What Customers Want (And Are Willing To Pay For) »
Print friendly
In the 1980's, as Americans were loading up on debt and living beyond their means, retailers and catalogers were loading up on services they thought customers wanted. Unfortunately, many of these services, such as subsidized next-day-delivery and 800 numbers, are expensive and, more importantly, may not be as valued by consumers as direct marketers think.
So, exactly how can you improve your understanding of what customers want in products and services? Traditional quantitative surveys either telephone- or mail-based) and qualitative focus groups provide a plethora of useful data, but developing a successful promotion requires a mixture of service and product attributes that may be hard for customers to choose using traditional market research tools. More importantly, having customers rank order a list of attributes doesn't require customers to trade off one attribute with another as they would in the real world when they decide which product to purchase or which promotion to respond to.
Conjoint Analysis
To get around this problem, market researchers adopted the "conjoint" testing methodology to help product developers select the best combination of product characteristics from a wide variety of alternatives. This methodology, which has historically been utilized to design consumer products such as cereals, cars, and toothpaste, has, in the past few years, been successfully adapted to designing the optimum mix of catalog attributes or services.
Using a simple example, a conjoint exercise can be developed to test three different catalog service policies to see which combination would result in increased consumer interest and willingness to purchase. The three policies in question are pricing (50 percent or 33 percent discount compared to full-price retail stores), ordering costs (toll-free versus a toll-call) and shipping method ("free" next day air delivery versus normal UPS or USPS service). In conjoint terminology, the policies are called "attributes" and the different service options within each attribute are called "levels."
Human nature being what it is, most people would choose all three if they could (50 percent off retail prices, toll-free ordering and next-day air delivery). Unfortunately for many catalogers, this service level may not be cost-effective. More importantly, the company may be incurring costs for service functions that don't factor into their customers' purchasing decisions.
How Does it Work?
In its modern form, a statistically valid sample of people is put in front of a personal computer and asked to answer questions about various catalog attributes to determine the "utility" provided by different levels of each attribute. In the first set of questions, customers are asked to provide preferences for different levels within an attribute. Figure 1 shows a conjoint test developed with software from Sawtooth Systems:
Figure 1
Which would be your FIRST CHOICE, assuming everything else to be equal.
1. All items delivered in one shipment (may delay delivery)
2. Items shipped as available (may be more than one shipment)
At this stage, customers will select the policy that provides the most benefit to them since they are viewing each attribute separately and are not required to trade off anything to achieve this attribute level.
Next, the customers are asked to indicate how important each of the policies is to them:
Figure 2
If two catalogs were acceptable in all ways, EXCEPT they had these policies, how important would the TOP POLICY ("A") be in ordering from a catalog?
"A": 800 toll-free order and customer service
versus
"B": Long distance call for order and customer service
4 Extremely Important (I could almost never accept a catalog without feature "A")
3 Very Important (A catalog without feature or policy "A" would have to be important in other ways)
2 Somewhat Important (I would not base my ordering from a catalog on this)
1 Not At All Important
Using each level that was previously selected, the customers are asked to rank them from 4 to 1 in importance. A rank of 4 would indicate that the shopper would not buy from the catalog unless it had the policy; a rank of 3 means that the catalog without the feature would have to be important in other ways; 2 indicates that the policy is not part of their purchasing decision; and a ranking of 1 indicates that it is not at all important to the consumer.
Next, a series of two different catalogs with varying attribute levels are presented to help determine which attribute levels are most important and which attribute levels will be traded off to achieve the desired service level.
Figure 3
Which Catalog Would You Prefer To Order From? Do you prefer catalog 1, 2, or do you have no preference?
Catalog 1... Catalog 2...
Long distance call for order800 toll-free order and customer and customer service service
Unconditional money back Money back on returns if items for all returned itemsunworn, credit if items worn
50% less than leading stores33% less than leading stores ($50 ($50 casual dress for $25) casual dress for $33.50)
CHOOSE A NUMBER TO SHOW YOUR PREFERENCE
CATALOG #1 NO PREFERENCECATALOG #2
1 2 3 4 5 6 7 8 9
As shown in Figure #3, a catalog with 50-percent lower prices and toll calls must be compared by customers with a catalog that has only 33 percent lower prices but doesn't require a toll call. After this task is performed, a series of increasingly attractive "virtual" catalogs are shown with a mixture of policies, developed from previous sections of the test, and the customers are asked to indicate the likelihood that they would shop from each catalog.
