13 Types of Conjoint Analysis Explained (With Image Examples)
Conjoint analysis is a popular research method for understanding which attributes are most important to customers when considering whether to purchase a product.
However, conjoint analysis is arguably the most misunderstood (and misused) of all research methods.
I’ve written previously about 10 research scenarios where you should NOT use conjoint analysis and 8 alternative methods for measuring people’s preferences that are easier, cheaper, and more flexible formats — I suggest you take a quick look at these two guides before this one!
But, if you’re determined to run a conjoint study, this post breaks down the 13 different types of conjoint analysis along with explanations and picture examples of each method. I hope it helps you find the right approach for your research project :)
Contents:
When To Use Conjoint Analysis?
What Is Conjoint Analysis?
Definition
Conjoint analysis is a survey format that measures the relative importance people feel towards different attributes (like price, brand, or features) when comparing products. Conjoint assumes that people evaluate products based on their combination of attributes. The result of a conjoint survey tells you the relative importance of each attribute (eg. price vs brand) and the options you included within each attribute (eg. Coca-Cola vs Pepsi).
What Does A Conjoint Analysis Survey Look Like?
Conjoint surveys show respondents three or four product “profiles” that represent variations of the same product. Each profile has the same list of attribute categories, but the options for each attribute vary by profile. When a respondent picks their favorite profile, their vote is recorded and a new set of profiles appears.
Four Components Of A Conjoint Analysis Survey
1. Question: Gives the respondent context about the product decision they’re simulating.
2. Profiles: A set of attribute values that make up a product offering.
3. Attributes: The characteristics of a product that are relevant to customers’ purchase decisions.
4. Levels: The range of values you include within each attribute (minimum 2, maximum 7).
Example Conjoint Analysis Survey And Results
You’re running a conjoint survey to decide what burger to prepare for your weekend barbecue. You pick three burger attributes: filling, sauce, and bun type. Each attribute has four ’levels’ — for burger fillings, you decide to include pork, chicken, vegan, and beef options.
Our barbecue conjoint can help us answer questions like:
Within the overall burger, which attributes are most important to people? Is the choice of filling more important than the sauce? Is the sauce more important than the bun type?
Within each attribute, which ’levels’ have the highest relative importance to people? For the filling, will people be more interested in having our burgers if we switch the pork filling to a chicken, beef or vegan patty instead?
Conjoint Analysis Result — Relative Importance of Attributes (left) and Levels (right)
Our results (above) show that:
When people make decisions about which burger they want, 50% of their decision is based on the filling. In comparison, the bun only accounted for 10% of people’s preferences.
Beef and chicken were the two most popular fillings.
When To Consider Using Conjoint Analysis?
Conjoint analysis simulates a purchase decision where a customer compares a group of similar products. It takes a very rational view that customers decide which product is their favorite by comparing individual attributes like price, brand, and features. That means that conjoint is really only suited to research scenarios:
That simulate a simple, well-considered purchase between similar products.
Where only one person is involved in the purchase decision.
Where the customer already knows what kind of product they need.
Where you know which attributes the customer use to compare products.
Where those attributes have no overlap in meaning or possible options.
These are very specific requirements because conjoint is a very specific research format! Examples of research projects that meet these criteria include:
Modeling market share scenarios against competitor products.
Informing a product bundling or product mix strategy.
Calculating willingness to pay for existing product attributes.
There are plenty of other research methods that model people’s preferences using forced trade-offs that give you a lot more flexibility than conjoint analysis for a much cheaper price.
10 Conjoint Analysis Methods
3 Conjoint Analysis Adaptations
— — —
10 Most Common Conjoint Analysis Methods
1. Choice-Based Conjoint Analysis (CBC)
Choice-Based Conjoint (CBC) is the most common approach to conjoint surveys today. Respondents are shown a set of product “profiles” (usually 3-6 at a time) and asked to choose the option they would “buy”. This is repeated multiple times, with the attribute “levels” in a profile changing each time the respondent votes.
2. Best-Worst Conjoint Analysis (BWC)
Respondents identify the “best” and “worst” options from the set of profiles shown. This approach combines the voting format of MaxDiff Analysis (which is a form of best-worst scaling) with the attribute-based profiles structure of Conjoint Analysis. It requires some extra work from respondents, but collects more data per set of profiles shown.
3. Ranking-Based Conjoint Analysis (RBC)
Instead of respondents selecting just one profile or the best/worst options, they rank all the profiles from 1st to last. It uses the format from ranked choice voting, which is one of the best-known survey formats for ranking things, and applies it to conjoint analysis.
