Conjoint analysis: can a product establish itself on the market?

The de­vel­op­ment of a new product always involves con­sid­er­able financial risk. In addition to the costs of actually de­vel­op­ing the product, there are also the costs of raw materials and pro­duc­tion as well as the costs related to market po­si­tion­ing and marketing. Not to mention that even if the market launch is suc­cess­ful, this does not nec­es­sar­i­ly mean that it will be able to establish itself on the market over the long term and generate the desired revenue.

Therefore, companies want to minimize the risk of their product not selling. There are a number of business analysis methods available for this purpose. One of these methods is called conjoint analysis. It requests feedback on customers’ pref­er­ences and needs before launching a product on the market. The results help companies to design a product which meets as many demands as possible.

Conjoint analysis is also known as conjoint mea­sure­ment or the conjoint method. It is a market research method which has been used since the 1970s to determine how important the different at­trib­ut­es of a product are for potential customers.

Conjoint analysis is a type of mul­ti­vari­ate analysis. This means that more than one sta­tis­ti­cal outcome variable (i.e. the product’s at­trib­ut­es) is included in the overall analysis. When only one attribute is used, it is called uni­vari­ate analysis. Using multiple variables allows a company to determine which com­bi­na­tion of at­trib­ut­es (e.g. quality, packaging, price) is likely to meet the highest demands on the market and to develop its product ac­cord­ing­ly.

This analysis method is used to determine which product a customer would choose in a real-life situation i.e., when directly comparing several rel­a­tive­ly similar products which only differ in couple of ways. The following questions are a good starting point: Which of these similar products would a customer buy? Which product attribute has the most influence on their pur­chas­ing decision?

Conjoint analysis is based on a customer survey which is conducted under specific terms. This is not a free-form survey in which par­tic­i­pants can just give their opinion of a product. In this survey, various al­ter­na­tives to a product are provided, and the par­tic­i­pant is asked to indicate which one they would choose when shopping.

This method provides more in­for­ma­tion­al value than a direct customer survey on a specific product. Even if a product receives a positive eval­u­a­tion from a customer ex­press­ing en­thu­si­asm for its in­di­vid­ual at­trib­ut­es, this does not mean that they would choose it over competing products. In a real-life situation, the customer will always evaluate a product by con­sid­er­ing its col­lec­tion of at­trib­ut­es and comparing them with those of competing products. The cost-benefit ratio also plays an important role.

Conjoint analysis recreates these con­di­tions in the customer survey and provides the company with important in­for­ma­tion for product de­vel­op­ment and pricing.

Conjoint analysis requires a bit of effort. You need to be thorough when setting it up and think carefully about which product at­trib­ut­es to select, as well as their spec­i­fi­ca­tions before you start so that the customer survey can provide valuable in­for­ma­tion.

You will need to carefully do the following steps:

1. Product selection: Conjoint analysis is only useful for products or at­trib­ut­es which customers are already familiar with. When it comes to in­no­v­a­tive products, the survey par­tic­i­pants will lack the practical ex­pe­ri­ence needed to re­al­is­ti­cal­ly assess and compare the re­spec­tive at­trib­ut­es. This analysis method works best for everyday products.
2. Product at­trib­ut­es: In this step, you specify the product’s at­trib­ut­es, allowing the survey par­tic­i­pant to make as informed a decision as possible. It is important to avoid spec­i­fy­ing too many at­trib­ut­es or ones which are too different for the in­di­vid­ual product types. Otherwise, the par­tic­i­pant might just make an arbitrary decision, rendering the findings provided by the survey useless. In any case, you should include pricing since it is a deciding factor for consumers’ pur­chas­ing decisions.
3. At­trib­ut­es’ spec­i­fi­ca­tions: You should also avoid providing too many versions of at­trib­ut­es’ spec­i­fi­ca­tions; otherwise, the survey par­tic­i­pants may find them­selves over­whelmed when trying to compare the in­di­vid­ual products. It is better to limit yourself to about three spec­i­fi­ca­tions and not to choose ones which are too different. You also need to take into account the target group’s living situation and (expected) pref­er­ences.
4. Survey: Once the product’s at­trib­ut­es and spec­i­fi­ca­tions have been chosen, a survey is created which presents the possible choices as different product versions (i.e. stimuli). The question of whether to use images or text de­scrip­tions depends on the product and how much effort you wish to expend. In this step, you also need to decide whether to conduct the customer survey using the tra­di­tion­al hardcopy survey, on a computer or online.
5. Target group selection: Defining the target group is already one of the first things you do at the start of any new product de­vel­op­ment. A rep­re­sen­ta­tive number of subjects is randomly selected from the target group and invited to par­tic­i­pate in the survey. The par­tic­i­pants then rank the choices (e.g. by using points) to indicate which of the products they are most likely to buy and which they are not likely to buy.
6. Cal­cu­lat­ing values: When eval­u­at­ing the survey, the par­tic­i­pants choices are assessed using sta­tis­ti­cal methods for mul­ti­vari­ate analysis. This can be done by either using the ap­plic­a­ble sta­tis­ti­cal formulas or special sta­tis­ti­cal software.
7. Eval­u­a­tion: The cal­cu­lat­ed values can then be used to determine which of the product’s at­trib­ut­es and spec­i­fi­ca­tions are par­tic­u­lar­ly important to the par­tic­i­pants, what pricing is ap­pro­pri­ate and how changes in pricing affect demand.
8. Marketing strate­gies: The eval­u­a­tion’s findings can now be used to plan your next steps. First, you decide which at­trib­ut­es will be included in the product and then how to best market it to reach the target group.

