Standard MaxDiff tells you that one feature beats another. It never tells you whether either one actually matters to respondents. Anchored MaxDiff scaling closes that gap by adding a single extra question to a standard MaxDiff analysis, turning a relative ranking into a scale a team can act on.
This distinction matters most in feature prioritization, product design, and message testing. Knowing that respondents prefer Feature A over Feature B is not the same as knowing whether either one clears a real bar of importance.
In this article, we’ll cover what anchored MaxDiff scaling is, how the Binary Direct and Dual Response models work, and how to set it up inside a live survey.
What is anchored MaxDiff scaling?
Anchored MaxDiff scaling is a research technique that upgrades a standard MaxDiff question from a relative measurement into an absolute one. It does this by adding a single follow-up question that identifies which attributes clear a real importance threshold.
MaxDiff, short for Maximum Difference scaling, is a survey method where respondents pick the best and worst option from a set of items instead of rating each one on its own. It is also known as Best-Worst scaling.
Traditional MaxDiff has one built-in limit. It ranks items against each other but never confirms whether any of them matter in an absolute sense.
- Feature A might score twice as high as Feature B on a standard MaxDiff report.
- That comparison alone says nothing about whether either feature is actually important to respondents.
- Neither feature might clear a meaningful threshold at all.
Anchored MaxDiff adds one extra question, often called the anchor or utility threshold, that asks respondents to draw a line between attributes that genuinely matter and those that do not.
Under the hood, standard MaxDiff scores come from a multinomial logit model, a statistical method that estimates the probability an item gets chosen as best or worst out of a set. That model produces clean relative scores, but it has no built-in zero point. Anchoring supplies that zero point by tying the scale to a real respondent judgment instead of an arbitrary midpoint.
Anchored MaxDiff vs standard MaxDiff vs conjoint analysis
These three methods get confused often because they all deal with preference and priority. Each one answers a different question, so picking the wrong one wastes a survey.
Standard MaxDiff is the right call when you only need a ranked list. Anchored MaxDiff is the right call when you need to know which items are worth acting on. Conjoint analysis is the right call when respondents need to trade off combined attributes, like price against features, rather than pick single best and worst items.
| Method | What it measures | Best used when |
|---|---|---|
| Standard MaxDiff | Relative preference only | You need to rank many items quickly |
| Anchored MaxDiff | Relative and absolute importance | You need to know which items truly matter |
| Conjoint analysis | Trade-offs across bundled attributes | Pricing or bundled feature decisions |
A common mix-up is treating anchored MaxDiff as a substitute for a rating scale. It is not. A rating scale asks respondents to score each item on its own, which invites bias when some people rate everything high and others rate everything low. Anchored MaxDiff keeps the comparative task intact and only adds one binary judgment, which keeps that bias out of the picture.
How does anchored MaxDiff scaling work?
Anchoring supports two models. Both insert as a simple extra step into the existing MaxDiff question type, so the added respondent effort is small.
Binary Direct model
The Binary Direct model is the simpler and more common of the two. Respondents see every option at once and mark which ones are must-haves.
- One clean action from the respondent, with no extra framing required
- A straightforward line between must-have and nice-to-have
- Fast to build on the research side
That single action draws a clean boundary between what genuinely matters to a respondent and everything else.
Dual Response model
The Dual Response model asks respondents to identify some, all, or none of the options using a standard multiple-choice question. It gives researchers more flexibility in how the anchor question is worded, at the cost of a slightly longer respondent task.
When should you use anchored MaxDiff scaling?
Anchoring earns its place any time a simple ranked list would hide the real answer. A smartphone brand deciding which features to build next needs more than a rank order. So does an apparel brand weighing which trends are worth chasing.
A credit card research team is a useful example. A researcher shows respondents six card brands using a standard MaxDiff question. Most respondents only ever select two of the six as their most and least preferred, leaving the other four unexplained. Adding an anchor question fixes that. If respondents mark Visa, Mastercard, and American Express as must-haves, the team learns that even though Visa ranked highest, the other two still clear the importance bar rather than sitting in an ambiguous middle.
A SaaS product team runs into the same problem during roadmap planning. Ranking twenty possible features against each other tells the team an order, but not how many of those features customers would actually miss if they were cut. An anchor question answers that directly.
The same logic applies in concept testing, where respondents rank which design concepts they like most and least. An anchor question then identifies every concept they would genuinely be satisfied with, not just the single top pick.
