“There are 13 named types of conjoint analysis, but almost every study today uses one of three. Choice-Based Conjoint (CBC), where participants pick their preferred product profile from a set, is the default for most studies and for pricing research. CBC rests on the discrete choice methods that won Daniel McFadden the 2000 Nobel Prize in Economics. Adaptive Choice-Based Conjoint (ACBC) tailors the survey to each participant and suits complex products with up to about 12 categories (attributes). Full-Profile Conjoint (FPC) asks participants to rank complete product profiles, and Paul Green and V. Srinivasan’s 1990 review of conjoint analysis says it “works very well when there are only a few (say, six or fewer) attributes.””
The other ten types of conjoint analysis are historical, niche, or parts of hybrid designs. Three questions decide which type a study needs: how many categories it tests, whether price is central, and how much statistical support the team has. For most product teams, the answer is CBC.
What is conjoint analysis?
Conjoint analysis is a survey method that works out which features drive people's decisions, and how much each option within those features is worth to them, by showing people realistic product profiles and recording which ones they choose. Rather than asking someone to rate how much they care about price, brand and battery life one at a time, a conjoint survey shows whole profiles that combine all three and reads the weight of each feature from the trade-offs people make when they can't have everything.
“💡 Academic papers call the features in a conjoint study “attributes” and the values inside them “levels”. OpinionX calls them categories and options, because those words need less explaining, and calls the output values (part-worth utilities) scores. This guide uses categories and options, except inside direct quotes.”
The math behind conjoint analysis goes back to Duncan Luce and John Tukey's 1964 paper on conjoint measurement. Paul Green and Vithala Rao turned it into a practical survey method for market research in their 1971 paper in the Journal of Marketing Research. Since then, researchers have built a whole family of conjoint types over more than fifty years, each for different numbers of categories, survey lengths and goals. Pick the wrong one and you spend participants' time on results that look authoritative and mislead you.
The 13 types of conjoint analysis overlap, because people named them along different lines. Some names describe what participants do (choose, rank or rate), some describe how much of the product each profile shows (every category or only some), and some describe whether the survey adapts to earlier answers. That's why Full-Profile Conjoint and Ranking-Based Conjoint appear as separate types even though a full-profile study often asks people to rank. This guide groups all 13 by what participants do and gives an example of what each one looks like on screen, then ends with a guide to choosing.
What are the 13 types of conjoint analysis?
The 13 types of conjoint analysis split into four families: choice-based, ranking-based, rating-based, and choice-adjacent or historical. The table below compares all 13 on what participants do, whether researchers still use them, what each type suits and what each one costs you.
| Type | What participants do | Still widely used? | Best for | Main drawback |
|---|---|---|---|---|
| 1. Choice-Based (CBC) | Pick one profile from a set, or none | 🟢 The default | Most studies, especially pricing | Profiles get hard to read when they carry many categories |
| 2. Adaptive Choice-Based (ACBC) | Build an ideal product, screen profiles, then choose in a tournament | 🟢 Yes | Complex products with up to about 12 categories | Longer interviews and specialist software |
| 3. Choice-Based with constant sum | Split points or purchases across the profiles in a set | 🟡 Niche | Repeat purchases and purchase volumes | More effort per question than a single pick |
| 4. Menu-Based (MBC) | Build a product from a menu of priced options | 🟡 Niche | Configurable products and add-ons | Heavy setup and specialist analysis |
| 5. Partial-Profile | Choose between profiles that show only some categories | 🟡 Niche | Very large numbers of categories | Needs more questions or more participants |
| 6. Full-Profile (FPC) | Rank or rate a fixed set of complete profiles | 🟡 Small studies | Six categories or fewer | The number of profiles grows fast with each category |
| 7. Adaptive (ACA) | Rate category importance, then compare tailored pairs | 🔴 Legacy | Many categories, no pricing | Tends to understate the weight people give to price |
| 8. Ranking-Based | Sort profile cards from best to worst | 🟡 Small studies | A few profiles, in-person sessions | Can't show whether anyone would buy at all |
| 9. Rating-Based | Score each profile, for example from 1 to 10 | 🔴 Largely replaced | Older academic work | People rate everything highly and skip the trade-off |
| 10. Self-Explicated | Rate each category and option directly | 🟡 Inside hybrids | Very many categories, quick studies | No trade-off between categories at all |
| 11. Hybrid | Self-explicated ratings, then a short set of profile tasks | 🟡 Specialist | More categories than a choice design can hold | Two stages to design and combine |
| 12. MaxDiff (best-worst scaling) | Pick the best and worst item from a subset | 🟢 Yes, as its own method | Prioritising a list of items | Doesn't model products built from categories |
| 13. Brand-Price Trade-Off (BPTO) | Pick a brand at a price, watch it rise, pick again | 🔴 Historical | Brand and price only | Two categories, and participants can see the price climbing |
Which types of conjoint analysis ask people to choose?
