Method · Priorities
MaxDiff: what really matters most to your customers?
In MaxDiff (best-worst scaling), participants repeatedly see a small selection from a longer list and pick the most and the least important item each time. Because they have to choose, not everything ends up as “very important”. 121surveys turns this into a clear ranking – for everyone together and, if you like, for each person.
Building and testing is free. To publish, you choose a paid plan.
MaxDiff explained
What it's for
Turns a long list – features, messages, ideas – into a clear ranking. Unlike “Rate everything from 1 to 5”, respondents have to choose, so not everything ends up as “very important”.
What respondents see
Several rounds showing a small selection from the list (usually four or five items) and the question: what matters most to you, and what matters least?
How many responses
Each person should see each item at least three times. The usual minimum is 200 respondents, and likewise 200 per group you want to compare.
Reading the results
To the right of the center is what respondents prefer, to the left what they reject; longer bars mean a stronger preference. The ranking is reliable. The gaps show a tendency but don't mean “twice as important”.
Limitations
Shows what's relatively more important – not whether anything is important at all. For that, there's the variant with an importance threshold.
What 121surveys does for MaxDiff
Balanced design
You decide how many items appear per task and how many tasks there are. 121surveys builds the sets so that every item appears roughly equally often.
Counts and Hierarchical Bayes
Counting best and worst choices is available straight away. Hierarchical Bayes also estimates preference scores for each individual person and runs in the background.
With an importance threshold
In the anchored variant, participants also say which items matter to them at all. That way you learn not only the order but also what falls below the threshold.
Segments from the preferences
From the scores per person, 121surveys forms groups with similar priorities – so you can see whether there are several camps rather than a single ranking.
Frequently asked questions
When is MaxDiff better than a rating scale?
When you want to compare many items that all sound somewhat important. On a scale almost everything gets a 4 or 5; with MaxDiff participants have to choose, and the differences become visible.
How many items can the list have?
Ten to thirty is typical. The longer the list, the more tasks each person needs to see every item at least three times – the method advisor in the editor works this out for your design.
What do the scores in the analysis mean?
They show how strongly an item is preferred or rejected relative to the others. The order is reliable, but you should not read statements such as “twice as important” into the gaps.
What is the difference between MaxDiff and conjoint?
MaxDiff ranks individual items on a list. Conjoint tests complete offers made up of several attributes with levels and calculates willingness to pay and choice shares.
What does a MaxDiff study with 121surveys cost?
Setting up the design and running through it with the test link is free. To publish it you need a paid plan; MaxDiff including Hierarchical Bayes is part of every one. What responses cost is on the pricing page.
Put your list in order
Create a MaxDiff question and run through it for free with the test link.
