Conjoint analysis is a survey method that measures how much each product feature is worth to buyers by asking them to choose between realistic combinations of features and prices, rather than rating features one at a time. Because people make trade-offs the way they do in real purchases, you learn which attributes actually drive decisions and how much each is worth. SegmentOS includes a Conjoint Analysis template that captures these trade-offs through ranking and choice questions.

What conjoint analysis measures

When you ask people to rate features individually, almost everything comes back as "important." Conjoint forces a choice instead: shown two or three product options that each bundle several features at different levels, respondents pick the one they prefer. From the pattern of choices across many respondents you can estimate the relative value, or utility, of every feature level, and infer how much price someone will trade for a better feature. The result is a ranked picture of what matters most and how attributes compete against each other.

Attributes and levels

Conjoint studies are built from two ideas:

  • Attributes are the dimensions of the product, such as price, brand, screen size, or warranty length.
  • Levels are the specific options within each attribute, such as $19/$29/$39 for price, or 1-year/2-year/3-year for warranty.

Keep it tight: around 4 to 6 attributes with 2 to 4 levels each is manageable. Too many combinations overwhelm respondents and dilute the signal.

Worked example

Imagine a project management tool with three attributes: price ($10, $20, $30 per user), storage (10 GB, 100 GB, unlimited), and support (email only, 24/7 chat). A respondent sees pairs like "$20, 100 GB, email" versus "$30, unlimited, 24/7 chat" and picks one. Repeat this across several screens and many people, and you can estimate that, say, 24/7 support is worth about $8 per user and unlimited storage about $5. That tells you which features justify a higher price.

How conjoint works, step by step

  1. Define your attributes and levels.
  2. Generate option profiles that combine those levels.
  3. Show respondents a series of choices between profiles, plus ranking and importance questions to anchor their priorities.
  4. Analyze the responses to estimate each level's utility and each attribute's overall importance.
  5. Use the utilities to simulate which product configuration and price would win the most preference.

When to use it (and when not)

Use conjoint when you are deciding which features to build, how to bundle a product, or how to price a feature relative to others. A typical sample is 300 or more respondents, since estimating many utilities needs more data than a simple rating survey. Avoid it for very simple products with one or two features, or when you only need an acceptable price range. To find a price range without modelling features, run a Van Westendorp pricing study instead. Conjoint also takes more design care than a standard survey, so budget time to define attributes well.

How to run this in SegmentOS

  1. Start a new study and choose the Conjoint Analysis template.
  2. Enter the product or service you are studying. The template builds a feature-importance ranking, importance ratings for price and brand, and a set of trade-off and payment-model choices.
  3. Open the survey builder to tailor the attributes, levels, and trade-off scenarios to your real product, using the question types that fit, such as ranking and single choice.
  4. Launch to your audience or the audience panel, aiming for at least 300 target buyers.
  5. When responses arrive, use segmenting your results to see how feature priorities differ across customer groups.

For how to launch any template, see Using templates.

Frequently asked questions

How is conjoint analysis different from just asking which features people want?

Direct rating makes nearly every feature look important because there is no cost to saying yes. Conjoint forces trade-offs between bundled options, so it reveals what people will actually give up and how much each feature is really worth.

How many attributes and levels should I use?

A practical range is 4 to 6 attributes with 2 to 4 levels each. More combinations increase respondent fatigue and reduce the reliability of your estimates.

How many responses does a conjoint study need?

Plan for 300 or more respondents from your target market. Conjoint estimates many feature utilities at once, so it needs more data than a simple satisfaction or awareness survey.

Does SegmentOS have a conjoint template?

Yes. The Conjoint Analysis template captures feature importance and trade-off choices and is ready to customize. It is a Premium template with ten questions and about an eight-minute completion time.