Choice-based types of conjoint analysis show participants sets of product profiles and ask them to choose. This is where most modern conjoint analysis happens, because choosing between whole products is how people actually buy. The statistics behind these types come from discrete choice modelling, and Daniel McFadden won the 2000 Nobel Prize in Economics for building the theory and methods used to analyse those choices. Jordan Louviere and George Woodworth brought choice tasks into market research in their 1983 paper on choice and allocation experiments.
Choice-based designs keep growing outside market research too. Health economists call these surveys discrete choice experiments, and Soekhai and colleagues' 2019 review in PharmacoEconomics counted an average of 60 published studies per year from 2013 to 2017, up from 45 per year from 2009 to 2012.
1. Choice-Based Conjoint (CBC)
Choice-Based Conjoint is the most widely used type of conjoint analysis. Participants see a set of two or more complete profiles, pick the one they'd choose (or a "none of these" option, if the survey includes one), and repeat this across several sets. CBC is the conjoint version of a discrete choice experiment, and our guide to discrete choice models explains the five survey formats that sit under that wider umbrella.
Example: a participant sees three smartphones side by side, each showing a brand, a storage size, a battery life and a price, and taps the one they'd buy. Then a new set of three appears with the options shuffled.
CBC is the default because it matches the way people buy, handles price well (price sits on every profile as one more category), and its results feed straight into a conjoint market simulator that predicts market share for products you haven't launched. If you don't have a specific reason to pick another type of conjoint analysis, CBC is the one to use.
CBC is the type behind the conjoint survey tool on OpinionX, which shows each participant sets of 2 to 10 profiles. You can add a none of the above button so people can reject every profile, and use profile logic to stop unrealistic pairings, like a budget phone with the top camera. OpinionX scores CBC results with Hierarchical Bayes, a statistical model that estimates scores for each participant rather than only an average. Those per-person scores let you filter results by customer segment and work out marginal willingness to pay, meaning how much extra someone would pay for one upgrade, like more storage.
Want to see what CBC results look like before you build a survey? A demo walks through the importance chart, the simulator and the willingness-to-pay report on a finished conjoint study.
2. Adaptive Choice-Based Conjoint (ACBC)
Adaptive Choice-Based Conjoint, which Sawtooth Software introduced in the late 2000s, customises the survey to each participant in three stages, which Cunningham, Deal and Chen's 2010 study in The Patient describes. Each person first sets up their ideal product in a build-your-own task, then marks which profiles they'd consider in a screening stage, and finishes with a choice tournament where the winning profile in each set moves on to the next.
Example: a participant first builds their ideal smartphone (Samsung, 256GB, all-day battery, $800). Next they see 20 phones close to that build and mark each as "a possibility" or "won't consider", then choose between the phones they kept, three at a time, until one phone wins.
