Smarter Segmentation for Real-World Marketing Impact

At SegmentationPro, we believe segmentation should be more than just a statistical exercise—it should be a strategic engine that drives real decisions and measurable results. Whether you’re refining your brand strategy, targeting high-value customers, or launching new products, our solution helps you uncover the customer groups that truly matter—and act on them with confidence

Segmentation Philosophy — Actionable,
Differentiated, and Grounded in Reality

 At the core of our segmentation philosophy is a simple idea: a true segmentation scheme must uncover groups of customers or prospects that would respond differently to changes in the marketing mix, based on their distinct needs, behaviors, or values. These segments must be more than just statistical clusters—they must be strategically useful.

To ensure this, we adhere to three essential criteria:

Be real — Can we recognize these groups in the real world? Have you encountered them before in your business?

Be well-differentiated — Do the segments lead to different marketing approaches? Would you talk to them differently?

 Be identifiable and actionable — Can the marketing team reliably find and reach these groups in real life, through media targeting and tailored communications?

We design segmentation to be a long-term investment. The output isn’t just a report—it’s a decision-making tool. We begin every project by aligning on your business objectives: what decisions will the segmentation guide?
What does success look like? What kind of segments would actually move the needle for your business?

Analytical Approach — Proprietary Bayesian Mixed Modeling

Our clustering approach is powered by a Bayesian mixed modeling algorithm developed in-house. This allows us to explore thousands of potential solutions in every study—systematically evaluating options across different input combinations, distance measures, and model configurations.

What makes our algorithm different is that it:

Identifies both within-cluster similarity
and between-cluster differences

 Handles different input types and
scales, including missing data

Incorporates weights and
accounts for correlations
between variables

 Statistically determines the optimal
number of segments

 Produces robust outputs for
accurate classifier models

Process & Tools — From Raw Data to Strategic Activation

Our segmentation process is comprehensive, hands-on, and designed to deliver actionable insights at every step. Here’s what it includes:

1

Input selection: Behavioral, psychographic, demographic, and geographic variables—focused on real differentiation

2

Survey design: Clean measurement using MaxDiff, semantic differentials, and “check all that apply” (no biased Likert scales)

3

Model exploration: We test thousands of models and retain only those with strong statistical structure

4

Solution evaluation: Detailed Excel workbooks with segment summaries, F-stats, classifier accuracy, and identity matrices

5

Classifier delivery: Excel-based typing tool with 80%+ accuracy in under 5 minutes

6

Optional integration: Mapping segments to your CRM for data enrichment and media targeting

7

Optional deliverables: From persona summaries in Word to PowerPoint decks for internal storytelling

 Ultimately, our goal is to make segmentation replicable, interpretable, and actionable—not just as a one-off exercise, but as a living tool your team can use over time

Let SegmentationPro Power Your Next Marketing Move

With SegmentationPro, you don’t just get insights—you get a framework for action. From strategy toexecution, we help you understand your customers, prioritize your efforts, and personalize your marketing at scale. Let’s turn segmentation into your most powerful competitive advantage.

Case Title

Background

L’étiquette d’une bouteille de vin est essentielle, capable à elle seule d’augmenter les

ventes ou de soutenir une hausse de prix sans impacter la demande.

Problem

Un graphiste a créé 14 étiquettes potentielles, mais laquelle aura le plus desuccès?

Solution

C’est le consommateur qu’il faut questionner! L’analyse MaxDiff permet de déterminer
quantitativement l’ordre de préférence des étiquettes parmi un échantillon de 200-400
consommateurs en comparant les différentes options. Cette méthode résout

efficacement l’incertitude du vigneron

Results

Example of result, some explaination

Yariv Levy, PhD
VP of Technology

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum

Moshe Vaizman
VP of Operations and 

Business Development

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum

Iris Levy
VP of Marketing

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum

Cindy Ford, PhD
VP of Customer Success

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum

Marc Uldry, PhD
Co-Founder / VP of Product

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum

Valerie Severin, PhD
Co-Founder / VP of Finance

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum

MaxDiff, or Best-Worst Scaling (BWS), is a remarkably insightful and productive measurement method used widely by researchers and practitioners across social science, business, and other disciplines. It is often preferred over traditional rating scales because it is effective at capturing preferences with known measurement properties.

Here is a comprehensive explanation of Best-Worst Scaling (BWS) and its origins as MaxDiff:

  1. Definition and Core Mechanism

Best-Worst Scaling (BWS) is defined as an extension of the method of paired comparison to multiple choices. The fundamental task requires participants to evaluate a collection, or “set,” containing three or more items or options. From this set, the participant must choose:

  • The best (or most attractive/most preferred) option.
  • The worst (or least attractive/least preferred) option.

