1. 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?
2. 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
But the algorithm is only one part of the solution. We manually review and refine solutions—dozens
per study—using experience and your business needs as a guide. For example, we can retain a
high-value segment while adjusting others that are less useful. The result is statistically valid
and strategically aligned.
3. 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:
- Input selection: Behavioral, psychographic, demographic, and geographic
variables—focused on real differentiation
- Survey design: Clean measurement using MaxDiff, semantic differentials, and
“check all that apply” (no biased Likert scales)
- Model exploration: We test thousands of models and retain only those with
strong statistical structure
- Solution evaluation: Detailed Excel workbooks with segment summaries, F-stats,
classifier accuracy, and identity matrices
- Classifier delivery: Excel-based typing tool with 80%+ accuracy in under 5
minutes
- Optional integration: Mapping segments to your CRM for data enrichment and
media targeting
- 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.