Clifford Stoll, the American astronomer and author, once said “Data is not information, information is not knowledge, knowledge is not understanding, understanding is not wisdom”. Yet we still treat data, especially numbers, with huge reverence. It’s certainly true that good data is the foundation for insight. But not all data is created equal.

A nuanced form of analysis

Traditionally, market research has relied on fairly straightforward tools like direct questioning: “Would you buy this product?”. But one of the limitations of this approach is that results do not reflect the trade-offs we typically have to make:

  • Most answers fall into “very important” categories.
  • Answers are sometimes useful for segmenting the market, but not very actionable.
  • Responses can also be manipulated by the way in which the questions are asked, as watchers of ‘Yes, Minister’ will remember.

In other words, it’s all a bit simplistic. In the real world, we weigh up the pros and cons of different options within the same product or service, and then make a choice. To understand and predict customer decisions, and what customers really value in a product or service, we need a research tool that forces people to do the same.

Welcome to the world of conjoint analysis. 

Conjoint analysis, also known as “discrete choice preference” or “stated preference” research, involves presenting people with choices in the form of product or service bundles and asking them to make specific trade-offs, just as they would do in real life.

By understanding how customer make decisions and what they value, product and marketing managers can work out which product or service features balance value to the customer against cost to the company. For example, do customers want low prices, or are they prepared to pay more for specific features? How much more? Does this additional price cover the costs of production? By understanding how many people favoured particular bundles, marketers can use conjoint analysis to forecast potential demand, comparing their own offerings with the competition.

Let’s consider a smartphone product manager who needs to assess which individual smartphone features will persuade more people to buy, and the likely sales revenues.

The first step in designing conjoint analysis is to identify product attributes, such as screen size, battery life, price, operating system and camera resolution. The next is to identify two or three options for each. So for battery life, you might select from 24, 48 or 72 hours.

Then, and this is the cunning bit, the software puts together ‘bundles’ of attributes, and asks customers to select the ideal one from each set. By subtly combining different attributes, marketers gain insight into trade-offs. An example question is shown in Figure 1. Each question would follow a similar format, just with slightly different bundles to choose between.

pastedGraphic.png

Figure 1: Example of a conjoint analysis question. Source: Trial Licence from Survey Analytics

The software then computes the different choices made by the consumers. The results are presented as a set of ‘preference scores’, also called ‘part-worth utilities’ for each attribute or feature level (see Figure 2).

pastedGraphic_1.png

Figure 2: Example of relative importance of attribute and level part-worth utility score. Source: Trial Licence from Survey Analytics

We can also predict the best and worst product or service bundle. All of this means that managers have a much better idea of how people will react, before they launch new products.

The potential of conjoint analysis

Conjoint analysis is widely used when designing new products and services, repositioning existing ones, and developing extensions to existing services or products. It can also be used to measure price elasticity, or forecast demand for products or services. It can even help you to predict profitability. Using a “what if” scenario, it helps to fine-tune product or service features, by forecasting their acceptability to consumers.

If you have a spare five minutes, this video from a conjoint software company summarises the technique.

Fundamentally, conjoint analysis helps you to ask the right questions, in the right way, so that you get actionable insights from your market research. It’s worth considering when you commission your next piece of market research, because it could help you to understand your customers’ behaviour and preferences in a much deeper and more meaningful way. And that could really help your bottom line.

Leave a Reply

Your email address will not be published. Required fields are marked *