Conjoint analysis design
Conjoint analysis and trade-off studies are amongst the most sophisticated forms of market research because they calculate the underlying values (utilities) that customers base their decisions on and so can provide detailed forecasts and estimates of demand which depend on specific product or service design features. But like any form of research, the quality of the output depends heavily on the quality of the design.
For conjoint analysis, in particular this means choosing the right flavour or type of conjoint to use and ensuring that the design of the attributes and levels, the way the profiles are shown and the number of choice tasks meets the research, analysis and business requirements.
The basis of conjoint design splits into three interrelated parts. Firstly there is the choice of the flavour or type of conjoint analysis that is to be used. Different types have different strengths and weaknesses. Some forms can be carried out on paper, some require the use of computer or internet-based interviews, some allow large complex designs, others are more robust in the face of issues like pricing. This also reflects the analysis method that will be used.
The second element of design are the attributes and levels that make up the product or service. In other words what options are there for different products in the market (see below). Large numbers of attributes and levels need different techniques to products or services with small numbers of attributes or levels.
The final inter-related element is the sample itself and the contact methods available. Can the interview be carried out face-to-face, or over the internet, or by post and phone, and how long will the questionnaire be?
What are attributes and levels?
Attributes and levels form the fundamental basis of conjoint analysis. The idea is that a product or service can be broken down into its constituent parts - so for instance a mobile phone has a size, weight, battery life, size of address book, type of ring. Each of these elements making up a generic mobile phone is known as an attribute.
When we compare between mobile phones each will have a different specification on each of these attributes. You might have choices in terms of battery life between 72, 108, 120 hours of battery life. Each of these options is known as a level of the battery life attribute.
So to give another example, a car might have an attribute 'colour' which is made up of the 'levels' red, blue, green, white.
This breaking down of products and services into attributes and levels is an extremely powerful tool for examining what a business offers and what it should be offering. For new product development, combining this product breakdown with an understanding of what the customer values most means that the business can focus its efforts on those areas of most importance to customers.
To identify attributes and levels, our standard approach is to run workshops that explore a typical customer journey and investigate all the potential uses and touch points for a product from finding initial information, to opening the box, using the product and down to factors like cleaning, security and problems. It's common to find more than 200 attributes which then need to be whittled down to the key attributes and levels that need to be measured.
In addition, each level can also be thought of as a performance target. If the customer, on the attribute 'delivery', wants the level "next day delivery" and we're perceived to be offering "48hr delivery" then there is a gap to make up. What is more with conjoint analysis, it is possible to calculate what the value of making up that gap is to the business. Compare this to the cost of doing it and the business can decide whether it is worth it.
Note that this is far more useful than simply knowing that we score 6.9 out of 10 on delivery. This has no meaning. It doesn't say where the customer thinks we are and it doesn't tell us what we need to do to get higher, or whether it is worth trying to improve this score, or what it will cost. For a lot of strategic work attributes and levels are far more powerful tools than scales and scores.
Attributes and levels in conjoint analysis
In conjoint analysis, attributes and levels have to follow certain rules to ensure the conjoint is valid but there are also best-practice design criteria to make them particularly useful.
Firstly, each attribute has to be independent, that is it should not overlap with other attributes. So colour and fuel economy are clearly not related, so they could appear together. However, some things like "car shape" and "number of passengers" aren't independent - a 7 seater coupe isn't realistic.
There are also more subtle effects - certain attributes have halo-effect on others around them. For instance, if one level were "gold-plated handle", many people would infer that the rest of the product was also of better quality when there is no other information to support this. The main difficulty this causes is that price and brand need to be treated extremely carefully in conjoint studies to produce valid results.
Each level also needs to be capable of being read and understood on its own. Although attributes are used to help break a product down and in analysis, when presented to respondents all the respondent sees are the levels.
Independent and readable levels are important from an analysis point of view, but for the conjoint to be useful it also needs to ensure that the range of attributes cover all the areas that are important to the customer, and that the range of levels cover all possibilities from worst-case to blue-sky. Having said that if you do want to focus on a particular element of choice, it is possible to use an "all other things being equal" scenario, but if you're looking at the type of car-radio this limitation won't tell you about the importance of the car-radio in the overall choice of car.
For many products, particularly in business markets, service can be more important than the actual product. By using both product and service attributes in the same conjoint it is possible to see how customers trade-off service against features. However, care has to be taken to balance the attributes to prevent biasing the outcome one way or another.
It is also vital that the levels used describe real and where possible measurable and actionable performance. You could run a conjoint with levels "very good fuel economy" "good fuel economy" "not so good fuel economy". But, as with the scales above, these don't mean anything. If you come out with "good fuel economy" and the customer wants "very good fuel economy" what do you do?
One good test is if you showed it to the shop-floor (or the call centre) would they know how to deliver it. A second test is whether it is a statement you could use on a packet or in a brochure.
For help and advice on carrying out conjoint analysis contact our specialists firstname.lastname@example.org or phone +44(0)20 7193 6640 or +1 713-983-8700.