# Conjoint Analysis

Conjoint analysis estimates how people trade off product attributes by analyzing their choices or evaluations of systematically varied offerings.

- Type: method
- Domain: Research & behavior
- Level: Advanced
- Last reviewed: 2026-10-20
- Original: [Soroush’s Product Wiki](<https://wiki.soroush.uk/topic/conjoint-analysis/>)

## What it means

Conjoint analysis is a family of research methods that represents offerings as combinations of attributes and levels, such as storage, support, and price. A statistical model estimates the contribution of those levels to preference. Choice-based conjoint uses repeated choices among profiles; traditional approaches may use ratings or rankings.

## Why it matters

Customers choose complete offerings rather than isolated features. Conjoint research helps a team explore packaging and pricing tradeoffs and simulate how changing one part of an offer might affect relative preference under explicit assumptions.

## When to use it

Use it when buyers understand the category and the team can describe realistic attributes and levels. Use qualitative discovery first if the important attributes are unknown. A simple list-prioritization question may be better served by MaxDiff.

## How it works

1. Define the target buyer, decision context, and attributes grounded in research.
2. Choose realistic levels and identify impossible combinations.
3. Design tasks and sample requirements with a qualified researcher.
4. Pilot comprehension and field the study, including a suitable no-choice option when relevant.
5. Estimate preferences and check model performance with validation tasks or other evidence.
6. Simulate plausible offers and report uncertainty and market assumptions.

## A practical example

A cloud-storage team compares plans varying in storage, recovery history, team access, and price. Simulations suggest that one segment trades storage capacity for longer recovery history. The team tests that package in a limited commercial trial instead of treating simulated share as a sales forecast.

## Where it breaks

Findings depend on the tested levels and model assumptions. Hypothetical choices omit awareness, distribution, switching costs, and other market conditions unless modeled. Attribute importance is relative to the range studied and should not be treated as a permanent customer trait.

## Common mistakes

- Including unrealistic combinations that respondents cannot evaluate.
- Reading simulated preference share as guaranteed market share.
- Generalizing attribute importance beyond the tested levels.

## Sources

1. [Why and How Conjoint Analysis Works So Well?](<https://sawtoothsoftware.com/resources/blog/posts/why-conjoint-analysis-works>) — Sawtooth Software · Method documentation
2. [Introduction to CVA](<https://sawtoothsoftware.com/help/lighthouse-studio/manual/introduction-to-cva.html>) — Sawtooth Software · Method documentation

## Related topics

- [MaxDiff](<https://wiki.soroush.uk/topic/maxdiff/>)
- [Willingness to Pay](<https://wiki.soroush.uk/topic/willingness-to-pay/>)
- [Survey Design](<https://wiki.soroush.uk/topic/survey-design/>)