4. Constant Sum Conjoint Analysis (CSC)
Respondents are given a fixed number of points that they can allocate amongst the profiles shown in whatever distribution they choose, allowing them to show the magnitude of their preference towards the profiles shown. This approach combines the Constant Sum (also known as Points Allocation) format of normal ranking surveys with profiles-based voting in Conjoint Analysis.
5. Rating-Based Conjoint Analysis
Respondents are asked to rate each profile shown on a scale from 0 to 100. Rating-based approaches are generally not recommended for understanding the relative importance people feel towards a set of options, as respondents tend to use a small range of the available points (eg. giving everything 60-80 points). Rating-based approaches also allow for ties, producing a higher proportion of useless data when it comes to calculating relative importance.
6. Volumetric Conjoint Analysis (VCA)
Respondents are given a scenario or task and told they can purchase as many of each product profile as they choose. Volumetric conjoint allows pricing to become a more realistic constraint in how respondents choose profiles — particularly when a limit is put on the budget they can spend on each set of profiles.
7. Full-Profile Conjoint Analysis (FPC)
Full-Profile Conjoint follows a similar format to Choice-Based Conjoint, except the attribute levels on each profile are a static set of values that do not change. In this sense, it’s more like Pairwise Comparison or MaxDiff Analysis where respondents vote on single options rather than variable-based profiles. FPC is often used when you want to explore the overall attractiveness or appeal of different product configurations as a whole.
8. Menu-Based Conjoint Analysis (MBC)
Respondents are then given a series of menus where they can build their preferred product or service by selecting one level from each attribute. The attributes available in each menu vary to test which attributes and levels have the highest relative importance to respondents.
9. Self-Explicated Conjoint Analysis (SEC)
The Self-Explicated Conjoint isn’t really a conjoint analysis at all — it uses the same setup as a conjoint, where you assign a number of attributes (categories) and the range of “levels” (options) for each attribute. But then it asks the respondent to rank the levels altogether using a standard choice-based ranking format like points allocation, constant sum, maxdiff analysis, or ranked choice voting. Some other formats of Self-Explicated Conjoint use slightly different approaches, but overall it generally just breaks attributes down into attribute-by-attribute ranking exercises.
10. Dual-Response Conjoint Analysis (DRC)
Respondents complete a normal conjoint question initially (usually following the choice-based format), and then a follow-up question appears based on their choice. This breaks each set of profiles down into two data points — the selection of profile measures attractiveness, while the follow-up question measures likelihood of purchase.
— — —
3 Adaptations To Conjoint Analysis
1. Adaptive Conjoint Analysis (ACBC)
Adaptive conjoint surveys learn from respondents’ choices and change what they show next. This can either be done through screening questions (ranking, rating, duel-response, etc.) or simply by learning from standard choice-based profile selection. You can apply this self-learning to any conjoint method, but Adaptive Choice-Based Conjoint (ACBC) is the most common adaptive method used. ACBC surveys take considerably longer for respondents to complete, are a lot harder to set up, and are much more expensive as you are always going to require expert support to design and analyze an adaptive conjoint survey.
2. Time-Series Conjoint Analysis (TSC)
Time-Series Conjoint is not actually a different format for asking conjoint questions — it is a different way to analyze conjoint results. A time-series analysis looks at respondents in “waves” based on when they completed the survey, helping researchers understand how preferences are changing over time.
3. Latent Class Conjoint Analysis
Latest Class Conjoint Analysis is not a survey format, it’s an approach to analyzing conjoint results. Latent Class is a way of looking at patterns in voting data that can be used to segment respondents into groups with similar preferences. This is a form of segmentation analysis (eg. needs-based segmentation).
— — —
Further Reading About Conjoint Analysis:
• 8 alternatives to conjoint analysis for trade-off research
• 5 criteria for assessing whether conjoint analysis suits your research scenario
• 10 examples of research scenarios that should NOT use conjoint analysis
• 10 most popular tools for creating conjoint analysis surveys (Free vs Paid)
• How to calculate conjoint analysis results in 8 Steps [Advanced]
About The Author:
Daniel Kyne is the Co-Founder of OpinionX, a free research tool for stack ranking people’s priorities — used by thousands of product teams to better understand what matters most to their customers. OpinionX has a bunch of free research methods for ranking people’s preferences — including Conjoint-style ranking methods like Pairwise Comparison and Constant Sum. Try it now!
Subscribe to The Full-Stack Researcher newsletter, where we share research strategies for founders and product teams: