Types of Conjoint Analysis: Which Method Fits Your Research Question — and What It Costs to Get It Wrong

Using the wrong type of conjoint for a specific decision is not a minor inefficiency. It produces preference data that is either less valid, less reliable, or less applicable to the commercial question than the right type would generate — and it produces it confidently, with the full statistical apparatus of a properly run study.

The types of conjoint analysis are not interchangeable. Each differs in its task design, data collection procedure, statistical estimation method, and the specific category of commercial question it is most suited to answer. Selecting between them is a foundational methodological judgment, not a default.

This blog covers each type — what it does, where it performs best, and where its structural limitations make it the wrong choice.

Traditional Full-Profile Conjoint

Traditional conjoint is the original form of the methodology — the starting point for understanding the family of techniques that has developed from it.

Each respondent evaluates a set of product profiles one at a time, rating each on a preference or purchase likelihood scale. Each profile is a complete product description defined by specific levels of all attributes in the design. Ordinary least squares regression analyses the rating data to estimate utility weights for each attribute level.

When it is the right choice: Simple product categories with four to six attributes. Analytical requirements that can be served by aggregate-level utilities without individual-level estimation. Straightforward studies where the smaller sample size requirements make it the most cost-efficient option.

The limitation that matters most: Traditional conjoint has poor ecological validity relative to choice-based designs. Rating each profile individually, without comparison to competitive alternatives, does not replicate the actual purchase decision where consumers compare multiple products simultaneously. The resulting utility estimates can be inflated relative to what consumers would reveal in a realistic competitive evaluation context — which makes them less reliable as inputs to pricing and product design decisions where competitive dynamics matter.

For most commercial brand research applications today, traditional conjoint has been superseded. Understanding it is useful for interpreting legacy research and for the smallest, simplest studies. It is not the right default for most decisions.

Choice-Based Conjoint (CBC)

CBC is the most widely used form of conjoint in commercial brand research — and for good reason.

Respondents are presented with sets of product profiles simultaneously, typically two to five per set, and choose which one they would purchase. That single design difference — choosing between alternatives rather than rating each in isolation — is what gives CBC its primary advantage: it replicates the structure of an actual purchase decision more closely than any other standard conjoint format.

The statistical analysis uses logistic regression or multinomial logit models to estimate utility weights from the pattern of choices. Individual-level utility estimation uses hierarchical Bayes methods, which produce reliable individual-level preference parameters even from modest individual respondent contributions — enabling segment analysis and market simulation across heterogeneous consumer populations.

When it is the right choice: Most commercial brand research applications involving competitive product choice. New product design, pricing research, portfolio optimisation, brand price premium estimation, and any situation where the research question concerns how consumers would choose among alternatives in a real market context. For a complete guide to how choice-based conjoint analysis works and what its outputs actually mean, the dedicated blog covers the full mechanics.

The limitation that matters most: CBC requires larger samples than traditional conjoint for equivalent statistical precision, and the experimental design management is more complex. In categories with very large attribute sets — more than seven or eight attributes — the cognitive complexity of evaluating full profiles can push respondents toward simplification strategies that undermine the validity of the choice data.

Adaptive Conjoint Analysis (ACA)

Adaptive conjoint analysis modifies the profiles presented to each respondent based on their responses to earlier questions, adapting the conjoint task to the individual respondent's preference space rather than presenting the same fixed profiles to all respondents.

ACA typically begins with an importance screening phase — respondents rate the relative importance of attributes — then proceeds to paired comparison tasks constructed specifically for the attribute space most relevant to that individual. This adaptive procedure concentrates choice tasks where each respondent's preference function is most discriminating, producing more precise individual-level utility estimates with fewer total tasks than a fixed design would require.

When it is the right choice: Complex product categories with large numbers of attributes where the full factorial design space cannot be covered efficiently with a fixed task structure. Research applications where individual-level utility precision is the primary goal — particularly when the resulting utilities will be used for detailed segment analysis or personalised product recommendation.

The limitation that matters most: ACA has lower ecological validity than CBC because the adaptive profile generation does not replicate a real competitive market context. The profiles respondents evaluate are built around their own preference space, not around the actual competitive alternatives in the category. For decisions where competitive dynamics matter as much as individual preference precision, this is a meaningful trade-off.

Adaptive Choice-Based Conjoint (ACBC)

ACBC is the most methodologically sophisticated form of commercially available conjoint. It combines the ecological validity of choice-based tasks with the adaptive efficiency of adaptive conjoint.

The design uses a sequence of phases: first identifying the range of product configurations each respondent considers acceptable, then presenting choice tasks using profiles concentrated within that acceptance space. The result is the most reliable individual-level utility estimates available from any standard conjoint format — particularly for studies with large attribute sets.

When it is the right choice: Complex product categories with many attributes where neither standard CBC nor ACA would efficiently produce the individual-level utility precision required for detailed segment analysis. When the research decision requires maximum precision and budget permits the premium investment.

The limitation that matters most: Operational complexity and cost. ACBC requires more sophisticated survey software, more complex design management, and longer respondent sessions than CBC. For most commercial brand research applications, the additional precision over well-designed CBC does not justify the additional investment. ACBC earns its cost when the decision genuinely requires it — typically in high-stakes product architecture or multi-tier pricing decisions where granular segment-level utility differences are commercially significant.