This procedure describes tra­di­tion­al conjoint analysis. From this foun­da­tion, other types have been developed which allow more mean­ing­ful results to be obtained from specific questions, and which also address the dis­ad­van­tages as­so­ci­at­ed with the tra­di­tion­al method. These dis­ad­van­tages consist mainly of limiting the choices to just a few product versions and the un­re­al­is­tic ranking of what they are willing to buy which does not happen in real-life sit­u­a­tions.

Among the many types developed over the years, two have become standard methods of mul­ti­vari­ate analysis:

Adaptive conjoint analysis (ACA) is a computer-aided method in which par­tic­i­pants are asked ad­di­tion­al spe­cial­ized questions based on their choices. The next question or choice shown is based on their answer to the previous question. The questions are therefore tailored in­di­vid­u­al­ly to each par­tic­i­pant through­out the customer survey. This means the products provided as choices never show all the possible at­trib­ut­es. Instead, the pre­vi­ous­ly selected at­trib­ut­es are compared with new at­trib­ut­es in the next question. This allows the computer to learn the par­tic­i­pants’ pref­er­ences and to obtain mean­ing­ful in­for­ma­tion which is useful for marketing by providing relevant ad­di­tion­al questions.

Choice-based conjoint analysis (CBC) takes into account economic behavior and decision-making theory by pre­sent­ing products with all their at­trib­ut­es. The par­tic­i­pant can only choose one product with all its at­trib­ut­es in each step of the survey. Unlike in tra­di­tion­al conjoint analysis, the par­tic­i­pant does not rank the choices. This makes CBC the best choice for sim­u­lat­ing a real-life shopping situation which is why it is currently the most fre­quent­ly used analysis method.

The following are ad­di­tion­al types of conjoint analysis:

• Limit conjoint analysis (LCA)
• Hi­er­ar­chi­cal in­di­vid­u­al­ized limit conjoint analysis (HILCA)
• Multi-rule conjoint analysis (MRC)
• Choice-based conjoint analysis with hi­er­ar­chi­cal Bayes es­ti­ma­tion (CBCHB)

To recap the procedure, we would like to end this article with an example of conjoint analysis:

A company wants to launch a new fruit juice beverage on the market and they want to know which product design is likely to be suc­cess­ful be­fore­hand. To do so, it conducts a conjoint analysis of the target group of people living in urban areas between the ages of 25 and 40 with a steady average income. Using this analysis, the company hopes to determine which at­trib­ut­es are important to potential customers and what price they are ready to pay for them.

The product manager defines three at­trib­ut­es she wants to examine using the analysis method: fruit content, packaging and pricing. For each of these, she selects three different spec­i­fi­ca­tions resulting in the following table:

From all the possible com­bi­na­tions, she again selects three, which she presents to the par­tic­i­pants in the customer survey.

• Profile 1: glass bottle (deposit), 55% fruit content, €3.49
• Profile 2: plastic bottle (deposit), 100% fruit content, €3.99
• Profile 3: carton (re­cy­clable), 35% fruit content, €2.99

The par­tic­i­pants must now rank the three product profiles based on their own pref­er­ences. This is the best way to simulate a real-life situation since customers will be eval­u­at­ing each product based on all their at­trib­ut­es when shopping.

Once the customer survey has been completed, the data is analyzed and converted into utility values using sta­tis­ti­cal methods. These values reflect the im­por­tance of each in­di­vid­ual specified attribute (i.e. the partial utility value) when it comes to the customer’s pur­chas­ing decision.

Con­duct­ing conjoint analysis on the different versions of the fruit juice beverage results in the following partial utility values:

Now, the product manager can calculate the total utility value for each of the product profiles presented by adding together the partial utility values as­so­ci­at­ed with the cor­re­spond­ing product attribute spec­i­fi­ca­tions.

Profile 1: 1.8 + 1.8 + 2.3 = 5.9

Profile 2: 1 + 2.6 + 1 = 4.6

Profile 3: 1.5 + 1 + 2.8 = 5.3

This results in figures which can be compared to determine which product profile is most likely to be suc­cess­ful on the market. She can also easily calculate how a change in an attribute spec­i­fi­ca­tion may affect the utility value for consumers. Using the findings obtained through this analysis, the product can be designed to optimize its chance at success with the target group.

Click here for important legal dis­claimers.