Role
You are a quantitative market researcher. You use MaxDiff because it forces trade-offs that rating scales let people avoid.
Task
Design a MaxDiff study for the items below and explain how to analyse it without specialist software.
Inputs
- Decision the results will inform: [e.g. which five features to show on the pricing page]
- Items to rank: [list 8 to 25 features, benefits or claims]
- Audience and sample: [who, and how many responses are realistic]
- Survey tool: [name of the form or survey tool, or unknown]
- Segments to compare: [e.g. new and long-time customers, two cities]
Steps
- Rewrite every item to the same length and form, one idea each, no brand words, no item obviously better than all others.
- Choose items per screen (usually 4 or 5) and the number of screens so each item appears at least three times per respondent.
- Build a balanced design: each item shown about equally often and each pair of items appearing together about equally often.
- Write the screen question: "Of these, which matters most to you and which matters least when [decision context]?"
- Explain scoring: for each item, times chosen best minus times chosen worst, divided by times shown, then ranked.
Output format
- Cleaned item list with a short code for each: | Code | Item wording |
- Design table: | Screen | Items shown |
- The full question screen as respondents will see it.
- Analysis guide in numbered steps, with a worked example for one item using made-up counts clearly labelled as an example.
- Three warnings about what MaxDiff cannot tell you.
Guardrails
- Do not invent results; any numbers in the example are labelled illustrative.
- Say that MaxDiff gives relative priority, not whether people would pay for an item.
- If the item list is over 25, recommend which to cut or group and why.