ACBC is more engaging than standard CBC and copes with more categories than a single choice set can hold. In Cunningham and colleagues' comparison, ACBC took longer to finish (a median of 7.4 minutes against 4.1 minutes for CBC), but participants rated it less boring. The authors suggest keeping ACBC studies under 12 categories, with no more than seven options in each. In a 2009 study, Google researchers Chris Chapman, James Alford and Michal Lahav ran CBC and ACBC with the same participants, and for a line of four products, ACBC's market share predictions had a 15% to 25% smaller error than CBC's when checked against actual market share.
Sadly though, ACBC costs you time: it needs more setup, longer interviews and specialist software. A 2020 review of ACBC by Benedikt Brand and Daniel Baier found that CBC gives you the main facts about the most preferred options with less effort, while ACBC suits more complex products and adds insight, such as which features people reject outright. OpinionX doesn't run ACBC, so if your study needs an adaptive format, you'll need a specialist adaptive platform.
3. Choice-Based Conjoint with constant sum
Choice-Based Conjoint with constant sum is a CBC variant where participants split a fixed number of points, or their next ten purchases, across the profiles in each set instead of picking one. Louviere and Woodworth's 1983 paper covered these allocation tasks alongside single choices. This type of conjoint analysis is useful when you care about how much of each option someone would buy rather than a single pick, which is important for groceries, office supplies and other repeat purchases.
Example: a participant sees three coffee subscriptions and splits their next ten orders between them, say six for the cheapest, four for the mid-priced one and none for the premium blend.
Constant-sum conjoint shares the logic of a constant sum survey, where people split 100 points across a list to show how much more they value one item than another. OpinionX's Points Rank question type runs that list-based version, which our guide to points allocation explains, but it allocates points across single items rather than across conjoint profiles, so it isn't a constant-sum conjoint.
4. Menu-Based Conjoint (MBC)
Menu-Based Conjoint asks participants to build their own product from a menu of options with prices attached, like configuring a car or a software subscription. Liechty, Ramaswamy and Cohen's 2001 paper in the Journal of Marketing Research set out the approach for mass-customised products, using a web-based information service as the test case. MBC models how people trade up and add extras, which standard CBC can't capture, so it's the right type of conjoint analysis when your actual product is a configurator. It takes a lot of setup though, and the analysis needs a specialist.
Example: a participant builds a phone plan by ticking a base plan at $25 per month, adding 50GB of data for $10 and skipping the $8 streaming bundle, with the prices changing from one menu to the next.
5. Partial-Profile Conjoint
Partial-Profile Conjoint is a choice-based method for products with too many categories to fit on one profile. Each choice task shows only a subset of the categories, and the subset changes from task to task, so the survey covers every category without overloading anyone. Keith Chrzan's 2010 paper in the International Journal of Market Research tested partial profiles as a way to handle large numbers of categories. Each answer tells you about fewer categories though, so a partial-profile study needs more choice tasks or more participants than a full-profile CBC study.
Example: a laptop study tests 15 categories, but each choice set shows three laptops described by only five of them, such as screen size, weight, price, battery life and warranty, with a different five in the next set.
Which types of conjoint analysis ask people to rank profiles?
Ranking-based types are the original forms of conjoint analysis. Participants put whole profiles in order from most to least preferred, which gets hard to manage as the number of profiles grows.
6. Full-Profile Conjoint (FPC)
Full-Profile Conjoint is the original form of conjoint analysis, where participants rank or rate a fixed set of complete profiles, each showing one option from every category. FPC gives you a lot of information per participant, but the number of profiles you need grows quickly with every category you add. Green and Srinivasan's 1990 review in the Journal of Marketing put the limit plainly: "The full-profile method of conjoint analysis works very well when there are only a few (say, six or fewer) attributes."
Example: a participant gets 12 cards, each describing a complete smartphone with all four categories filled in, and drags them into order from the phone they'd most like to the one they'd least like.