A healthcare provider evaluating patient service improvements faces a similar decision. Ranking twelve possible changes tells leadership an order of preference, but a hospital cannot act on all twelve at once. Anchoring narrows the list to changes that clear a real importance bar, which is a more useful input for understanding customer needs than a rank order alone.
How to set up anchored MaxDiff scaling in a survey
Anchoring is available as a setting on the MaxDiff question type inside most survey platforms, including QuestionPro, but it is off by default. Turning it on adds the anchor question automatically to the standard MaxDiff flow.
- Open the MaxDiff question in your survey builder and locate the anchoring setting.
- Choose between the Binary Direct or Dual Response model based on the flexibility your study needs.
- Customize the anchor question wording to match the study, such as must-haves, would definitely buy, or another phrase that fits the context.
- Confirm the anchor question runs as part of the same flow as the standard MaxDiff task, so respondents see it right after the ranking exercise.
- Launch the survey and review the raw output, which includes the anchor results alongside the standard MaxDiff scores in the same report.
Teams running this kind of feature-prioritization study inside Market Research Software can build the anchor question directly into the same MaxDiff flow without exporting data to a separate tool.
How to interpret anchored MaxDiff scores
A standalone MaxDiff score means little without a reference point. Saying one feature scored a 10 and another a 5 only tells you the first is twice as preferred. It does not tell you whether either one is worth building.
The anchor question gives that missing context. Items that fall above the anchor threshold are treated as important in an absolute sense. Items below it are not, regardless of how they rank against each other.
| Feature | Relative MaxDiff score | Above or below anchor |
|---|---|---|
| Fast checkout | 42 | Above |
| Live chat support | 28 | Above |
| Loyalty points | 19 | Below |
| Custom themes | 11 | Below |
A peer-reviewed study in Marketing Letters found that anchored MaxDiff approaches help address a core limitation of traditional MaxDiff by bringing results onto an absolute scale that supports real purchase predictions, not just relative rankings. Read the full study.
This table format is a simplified illustration. Actual output depends on which anchoring model was used and how the underlying utility scores were estimated, but the core read stays the same: a score above the anchor means the feature clears the bar, and a score below it does not.
Common mistakes to avoid with anchored MaxDiff surveys
Anchoring adds real value, but it is easy to misuse. A few mistakes show up often enough to call out directly.
- Adding an anchor question to every MaxDiff study out of habit, even when a simple rank order already answers the research question
- Wording the anchor question vaguely, which lets respondents interpret must-have differently from one another and muddies the threshold
- Treating the anchor score as a rating scale rather than a threshold, which misreads what the data is actually showing
- Skipping a pilot test of the anchor wording before fielding the full study, which risks discovering a confusing phrase after data collection is already underway
- Ignoring sample size when anchoring, since a small sample can make the threshold unstable across subgroups and lead to shaky conclusions
Most of these mistakes come down to treating anchoring as an automatic upgrade rather than a deliberate choice tied to a specific research question. A quick pilot with a handful of respondents usually catches wording problems before they affect the full dataset.
Getting the fuller picture from a MaxDiff study
A standard MaxDiff survey works well when a high-level read on preferences is all a team needs. Anchored MaxDiff earns its place when the real question is which items clear a genuine bar of importance, not just which ones rank higher. Ranking everything relative to everything else is not always enough to make a confident call.
Frequently Asked Questions (FAQs)
No. The underlying MaxDiff utility scores are calculated the same way. The anchor question runs alongside the standard exercise and adds a separate reference point, so the original relative rankings stay intact and comparable.
Yes. Anchoring works with small item sets and large ones alike. Smaller lists just need fewer choice sets per respondent, since MaxDiff already adjusts the number of comparisons to the number of items being tested.
It typically adds well under a minute. The anchor question is a single multiple-choice step placed after the standard MaxDiff task, so it does not meaningfully add to overall survey length or respondent fatigue.
Yes, especially for software and enterprise feature prioritization. US B2B teams often need to justify a roadmap decision to stakeholders, and an absolute must-have threshold gives them a clearer case than a relative ranking alone.
There is no single fixed number, but most researchers treat 150 to 300 completed responses per key subgroup as a reasonable starting point. Smaller samples can still work, though the anchor threshold becomes less stable across subgroup comparisons.