The terms “best” and “worst” are considered metaphors representing the extremes of an underlying, latent, subjective continuum of interest.

By collecting both extreme choices from a set, BWS obtains extra choice information. Crucially, this method gathers valuable information about both the attractive options (via the “best” choices) and the less attractive options (via the “worst” choices), providing richer insights into the respondent’s utility function.

  1. Nomenclature and History (MaxDiff vs. BWS)

The terms MaxDiff and BWS refer to the same technique, though BWS is now the generally accepted name, particularly within academic circles.

  • Origin as MaxDiff: The approach was pioneered by Jordan J. Louviere, who developed it in 1987. He initially called it “maximum difference scaling” (or “MaxDiff”).
  • Hypothesized Process: This original name reflected Louviere’s hypothesis that the underlying decision process involved choosing the pair of options that exhibited the largest subjective difference on the latent scale of interest. The two chosen options (best and worst) would thus be maximally different.
  • Shift to BWS: The name was subsequently changed to Best-Worst Scaling because subsequent academic research demonstrated that individuals generally do not use a maximum difference choice process to perform the task. BWS is considered a more inclusive general term, as MaxDiff only represents one of several possible psychological processes a person might use to provide best-worst data.
  1. Advantages over Rating Scales

A significant benefit of using BWS is its ability to avoid many problems inherent to traditional rating scales.

  • Forced Discrimination: BWS forces respondents to discriminate between items by selecting only one best and one worst option, thereby reducing response biases (such as acquiescence bias or consistently using only mid-points or end-points of a scale).
  • Measurement Properties: BWS offers a theoretical framework to measure latent, subjective quantities. The derived measurement values possess known properties—specifically, they often produce ratio-level scales (if specific conditions regarding choice probabilities are met).
  • Simplicity and Quality: The task itself is simple and undemanding for respondents, which typically improves data quality.
  1. Theoretical and Analytical Foundation

The conceptual framework underpinning BWS is random utility theory (RUT). RUT assumes that when people choose repeatedly, their observed choice frequencies indicate how much they value the items under consideration.

Analytically, BWS often involves straightforward calculations, making it accessible even for practitioners with minimal prior knowledge. A simple way to summarize the data is by calculating best-minus-worst scores for each option, which is the count of times an item was chosen as “best” minus the count of times it was chosen as “worst”.

These best-minus-worst scores empirically correlate highly with estimates from more sophisticated models, often serving as a sufficient statistic for conditional logit regression models if the MaxDiff model holds.

  1. Categorization of Applications

BWS applications are typically divided into three main classes, differing based on the complexity of the items being evaluated:

Case

Name

Description of Items

Choice Task Focus

Case 1

The Object Case

Simple items, objects, statements, or people.

Measuring a list of items on a single underlying subjective scale (e.g., measuring priority, importance, or agreement).

Case 2

The Profile Case

Attribute levels described/displayed as a single profile (a combination of attribute levels).

Choosing the best and worst attribute levelswithin a single profile.

Case 3

The Multi-Profile Case

Sets of three or more profiles (each profile being a complex combination of attributes and levels, akin to a traditional DCE).

Choosing the best and worst whole profiles (options) among multiple choices.

In summary, BWS is an invaluable choice-based measurement method that leverages the finding that individuals are highly reliable and accurate when identifying extreme options. It moves beyond the MaxDiff hypothesis to provide a robust framework for measuring subjective preferences.

Reference: Louviere, J. J., Flynn, T. N. & Marley, A. A. J. Best–Worst Scaling:Theory, Methods, and Applications (Cambridge Univ. Press, 2015).

 

 

Pierre Uldry, PhD
CEO / Founder

Pierre, originally from Switzerland, began his journey as an economist before transitioning into marketing, statistics, and becoming a self-taught programmer.
He moved to Australia to pursue his PhD at the University of Sydney, studying under the guidance of Professor Jordan Louviere, the mastermind behind BWS, nowadays MaxDiff. In a significant contribution, Pierre introduced the dual response to this technology. With a solid foundation in mathematics, he’s dedicated over two decades to discrete choice modeling, consistently thinking outside the box.
Passionate about movies, running, and his former stint as a Swiss mogul skier, he’s deeply connected to Australia, where he
continues to reside. His expertise in Bayesian Methods has seen him craft processes for diverse industries and tailor models for numerous studies

How does it work?

1. Reach out

You: Fill out our form and tell us what you’d like to test.

We: Review your request and confirm feasibility.

2.⁠ ⁠Free Consultation

You: Join a 30-minute call to discuss your study.

We: Design and launch your study.

3.⁠ ⁠Approval

You: Approve the quote and study plan.

We: We design and launch the study.

4. Results

You: Sit back and relax — your insights are on the way.

We: Collect the data, analyze it, and share your report.