MaxDiff Scaling: Related But Different

MaxDiff — maximum difference scaling — is sometimes grouped with conjoint methods because it shares the trade-off task structure. Respondents identify the most and least preferred items from subsets of a larger item set, producing relative preference scores for each item.

But MaxDiff and conjoint are answering fundamentally different questions.

MaxDiff produces relative preference ordering of individual items — features, messages, brand attributes, benefits. It does not estimate how attribute levels combine to produce product preference, and it does not produce the utility function needed for product design trade-off analysis or willingness-to-pay estimation.

Conjoint analysis produces attribute-level utility weights from product profile choices. It answers trade-off questions about product design, pricing, and competitive positioning that MaxDiff cannot address.

MaxDiff is appropriate for: Brand perception scaling, feature prioritisation, message hierarchy development, and any research question concerned with the relative importance or preference ordering of a set of items. For a direct comparison of the two methods and the specific decision contexts where each applies, conjoint analysis vs MaxDiff covers the full distinction.

MaxDiff is not appropriate for: Product attribute trade-off analysis, pricing research, portfolio optimisation, or willingness-to-pay estimation. It is not a substitute for conjoint in these applications.

How to Choose — A Practical Decision Framework

The type selection should be driven by three factors, in this order:

The commercial decision and the research question it requires. If the question is how consumers would choose between competing product alternatives — with price as one attribute — CBC is almost always the right answer. If the question is how consumers rank-order a set of features or messages, MaxDiff is the right answer. If the question involves a very large number of product attributes with complex individual-level preference heterogeneity, ACA or ACBC may be more appropriate.

The number of attributes. Four to six attributes: traditional conjoint or CBC. Seven to ten attributes: CBC with careful design management or ACA. More than ten: ACBC or ACA.

The precision requirements of the output. Aggregate-level utilities for population-level decisions: traditional conjoint or CBC with standard analysis. Individual-level utilities for segment analysis or market simulation: CBC with hierarchical Bayes or ACA. Maximum individual-level precision for complex multi-segment decisions: ACBC.

For the complete framework connecting <u>conjoint analysis to specific commercial decisions and what each type of output enables</u>, the interpretation guide covers the full picture. And for the foundational understanding of what conjoint is measuring and why utility weights are the primary output, the conjoint analysis definition blog covers the conceptual foundation that makes type selection genuinely informed.

Frequently Asked Questions

What are the main types of conjoint analysis?

Traditional full-profile conjoint, choice-based conjoint (CBC), adaptive conjoint analysis (ACA), adaptive choice-based conjoint (ACBC), and MaxDiff scaling. Each differs in task design, statistical method, ecological validity, and the types of commercial decisions it best supports.

Which type of conjoint analysis is most commonly used?

Choice-based conjoint (CBC) is the most widely used in commercial brand research. Its combination of ecological validity — replicating how consumers actually choose between alternatives — and analytical flexibility through hierarchical Bayes individual-level estimation makes it the standard recommendation for most product design, pricing, and portfolio decisions.

What is the difference between traditional conjoint and choice-based conjoint?

Traditional conjoint asks respondents to rate profiles one at a time on a preference scale. Choice-based conjoint asks respondents to choose between profiles presented simultaneously — replicating the structure of a real purchase decision. CBC produces utility estimates with higher ecological validity because the choice task is closer to what consumers actually do when they buy.

When should you use adaptive conjoint (ACA) instead of CBC?

When the product category involves more attributes than CBC can handle efficiently — typically more than seven or eight — and when individual-level utility precision is more important than competitive market replication. ACA is more efficient for large attribute sets but has lower ecological validity than CBC.

What is ACBC and when is it worth the extra investment?

Adaptive choice-based conjoint combines the ecological validity of CBC with the adaptive efficiency of ACA, producing the most precise individual-level utilities of any standard conjoint format. It is worth the additional cost when the decision requires maximum individual-level precision across large attribute sets — typically complex product architecture or multi-tier pricing decisions where granular segment differences are commercially significant.

Is MaxDiff a type of conjoint analysis?

MaxDiff uses a similar trade-off task structure to conjoint but is a distinct method. It produces relative preference ordering of individual items — features, messages, brand attributes — rather than attribute-level utility weights from product profiles. It cannot be used for product design trade-off analysis, pricing research, or willingness-to-pay estimation. The two methods answer different questions.

How do you decide which type of conjoint to use?

By working through three questions in order: what commercial decision does the research need to support, and which task structure best replicates the real decision context that question involves? How many attributes does the design require? What level of analytical precision does the output need — aggregate or individual-level? The answers to these three questions will point to one type more clearly than the others.

Selecting the right type of conjoint is the methodological decision that most directly determines whether the resulting preference data is fit for the commercial decision it is meant to inform. Pulse AI Research designs and runs conjoint studies matched to the specific research question — with the type, design, and analytical approach determined by what the decision requires, not by what is most familiar.

Related reads: Conjoint Analysis: A Complete Guide for Brand Teams | Choice Based Conjoint Analysis: When and How to Use It | Conjoint Analysis vs MaxDiff: Which Method Fits Your Question