MaxDiff & TURF analysis
Rank what matters with real separation, then build the smallest set that appeals to the most people.
Which benefit matters most to you, and which matters least?
| Most | Benefit | Least |
|---|---|---|
| No foreign transaction fees | ||
| Instant fraud alerts | ||
| Metal card design | ||
| Airport lounge access |
Sample question, illustrative: from four benefits, the respondent has chosen “Instant fraud alerts” as most important and “Metal card design” as least important.
Questions this answers
- Out of 20 to 60 possible features, which matter most, and by how much?
- Which benefits or claims should lead our positioning?
- Which combination of features, perks, or messages reaches the widest audience?
- Do priorities differ by segment?
- Where can we cut without losing customers?
Use it when
You have a long list to narrow down, need to choose positioning or claims, or are deciding what goes into a lineup or bundle.
Look elsewhere when
You need prices or full product trade-offs. Take the top items into a conjoint study.
What we prioritize with it
- Features
- What to build first, and what to cut from the roadmap.
- Benefits and perks
- Which rewards and services earn their cost.
- Positioning and claims
- Which message should lead, overall and by segment.
- Lineups and portfolios
- The smallest range that reaches the most customers.
MaxDiff
Respondents pick the most and least important item from small sets. Across many sets, that forced choice produces clear, ratio-scaled priorities, without the rating-scale problem where everything looks important.
- Best–worst scaling
- Small, balanced sets so every item is seen fairly.
- Anchored MaxDiff
- Adds an absolute threshold, so you know which items truly matter, not just their rank.
- Sparse and express designs
- Handle lists of 40 or more items without overloading respondents.
- Hierarchical Bayes estimation
- Individual-level scores for segment cuts and TURF.
Text summary of the chart: indexed against the top item, instant fraud alerts scores 100, no foreign transaction fees 82, cash back on groceries 74, travel insurance 55, free credit score 48, airport lounge access 36, concierge service 22, and metal card design 12. The anchored threshold sits at 50, so the top three items clear it comfortably. Figures are invented for illustration.
TURF
Total Unduplicated Reach and Frequency finds the combination of items that appeals to the largest share of customers, and shows where adding more stops paying off. It runs on the same MaxDiff data, so no extra survey is needed.
- Optimal portfolios
- The best combination of 2, 3, 4, or more items.
- Diminishing returns
- How much each additional item adds to reach.
- Constraints
- Force in must-haves or exclude items you can't offer.
- Segment reach
- Which combination wins within each target segment.
Best 4: instant fraud alerts, no foreign transaction fees, cash back on groceries, free credit score. Adding a fifth reaches only 4 more points.
Text summary of the chart: the best combination of one benefit reaches 48% of customers, two reach 66%, three reach 77%, four reach 84%, five reach 88%, six reach 90%, and seven reach 91%. Returns flatten after four. Figures are invented for illustration.
How we run it
Typically 2 to 3 weeks from item list to recommendations.
One team, end to end. The same Sibilance researchers build your item list, field the study, run the MaxDiff and TURF, and make the recommendation, and can take the leaders into a conjoint without re-briefing anyone.
-
Build the item listWritten with you, in the language customers use.
-
Design balanced setsSo every item is shown fairly and often enough.
-
Field and estimateIndividual-level scores, with quality controls.
-
Rescale the scoresAn intuitive 0 to 100 share of preference.
-
Run TURFFind the combinations that maximize reach.
-
RecommendPriorities for the total market and each segment.
Case example: focusing an app on what counts
Client: MassMutual (Society of Grownups)
Challenge
MassMutual was building a mobile financial education app for younger customers, with more than 65 possible features on the table.
Approach
We prioritized the features against target segments, focusing the first release, then tested prototypes against market leaders.
Result
Prototypes outperformed leading apps in quantitative testing, and pre-launch landing page tests beat benchmarks.
Questions we hear
Why not just use rating scales?
Ratings tend to make everything look important. MaxDiff forces choices, which gives much clearer separation.
How many items can we test?
Commonly 15 to 40. Larger lists are possible with sparse designs.
Does TURF need a separate survey?
No. We run it from the same MaxDiff data.
Can MaxDiff test messages?
Yes. It's one of the best ways to prioritize claims and positioning statements.
What sample size do we need?
Usually 300 to 800 respondents, more if you need to read several segments.
Talk to us about MaxDiff & TURF
Send us your list. We'll suggest a design, a sample, and a timeline.