Feature importance research: a guide to product feature analysis

Feature importance research tells you which product characteristics customers actually value, so you can prioritize development and marketing around what matters instead of what is loudest in the room. Product feature analysis is the umbrella term for this work, and Maximum Difference analysis, or MaxDiff, is the method most commonly used to run it.

Imagine you are developing a new service, like a private members' club lounge or a book club. How do you decide which genre of books to stock, or what drinks to keep behind the bar? This guide covers what a feature is, what feature importance research measures, how MaxDiff analysis works, and how to run a product feature analysis of your own.

A feature is a characteristic of a product, such as its size, color, variation, taste, or speed. A benefit is the advantage that feature delivers to the customer. A phone's advanced camera is a feature; sharper photos are the benefit.

That distinction matters because features without a valued benefit waste development budget. A phone maker gains little from investing heavily in a camera feature if customers do not care about the resulting benefit. Feature importance research, a specific form of market research, exists to test that assumption before you build.

Feature importance research measures how much customers value each characteristic of a product, service, package, or message relative to the others, so you know which ones to prioritize and which to cut. It typically covers four types of characteristics:

Feature importance research is usually carried out using MaxDiff analysis, also known as Best-Worst Scaling. Respondents see a series of small sets of product features and are asked to pick the one they like best and the one they like least from each set. Research suggests a choice set of 3 to 5 features works best. Too many options in one set makes the trade-off harder for respondents and weakens the data.

Analysis of the results converts preferences into a 0 to 100 relative importance scale, or into percentage scores that add up to 100 across all features, so you can see exactly how much more one feature matters than another rather than a vague ranked list.

In an illustrative Feature Prioritization example from SurveyMonkey LaunchPad, a MaxDiff and TURF simulation on ice cream flavor preference found that vanilla and chocolate alone covered 70% of the market, the kind of stack-ranked output an automated analysis produces once responses come in.

Feature importance in market research is a measure of how much customers value one product characteristic relative to others, used to decide which features, packaging concepts, or messages to prioritize or cut.

Feature importance is the specific metric, a relative score for each characteristic. Product feature analysis is the broader process of researching, ranking, and acting on those characteristics, of which feature importance is the core output.

Feature importance research in market research measures how much customers value a product characteristic, gathered through survey methods like MaxDiff. Feature importance in machine learning measures how much a model's input variable contributes to its predictions, calculated through statistical methods like SHAP or permutation importance. The two share a name but answer completely different questions.

No. MaxDiff analysis handles the statistical modeling in the background. Feature Prioritization from SurveyMonkey LaunchPad builds the survey and runs the analysis automatically, so no quantitative research background is required.

Feature importance research only pays off once it changes what you build or promote next. Feature Prioritization from SurveyMonkey LaunchPad runs MaxDiff analysis automatically, turning trade-off responses into a stack-ranked list of what your customers value most, without any manual modeling.

If you'd rather hand the legwork to someone else, SurveyMonkey Market Research Services can design, field, and analyze a custom study for you, from questionnaire design through the final report.