Because you fix the profiles in advance, you can run Full-Profile Conjoint with ordinary ranking questions, such as rank order, pairwise or MaxDiff. Our guide on how to create a full-profile conjoint survey walks through four voting formats and when FPC is still the better choice over CBC.
Got six or fewer categories and can't decide between Full-Profile Conjoint and CBC? Book a free setup call and a research expert will look at your categories and options and tell you which format fits.
7. Adaptive Conjoint Analysis (ACA)
Adaptive Conjoint Analysis is one of the older conjoint methods, released by Sawtooth Software in the 1980s for studies with many categories. ACA starts by asking each participant to rate how important each category is, then shows them pairs of partial profiles built around the categories they care about most and asks how strongly they prefer one over the other. ACA spread fast, and within five years of launch, Green, Krieger and Agarwal's 1991 critique described it as "one of industry's most popular software packages" for conjoint studies.
Example: a participant first rates storage, battery life, brand and price on a 1 to 4 importance scale. Then they see two phones described by only two or three categories, and move a slider from "strongly prefer left" to "strongly prefer right".
ACA is a legacy method today. It's weak in pricing research because it tends to understate how much weight people give to price, and CBC and ACBC have replaced it for most studies with many categories. It's worth knowing about if you're reading older research, but you'll rarely have a reason to choose it for a new study.
8. Ranking-Based Conjoint
Ranking-Based Conjoint asks participants to rank profiles in order of preference without the adaptive or choice mechanics. It's intuitive and works offline with printed cards on a table, which makes it practical for in-person sessions and workshops. It doesn't scale past a handful of profiles, though. You can include price as a category, but a ranking can't tell you whether someone would buy any of the products at all, so it's weak for setting a price.
Example: a workshop group gets eight printed cards for snack bars, each with a flavour, a size, a sugar level and a price, and sorts them into order on the table.
If your profiles are fixed and few, an online rank order question does the same job as cards on a table, and our guide to stack ranking covers six situations where ranking a list is the right method.
Which types of conjoint analysis ask people to rate profiles?
Rating-based types of conjoint analysis ask participants to score profiles on a scale rather than choose or rank between them. These types are largely historical, because a rating scale brings back the problem conjoint analysis was built to avoid: people don't make trade-offs when they can give everything a high score. Our guide to survey straightlining explains why participants give the same rating all the way down a list, and how to stop it.
9. Rating-Based Conjoint
Rating-Based Conjoint asks participants to rate each profile, usually on a 1 to 10 scale. Rating-based designs were common in the conjoint research of the 1970s, which Green and Srinivasan's 1978 review in the Journal of Consumer Research covers, and choice-based methods have since replaced them for most purposes, for the trade-off reason above. If you're looking at Rating-Based Conjoint, CBC almost certainly serves the same goal better.
Example: a participant sees one smartphone profile at a time and rates how likely they'd be to buy it from 1 to 10, then moves to the next profile.
10. Self-Explicated Conjoint
Self-Explicated Conjoint isn't really a conjoint method in the trade-off sense. Participants rate the importance of each category and how much they like each option directly, and multiplying the two ratings gives a score for every option. Green and Srinivasan's 1990 review credits self-explication with "its simplicity and thus one's ability to use it even when the number of attributes is large". It's most useful as one stage inside a hybrid design rather than on its own.
Example: a participant rates how important battery life is to them on a 1 to 10 scale, then rates how much they like 12-hour, 18-hour and 24-hour batteries, and repeats this for every category.
11. Hybrid Conjoint
Hybrid Conjoint combines self-explicated ratings with a short set of full-profile or choice tasks. The quick self-explicated stage covers every category, and the heavier trade-off stage then refines the scores, an approach Green, Goldberg and Montemayor set out in 1981 in the Journal of Marketing. It's a specialist approach for studies with more categories than a pure choice design can hold.
