← All articles
How-to

A lightweight conjoint survey for pricing a brand-new product

You don't need a market-research firm to run conjoint analysis. Here's a scaled-down version a small pricing or product team can run in about a week.

Sarah Kwon · July 30, 2026
Advertisement · 320×100

Van Westendorp is the fastest way to get a defensible price range for a new product, but it can't tell you how price trades off against specific features — which matters when you're deciding, say, whether to bundle a feature in or sell it as an add-on. That's where conjoint analysis earns its keep: it asks respondents to choose between full product packages that vary on several attributes at once, including price, and infers how much each attribute is actually worth to buyers.\n\nA full commercial conjoint study can run into the tens of thousands of dollars and take weeks. A scaled-down version is workable for most teams: pick no more than 4-5 attributes that matter most (including price as one of them), define 3-4 realistic levels for each, and generate a set of 8-12 package comparisons using a simple fractional-factorial design — free tools and open-source R packages (such as the `conjoint` or `support.CEs` packages) can generate a balanced design without a market-research background.\n\nField the survey to at least 100-150 respondents from your actual target segment if you can get them; fewer than that and attribute-level estimates get noisy. For each comparison, ask respondents to pick which full package they'd choose, not to rate features individually — the forced trade-off between price and features is the entire point, and it produces far more realistic answers than asking people to rate importance on a 1-5 scale.\n\nAnalyze results with a simple choice model (multinomial logit is standard and doesn't require a data scientist to run in R or Python) to get relative importance weights for each attribute and a willingness-to-pay estimate for each feature level. Treat the output as a planning input for pricing and packaging decisions, not a guaranteed revenue forecast — conjoint measures stated preference in a survey, not actual purchase behavior, and tends to overstate willingness to pay for feature-rich, high-price bundles relative to what people do with real money.

Advertisement · 300×250
Never miss an issue

Get the bi-weekly brief

Teasers of the latest articles, delivered every two weeks.