> ## Documentation Index > Fetch the complete documentation index at: https://docs.exoid.ai/llms.txt > Use this file to discover all available pages before exploring further. # MaxDiff > Measure relative preference across a set of statements, features or concepts by asking respondents, set after set, which one convinces them most and which convinces them least. Exoid supports MaxDiff as a question type across the entire survey flow: from configuring it in the Builder, to answering it in the Player, to analyzing it in Results and in the data export. ## What MaxDiff is MaxDiff (Maximum Difference Scaling, also called Best-Worst Scaling) is a method for measuring relative preference across a list of statements, features, claims or concepts, for cases where a direct scale doesn't discriminate between them well enough. Asking people to rate ten claims one by one on a five-point scale often produces high, flattened ratings on almost everything: the respondent isn't forced to choose. MaxDiff solves this by showing the statements in small groups: in each group, the respondent only has to say which one convinces them most and which convinces them least. Repeated across several groups, the implicit pairwise comparison across every statement produces a far sharper preference ranking than any direct scale. ## Add a MaxDiff question Add a **MaxDiff** block from the question type menu in the Builder, alongside Ranking, Multiple choice and the other option-based blocks — see [Question types](/en/builder/question-types). ## Configure the question Besides the fields shared by every block, the MaxDiff block's right-hand panel carries settings specific to this exercise. The question text the respondent reads above every set, for example "Which of these statements convinces you most to choose a laundry detergent?". The text that explains the task, for example "For each group, select the statement that convinces you most and the one that convinces you least." An image, video or audio attached to the question, same as on other blocks — see [Block options](/en/builder/block-options). The same timer shared with other blocks, useful when you want to give respondents time to read the statements before they can answer. The text of the button or label the respondent uses to mark the statement they prefer most, for example "Convinces me most". The matching text for the statement they prefer least, for example "Convinces me least". What you want to call the individual items being compared in the interface — for example "Comparison". What you want to call each individual set shown to the respondent — for example "Statements". ### Statements to compare This is where you enter the full list of statements, features, claims or concepts to test, using **Add Bulk Options** to paste them all at once or one at a time. Add at least two statements. For a meaningful comparison between Best and Worst, the interface recommends at least three. From the configured list, Exoid automatically generates the sets — the "tasks" — that make up the exercise: each set shows a subset of the statements, and over the course of the question the respondent completes more than one, so every statement gets seen and compared several times. ## What respondents see Respondents see one set at a time, with a small group of statements (for example four). For each set, they indicate which statement convinces them most and which convinces them least, then move to the next set. The exercise repeats across every generated set, always with a small number of statements visible at once. ## MaxDiff results ### Summary In [Summary](/en/analyze/summary), the MaxDiff question shows a table with one row per statement: | Column | What it means | | - | - | | Statement | The text of the statement being tested. | | Net score | The difference between how often it was picked as best and how often as worst, as a percentage of the number of times it was shown. | | Shown | How many times the statement appeared in a set, across every collected response. | | Best / Best % | How many times it was picked as best, in absolute terms and as a percentage of the times it was shown. | | Worst / Worst % | How many times it was picked as worst, in absolute terms and as a percentage of the times it was shown. | The net score is the fastest way to rank statements from most to least convincing: a high, positive value marks a statement that was often picked as best and rarely as worst. ### Responses In a single response's detail view, in [Responses](/en/analyze/responses), the MaxDiff question shows two levels of detail: * The net score per statement for that single respondent, calculated the same way as in Summary but limited to the sets they were actually shown. * The list of every task they completed, with the statements presented in that set, the one picked as best highlighted in green and the one picked as worst highlighted in red. This is the view to use when you want to see how one respondent moved across sets, not just their aggregated result. ## Exporting MaxDiff data The Excel export from [Responses](/en/analyze/responses) includes MaxDiff data at two levels: * In the main sheet, for every configured statement you get three columns per respondent — how many times it was shown to them, how many times they picked it as best and how many as worst — plus a text summary column detailing every task they completed (the statements shown, and which were picked as best and worst). * A sheet dedicated to the question, titled with its question code (for example "MaxDiff D4"), with one row per task per respondent: the statements shown in that set and which ones were picked as best and worst. This second sheet is the per-task detail that enables more granular analysis directly from the exported workbook — see [Exporting data](/en/analyze/export). ## Next steps Every available block in the Builder and when to use each one. How to take MaxDiff data out of Exoid with per-task detail. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.