Example: a participant rates 15 laptop categories for importance, then answers six choice tasks showing complete laptops built from the categories they rated highest.
Which types of conjoint analysis are choice-adjacent or historical?
The last two types sit at the edge of the conjoint family: one is a separate method that researchers often group with conjoint, and the other is an early pricing method that CBC has absorbed.
12. Max-Diff (best-worst scaling)
MaxDiff, also called best-worst scaling, shows participants a subset of items from a list and asks them to pick the best and the worst, repeating this until every item has appeared several times. Researchers often cite Adam Finn and Jordan Louviere's 1992 food safety study as the first published application of the method. MaxDiff often gets grouped with conjoint analysis because both are choice-based and share some math, but strictly it isn't a type of conjoint analysis, because it ranks standalone items rather than modelling trade-offs between categories inside a product.
Example: a participant sees five possible smartphone features (longer battery, better camera, more storage, faster charging, a lower price) and picks the most important and the least important one, then gets a new set of five.
Use a MaxDiff survey when you want to prioritise a list, and conjoint analysis when you want to design a product from categories and options. Our guide to MaxDiff analysis explains how the scores work and how to size a study, and our guide to conjoint analysis alternatives compares MaxDiff and seven other methods against conjoint directly.
Want to see how a MaxDiff ranking and a conjoint importance chart differ on screen? A demo shows both results views on finished surveys, so you can see which output answers your question.
13. Brand-Price Trade-Off (BPTO)
Brand-Price Trade-Off is a historical precursor to modern CBC that tests only two categories: brand and price. Participants pick their preferred brand from a set shown at starting prices, then the price of the brand they picked goes up and they choose again, repeating until the prices run out. CBC now does everything BPTO did while hiding the price changes better, because in BPTO participants can see the price climbing and some start answering strategically. The Gabor Granger method still uses a price that rises after each "yes", but it tests one product at a time rather than a set of brands.
Example: a participant sees five coffee brands at $8 each and picks one, then that brand rises to $9 while the others stay at $8, and they pick again.
How do CBC, ACBC and Full-Profile Conjoint compare?
CBC, ACBC and Full-Profile Conjoint cover almost every conjoint study run today, and they differ most on how many categories they handle, how long the survey takes and how well they support pricing decisions. The table below compares them side by side.
| Choice-Based Conjoint (CBC) | Adaptive Choice-Based Conjoint (ACBC) | Full-Profile Conjoint (FPC) | |
|---|---|---|---|
| What participants do | Pick one profile from each set of 2 or more, repeated across several sets | Build an ideal product, screen profiles in or out, then choose in a tournament | Rank or rate a fixed set of complete profiles |
| Categories it handles | As many as fit readably on one profile | Fewer than 12, with up to 7 options each (Cunningham et al., 2010) | Six or fewer (Green and Srinivasan, 1990) |
| Survey length | Median 4.1 minutes in Cunningham et al.'s 2010 comparison | Median 7.4 minutes in the same comparison | Grows with every profile you add |
| Pricing research | 🟢 Strong: feeds simulators and willingness-to-pay charts | 🟢 Strong: 15% to 25% lower share-prediction error than CBC in Chapman et al. (2009) | 🟡 Weak: a ranking can't show whether anyone would buy at all |
| Analysis effort | Self-serve on modern tools | Specialist design and analysis | Simple, works with ordinary ranking questions |
| Where you can run it | Most conjoint platforms, including OpinionX's free plan | Specialist adaptive platforms only, not OpinionX | Any tool with rank order, pairwise or MaxDiff questions |
How do you choose the right type of conjoint analysis?
Three questions settle the choice of conjoint type for almost every project.
How many categories are you testing? With six or fewer, Full-Profile Conjoint or CBC both work. With more than six, use CBC. Past roughly ten categories, look at ACBC, partial-profile or hybrid designs. For ten or more categories, Green and Srinivasan's 1990 review said "the combination methods (hybrid and ACA) are likely to be more appropriate", and ACBC has since replaced ACA in that role. The number of categories is the biggest single factor in which type of conjoint analysis is even possible.