The end result of the conjoint test is a series of statistics which, among other things, help a marketer determine the following:
1. The overall importance of a given attribute compared to other attributes. This is simply the summation of the utility for each level of the attribute. When compared to the total utility of other attributes, the marketer can determine if a given attribute is important in the consumer's consideration of a catalog or promotion (for example, a group of consumers who are not time sensitive to delivery will rank service delivery time as an unimportant attribute). Thus, if needed, management can de-emphasize or look for ways to eliminate services that provide little "bang for the buck."
2. The gain in customer utility by moving from one attribute level to another. For example, a cataloger deciding on switching from toll service to 800 service would be in a better position to weigh the costs of 800 service with the increase (if any) in utility provided by enhancing the service level of this attribute.
It should be noted that 800 service has experienced a decline in value over time as more employees "steal" calls from their employers and the cost of long-distance calling has come down.
The reason conjoint analysis is so useful is that it simulates the decisions that consumers make as they evaluate different products in the marketplace. In this example, the consumer may be willing to pay for the toll call because the value of the purchase discount far outweighs the $3.00 long-distance call required to achieve the increased discount.
Other applications of conjoint analysis include determining the mix of product categories offered (sweaters vs. blouses), refining customer return policies (unlimited return period, customer pays shipping versus 30-day return period and customer does not pay shipping) or the overall price orientation (high-end products exclusively or medium and high-end products). But, it is important to remember that the number of attributes testing in a conjoint exercise should not be more than 10, since this leads to a large number of attribute combinations and may be difficult for customers to evaluate. The smaller the number of attributes tested, the easier the test is to administer.
What should be tested?
Customer focus groups are good places to start when trying to determine which attributes and levels should be tested. Focus groups, by their nature, do not provide statistically valid data on consumer attitudes and preferences, but they do provide a forum for surfacing issues and judging the level of consumer emotion surrounding specific issues.
Beyond researching the needs of your customers, remember to perform the same analysis on former customers (those who were steady customers but defected) and on customers who appear as though they will defect (those who were steady, but have cut back on purchases). Why should you be interested in your former customers? Former customers can tell you what you are doing wrong and can help you develop a customer reactivation program.
When combined with a program of regular focus groups and quantitative surveys, conjoint analysis can help direct marketers stay in close touch with their market and their customers. This results in more targeted promotions, increased customer satisfaction and lower prospecting costs.
__________________
This resource is (c) The Direct Marketing
Association, Inc.
by Jeffrey A Steinberg
In the 1980's, as Americans were loading up on debt and living beyond their means, retailers and catalogers were loading up on services they thought customers wanted. Unfortunately, many of these services, such as subsidized next-day-delivery and 800 numbers, are expensive and, more importantly, may not be as valued by consumers as direct marketers think.
So, exactly how can you improve your understanding of what customers want in products and services? Traditional quantitative surveys either telephone- or mail-based) and qualitative focus groups provide a plethora of useful data, but developing a successful promotion requires a mixture of service and product attributes that may be hard for customers to choose using traditional market research tools. More importantly, having customers rank order a list of attributes doesn't require customers to trade off one attribute with another as they would in the real world when they decide which product to purchase or which promotion to respond to.
Conjoint Analysis
To get around this problem, market researchers adopted the "conjoint" testing methodology to help product developers select the best combination of product characteristics from a wide variety of alternatives. This methodology, which has historically been utilized to design consumer products such as cereals, cars, and toothpaste, has, in the past few years, been successfully adapted to designing the optimum mix of catalog attributes or services.
Using a simple example, a conjoint exercise can be developed to test three different catalog service policies to see which combination would result in increased consumer interest and willingness to purchase. The three policies in question are pricing (50 percent or 33 percent discount compared to full-price retail stores), ordering costs (toll-free versus a toll-call) and shipping method ("free" next day air delivery versus normal UPS or USPS service). In conjoint terminology, the policies are called "attributes" and the different service options within each attribute are called "levels."
Human nature being what it is, most people would choose all three if they could (50 percent off retail prices, toll-free ordering and next-day air delivery). Unfortunately for many catalogers, this service level may not be cost-effective. More importantly, the company may be incurring costs for service functions that don't factor into their customers' purchasing decisions.
How Does it Work?
In its modern form, a statistically valid sample of people is put in front of a personal computer and asked to answer questions about various catalog attributes to determine the "utility" provided by different levels of each attribute. In the first set of questions, customers are asked to provide preferences for different levels within an attribute. Figure 1 shows a conjoint test developed with software from Sawtooth Systems:
Figure 1
Which would be your FIRST CHOICE, assuming everything else to be equal.
