A tool to help product managers understand what features customers value
Today we are talking about conjoint analysis, which is a tool you can use to make informed decisions about what customers value and what they will pay for.
If you have to make decisions about what features to include in a new product or the next version of a product, what price to charge for a product, or what the impact on market share will be by introducing a new product, then this discussion is for you.
To learn about Conjoint Analysis, we are talking with Patty Yanes, a market researcher who has led numerous research projects that resulted in new insights about customers and a deeper understanding of their needs. Patty is with Applied Marketing Science (AMS), a firm dedicated to helping product managers with market research. AMS was founded by an MIT professor and is well respected for the work it does.
Summary of some concepts discussed for product managers
[1:49] What kind of problems does Conjoint Analysis solve?
Conjoint Analysis is the industry standard for understanding the features to include in a product. It helps you find out how much people are willing to pay and is the best way to figure out pricing.
[2:56] What are some other tools similar to Conjoint Analysis?
- Van Westendorp: a quicker and easier but less reliable method of pricing
- Gabor Granger: another pricing test
- Max/Diff (maximum difference scaling): a method of ranking features without considering price
- Turf Analysis: a method to understand which bundle of features will allow you to reach the most customers
[5:29] Take us through an example of Conjoint.
Let’s take the example of a pair of headphones. First, we must make sure we have the right inputs. Inputs that affect the customers’ decision to purchase the headphones are: brand, how it fits on the ear, whether it’s wireless or noise-cancelling, and the microphone. You’ll also need to determine your price, which could vary from $20 to $500.
[7:54] How does segmentation fit in?
Segmentation can be part of the process of designing your study or it can be part of cutting the data on the back-end to see how the results vary. For instance, you might create two separate studies for two different headphones, one for gamers and one for audiophiles. Or if you want to develop just one product that you will market differently to different segments, you would design one study with all the attributes and then cut on the back-end to see what’s more important to a gamer vs. an audiophile.
[9:27] What’s the next step in Conjoint Analysis?
Make sure you’re talking to the right people. You need a significant sample size including all the people who are part of the decision to purchase your product. We recommend 300 people, with a minimum of 100 per segment.
Once we’ve designed our Conjoint study, we pre-test our survey. We talk to people in the field to make sure they’re understanding our survey and we’re understanding their answers.
[12:23] How do you recruit and incentivize study participants?
We use research panels that are already in existence. The incentives may have to be quite high since people’s time is valuable. Scrappy solutions include talking to customers in a store or reaching out to friends and family.
[15:29] Tell us more about the pre-test.
We recruit a small sample of the same type of people who would be taking our general survey and run through the survey with them. We ask them questions to discover what assumptions they’re making about the survey and whether they understand the survey in the way we expect.
[17:35] How do we field the study?
We go into the field with a panel, send the survey out, and get the data back. Then, before analysis, we clean the data, making sure everyone paid attention appropriately and that we have the right people who will give relevant answers.
[19:36] What do we do with the data?
There are many different ways you can analyze the data, and today almost all of them can be done with software. The industry standard for Conjoint software is Sawtooth Software.
First, perform a regression to see what’s most important to people. We use the hierarchical Bayes regression, but there are many different options. The regression equation tells you what’s driving the overall purchase decision for the product. This allows you to look at which features are more and less important.
Next calculate willingness-to-pay for each feature. For example, we could look at willingness-to-pay for two different types of noise cancellation.
[24:20] What kind of questions would we ask in the survey to determine which features are more important?
Conjoint is all about simulating choices and tradeoffs people are making. You present survey participants with three to five option of different headphones that all have different attributes and different prices, and you ask them to select the one they prefer. You repeat this about 12 times with different combinations, and the software determines which features are most valuable.
[27:08] How do you choose which combinations to put in the survey?
We avoid putting in combinations that don’t ever happen. For example, noise-cancelling headphones probably can’t be feather-light. However, you might still want to include them in the study so that if the technology becomes available in the future, you have information about whether customers are interested.
[29:45] How do we choose which prices to include in our survey?
We present different buckets of pricing. The different products vary by seven or eight different price levels. One pair of headphones has noise cancellation and costs $60, and another does not have noise cancellation but costs $40. The customer decides how much it’s worth to them.
- Learn more about Applied Market Science
- Check out the AMS blog
- Connect with Patty and AMS on Twitter
- Connect with Patty on LinkedIn
- Learn more about Sawtooth Software for Conjoint Analysis
“The customer’s perception is your reality.” – Kate Zabriskie
Thank you for being an Everyday Innovator and learning with me from the successes and failures of product innovators, managers, and developers. If you enjoyed the discussion, help out a fellow product manager by sharing it using the social media buttons you see below.