Your product team has a list of fifteen features they want to build. Your marketing team insists all twelve brand messages are equally compelling. Your leadership team believes every strategic initiative deserves equal investment. The truth is, not everything matters equally to the people who actually buy from you -- and guessing which items rise to the top is a recipe for wasted resources.
MaxDiff analysis -- sometimes called maximum difference scaling or best-worst scaling -- is the most reliable way to force-rank preference and importance across a long list of items. At Meraki Research, we use MaxDiff to cut through internal opinion and give you a clean, statistically defensible hierarchy of what your customers actually care about. Whether you are prioritizing features, refining messaging, or narrowing a product roadmap, MaxDiff delivers the discrimination power that simple rating scales simply cannot.
We have designed and fielded hundreds of MaxDiff studies across consumer packaged goods, healthcare, financial services, and technology. Our researchers pair rigorous experimental design with practical business framing so the data you receive is not just analytically sound -- it is immediately actionable. From an intro call to final deliverables, our process is built around one principle: the analysis only matters if it changes a decision.
How MaxDiff Works: Our Structured Analysis Approach
MaxDiff was originally developed by Jordan Louviere in the early 1990s as an alternative to traditional rating and ranking exercises. The core insight is elegant: when respondents are shown a subset of items and asked to pick the best and worst in each set, the resulting choices produce far greater discrimination among items than conventional Likert scales ever could.
Our team follows a proven five-phase methodology we call Precision Priority Mapping to ensure every MaxDiff study we run delivers maximum strategic value.
Phase 1: Item Development and Strategic Framing
Before we write a single survey question, we work with your team to define the decision that will be informed by the results. Are you choosing which features to fund in the next sprint? Selecting which value propositions to lead with in a campaign? Narrowing a portfolio of concepts? The decision context shapes everything from item wording to sample design.
We typically refine the item list to between 10 and 25 items. Fewer than 10 and you rarely need MaxDiff at all -- a simple rank order will do. More than 25 and respondent fatigue begins to erode data quality. Our researchers pressure-test each item for clarity, mutual exclusivity, and comparable abstraction level, because poorly written items produce misleading hierarchies.
Phase 2: Experimental Design
This is where the analytical rigor lives. We construct a balanced incomplete block design that determines how many sets each respondent sees, how many items appear per set, and how items rotate across the full sample. The goal is ensuring every item appears with every other item a roughly equal number of times, which is what allows us to compute reliable utility scores.
For example, a study with 16 items might show each respondent 12 sets of 4 items each. Our team uses advanced design algorithms -- not off-the-shelf defaults -- to optimise statistical efficiency for your specific item count and sample size. The design phase is invisible to respondents but critical to the quality of downstream analysis; a poorly balanced design forces researchers to make statistical compromises that reduce confidence in the final hierarchy.
Phase 3: Fieldwork and Quality Controls
We program the MaxDiff exercise within a broader survey that includes context-setting questions, screening criteria, and behavioral anchors. Respondents see each set and select the best (most important, most preferred, most appealing) and worst option. The forced-choice format means there is no satisficing -- every choice carries information.
Our quality protocols flag respondents who complete the exercise too quickly, show random response patterns, or contradict their own stated preferences. We remove these records before analysis so your results reflect genuine preference, not noise.
Phase 4: Hierarchical Bayes Estimation
We estimate individual-level utility scores using hierarchical Bayes (HB) modelling, which produces a preference score for every item for every respondent. This is critical because it allows us to segment your audience by priority structure, not just look at averages. Two customer segments might rank the same item at the top on average but for entirely different reasons -- HB-level data reveals those differences.
The output is a ratio-scaled score for each item, meaning you can say "Item A is twice as important as Item B" with statistical confidence. That level of discrimination is what makes MaxDiff superior to simple rating scales where everything clusters around 7 or 8 out of 10.
Phase 5: Strategic Translation and Activation
Numbers without narrative are just a spreadsheet. We translate the utility scores into a prioritized action framework tied to your original decision. We show you which items are table stakes (high importance, low differentiation), which are true differentiators, and which are irrelevant distractions. We often pair MaxDiff results with TURF analysis (Total Unduplicated Reach and Frequency) to answer follow-up questions like "What is the smallest set of features that appeals to the largest share of our target market?"
Our researcher team also maps MaxDiff utility scores against competitive positioning data when available. If you know which features or benefits your competitors emphasize, layering that intelligence onto your priority hierarchy reveals gaps -- high-importance items that no one in your category is owning. That is where differentiation lives.
When MaxDiff Analysis Is the Right Tool
MaxDiff is the right method when you need to prioritize a list of roughly 10 to 25 items and want more discrimination than ratings or rankings provide. It works exceptionally well for:
- Feature prioritization: You have a backlog of product enhancements and need data-driven prioritization that the entire organisation trusts.
- Message testing: You want to know which brand claims resonate most, not whether each claim is "somewhat appealing."
- Benefit hierarchies: You are building a positioning strategy and need to know which benefits actually drive preference.
- Portfolio optimisation: Paired with TURF, MaxDiff identifies the minimum set of items that covers the maximum audience.
MaxDiff is not the right method when you need to understand how items interact with each other or when price is a key variable. For those questions, conjoint analysis is the better fit. MaxDiff tells you what matters most in isolation; conjoint analysis tells you how people trade off multiple attributes simultaneously. If your primary question is about optimal pricing, consider our Pricing Research practice, which includes Van Westendorp, Gabor-Granger, and Discrete Choice Modeling approaches.
What We Deliver
Every MaxDiff engagement produces a clear set of deliverables designed to move decisions forward, not sit on a shelf.
- Priority Scorecard: A ratio-scaled ranking of all items with confidence intervals, presented in a visual format your leadership team can absorb in minutes.
- Segment-Level Priorities: If your audience contains distinct customer segments (and it almost always does), we show how priority structures differ across groups. This pairs naturally with our Market Segmentation Research capabilities.
- TURF Optimization (when applicable): The optimal subset of items -- features, messages, products -- that maximizes unduplicated reach across your audience.
- Interactive Data Dashboard: A filterable tool your team can use to explore results by segment, demographic, or behavioral group without waiting for a researcher to re-run the numbers.
- Executive Presentation: A 30-to-45-minute read-out with clear recommendations tied to your business context.
Timelines vary by complexity, but a typical MaxDiff study moves from kickoff to final deliverables in four to six weeks. Studies requiring custom panel recruitment or international fielding may add one to two weeks. For teams operating on tighter timelines, we offer an accelerated design track that compresses the intro alignment and design phases into a single week without sacrificing analytical rigor.
Industry Applications
Consumer Packaged Goods: Prioritizing product claims for packaging and advertising across dozens of potential messages.
Technology and SaaS: Determining which features to build next based on actual user preference rather than the loudest internal voice.
Healthcare and Pharma: Ranking patient needs, physician messaging hierarchies, and treatment attributes in regulated environments where every word must be defensible.
Financial Services: Identifying which product benefits drive consideration among different customer segments -- retirees versus young professionals, for example.
Retail and E-Commerce: Establishing which store or site features matter most to buyers when choosing between competitors.
B2B and Professional Services: Prioritizing product capabilities, service attributes, and partnership criteria among business buyers whose purchase decisions involve multiple stakeholders with competing priorities.