Is pricing central to the study? If yes, use CBC or ACBC, since both are built for it and the older ranking and rating methods (ACA especially) are weak on price. Willingness to pay is where the wrong type of conjoint analysis costs you most. If price is the only question you're asking, our comparison of Van Westendorp and conjoint analysis explains when a simpler pricing survey is the better fit, and our Van Westendorp pricing guide shows how the most popular of those simpler surveys works.
How much statistical support do you have? CBC on a modern tool is self-serve, because the tool calculates the scoring, the simulator and the willingness-to-pay charts for you. ACBC, Menu-Based and Partial-Profile Conjoint work best with a specialist on the team, and often need one. If you're running the study in-house without a statistician, CBC is the realistic choice, and it's rarely the wrong one.
The other twelve types exist for specific constraints: very high category counts, configurable products, purchase volumes, or comparability with older academic work. If none of those describe your study, CBC is your answer and you can stop reading here. Still not sure conjoint is the right method at all? Our guide to when to use conjoint analysis (and 10 times not to) walks through the cases where conjoint breaks down, and our free research planner asks a few questions about your study and suggests a survey design to match.
Picked CBC and now staring at a blank list of categories and options? Book a free 30-minute setup call and a research expert will go through your categories, options and number of choice sets with you.
Frequently asked questions
What is the most common type of conjoint analysis? Choice-Based Conjoint (CBC) is the most common type of conjoint analysis. Participants pick a preferred product profile from a set, which mirrors how people buy, handles pricing well and feeds directly into market share simulations. CBC is the default on almost every modern conjoint platform.
What is the difference between CBC and ACBC? Choice-Based Conjoint (CBC) uses a choice design fixed before the survey starts, and it doesn't change based on answers. Adaptive Choice-Based Conjoint (ACBC) adapts to each participant through a build-your-own step, a screening step and a choice tournament. ACBC suits complex products, but in one 2010 study it took a median of 7.4 minutes against 4.1 minutes for CBC.
Which type of conjoint analysis is best for pricing? Choice-Based Conjoint (CBC) and Adaptive Choice-Based Conjoint (ACBC) are the best types of conjoint analysis for pricing. Both put price on every profile as a category, which feeds willingness-to-pay and market share simulations. Older ranking and rating methods, and Adaptive Conjoint Analysis (ACA) in particular, tend to understate how much weight people give to price.
Is MaxDiff a type of conjoint analysis? MaxDiff isn't strictly a type of conjoint analysis. It ranks standalone items by asking for the best and worst in a set, while conjoint analysis models trade-offs between categories inside a product. Researchers often group the two because both are choice-based and share some math, but they answer different questions.
How many types of conjoint analysis are there? There are 13 named types of conjoint analysis, but most projects use one of three: Choice-Based Conjoint (CBC), Adaptive Choice-Based Conjoint (ACBC) or Full-Profile Conjoint (FPC). The other ten types are historical, niche, or components of hybrid designs, such as self-explicated ratings.
Once you've picked a type, our complete guide to conjoint analysis covers designing the survey, from choosing categories and options to deciding how many participants you need. Our guide on how to calculate conjoint analysis results works through the scoring step by step, and our review of conjoint analysis software compares ten tools on price so you can pick where to run it. If most of your participants will answer on a phone, our test of mobile conjoint analysis shows how eight survey tools display profiles on a small screen.
CBC runs on the OpinionX free plan: $0, unlimited surveys and unlimited researcher seats, with up to 25 participants per survey, then $900 per year on Analyze for unlimited participants (see our pricing). Create a free conjoint analysis survey and test it on your own team before you send it out.
Want to see a finished conjoint study before you start your own? Book a demo and we'll walk through the results, the simulator and the segment filters on a finished survey.