1 styles' share increase due to the increase in price by other Brand 1 products.

Table 1

Raise all Brand 1 garments price
Hold all competitors Constant

Index Values*

*Indices indicate changes in volume

Another scenario involved the pricing of different brands and styles of Brand 1 garments at a variety of different price levels. Table 2 illustrates what might happen in this situation.

Table 2

Raise Price of Styles A,D in Brand 12
Raise Price of Styles B,C,E in Brand 12
Raise Price of Styles A,D in Brands 1 thru 6
Hold Brands 7 thru 11 Constant
Index Values*

*Indices indicate changes in volume

Figure 2 further illustrates how a wide variety of pricing scenarios can have varying impact on company profit. As illustrated, in some cases there was a significant loss in both share and profit. In other cases, there was a significant drop in share and increase in revenue; and still in other cases there was an increase in both share and profit. The goal is to choose the pricing strategy that will increase profits the most while minimizing any negative impact on sales.

FIGURE 2

Comparison of Profit and Unit Change Indices By Model

Conclusions

While all these simulations did not lead us to the perfect scenario, they did help Levi Strauss & Co. develop rules to help guide pricing decisions, as well as providing an ability to model potential competitive action with an understanding of their dollar sales and profit impact on Levi Strauss & Co.'s business.

Two important marketing issues were also confirmed by this research:

1. Strategic pricing changes may not be detrimental to a company's profit margin if the decrease in unit sales is offset by a larger increase in revenue due to price increases.
2. Strategic pricing within a product line can cause some cannibalization within brand and increase the unit sales on the lower volume products in the line.

Several additional observations can be made about conducting a strategic pricing study using conjoint analysis.

• Conjoint utilities for each attribute and level are excellent segmentation variables because they illustrate behavioral differences among consumers.
• Using segmentation in conjoint studies also allows the marketer to analyze price and market shifts within different segments.
• The use of readily available statistics such as gross margins, market share estimates, and unit volume will further enhance the research results and allow the marketer to determine the impact of the price changes on the company's bottom line.

Next Steps

Most research is not the "final say" as to what will happen in the marketplace. There are many issues external to the research that can not be controlled for. Conjoint analysis and market simulations are snapshots of a static market and usually can't control for advertising, competitor promotions, and attitudinal shifts in consumer perceptions. Additional testing of any specific pricing strategy should be conducted before national pricing decisions are made. These tests include:

• Test markets
• Controlled store results
• Laboratory simulations

In addition, the following research studies should be conducted to monitor the impact of any price change.

• Brand image tracking should be conducted before and after price changes to monitor long-term effects.
• Pricing studies should be repeated periodically to track changing price sensitivities due to changes in fashion.
• Retailer surveys should be conducted to estimate retailer response prior to pricing changes.

MACRO CONSULTING
Intelligent Decisions