1. All items delivered in one shipment (may delay delivery)
2. Items shipped as available (may be more than one shipment)
At this stage, customers will select the policy that provides the most benefit to them since they are viewing each attribute separately and are not required to trade off anything to achieve this attribute level.
Next, the customers are asked to indicate how important each of the policies is to them:
Figure 2
If two catalogs were acceptable in all ways, EXCEPT they had these policies, how important would the TOP POLICY ("A") be in ordering from a catalog?
"A": 800 toll-free order and customer service
versus
"B": Long distance call for order and customer service
4 Extremely Important (I could almost never accept a catalog without feature "A")
3 Very Important (A catalog without feature or policy "A" would have to be important in other ways)
2 Somewhat Important (I would not base my ordering from a catalog on this)
1 Not At All Important
Using each level that was previously selected, the customers are asked to rank them from 4 to 1 in importance. A rank of 4 would indicate that the shopper would not buy from the catalog unless it had the policy; a rank of 3 means that the catalog without the feature would have to be important in other ways; 2 indicates that the policy is not part of their purchasing decision; and a ranking of 1 indicates that it is not at all important to the consumer.
Next, a series of two different catalogs with varying attribute levels are presented to help determine which attribute levels are most important and which attribute levels will be traded off to achieve the desired service level.
Figure 3
Which Catalog Would You Prefer To Order From? Do you prefer catalog 1, 2, or do you have no preference?
Catalog 1... Catalog 2...
Long distance call for order800 toll-free order and customer and customer service service
Unconditional money back Money back on returns if items for all returned itemsunworn, credit if items worn
50% less than leading stores33% less than leading stores ($50 ($50 casual dress for $25) casual dress for $33.50)
CHOOSE A NUMBER TO SHOW YOUR PREFERENCE
CATALOG #1 NO PREFERENCECATALOG #2
1 2 3 4 5 6 7 8 9
As shown in Figure #3, a catalog with 50-percent lower prices and toll calls must be compared by customers with a catalog that has only 33 percent lower prices but doesn't require a toll call. After this task is performed, a series of increasingly attractive "virtual" catalogs are shown with a mixture of policies, developed from previous sections of the test, and the customers are asked to indicate the likelihood that they would shop from each catalog.
The end result of the conjoint test is a series of statistics which, among other things, help a marketer determine the following:
1. The overall importance of a given attribute compared to other attributes. This is simply the summation of the utility for each level of the attribute. When compared to the total utility of other attributes, the marketer can determine if a given attribute is important in the consumer's consideration of a catalog or promotion (for example, a group of consumers who are not time sensitive to delivery will rank service delivery time as an unimportant attribute). Thus, if needed, management can de-emphasize or look for ways to eliminate services that provide little "bang for the buck."
2. The gain in customer utility by moving from one attribute level to another. For example, a cataloger deciding on switching from toll service to 800 service would be in a better position to weigh the costs of 800 service with the increase (if any) in utility provided by enhancing the service level of this attribute.
It should be noted that 800 service has experienced a decline in value over time as more employees "steal" calls from their employers and the cost of long-distance calling has come down.
The reason conjoint analysis is so useful is that it simulates the decisions that consumers make as they evaluate different products in the marketplace. In this example, the consumer may be willing to pay for the toll call because the value of the purchase discount far outweighs the $3.00 long-distance call required to achieve the increased discount.
Other applications of conjoint analysis include determining the mix of product categories offered (sweaters vs. blouses), refining customer return policies (unlimited return period, customer pays shipping versus 30-day return period and customer does not pay shipping) or the overall price orientation (high-end products exclusively or medium and high-end products). But, it is important to remember that the number of attributes testing in a conjoint exercise should not be more than 10, since this leads to a large number of attribute combinations and may be difficult for customers to evaluate. The smaller the number of attributes tested, the easier the test is to administer.
What should be tested?
Customer focus groups are good places to start when trying to determine which attributes and levels should be tested. Focus groups, by their nature, do not provide statistically valid data on consumer attitudes and preferences, but they do provide a forum for surfacing issues and judging the level of consumer emotion surrounding specific issues.
Beyond researching the needs of your customers, remember to perform the same analysis on former customers (those who were steady customers but defected) and on customers who appear as though they will defect (those who were steady, but have cut back on purchases). Why should you be interested in your former customers? Former customers can tell you what you are doing wrong and can help you develop a customer reactivation program.
When combined with a program of regular focus groups and quantitative surveys, conjoint analysis can help direct marketers stay in close touch with their market and their customers. This results in more targeted promotions, increased customer satisfaction and lower prospecting costs.
__________________
This resource is (c) The Direct Marketing
Association, Inc.