Segmentation analysis is the ultimate simplification for survey data, as we are grouping individuals using common patterns we are interested in. We can make predictions about how an individual will respond to questions simply by knowing what segment they’re in.
Successful segmentations need strong planning, good communication and the use of best-in-class statistical models. We are power users of Latent Class Analysis, considered one of the most robust and flexible methods for identifying distinct groups with different patterns and deploy simpler methods such as CHAID and hierarchical clustering where appropriate.
We run segmentation analysis across a wide range of sectors, audience types and content and know that segmentation is a soft as well as a hard science. Our decades of research experience put us in a great position to advise our clients on survey inputs, ensuring they are optimised and well balanced.
We help our clients visualise differences between alternative segment models using excel mini-profiles. When a model has been agreed we also regularly produce golden question algorithms for clients to help them predict segment membership for new cases (outside of the survey).
Examples of different types of segmentation models:
- Highly targeted marketing and product development strategies: to unlock a better understanding of potential customers, provide critical insights and maximise the appeal to key groups in the market.
- Stakeholder segmentation: groups business decision-makers into segmented clusters based on their background, interests, motivations and attitudes.
- Medical treatment and medical device segmentation: can illuminate preferences and attitudes towards new types of treatment and devices among medical healthcare professionals.
- Thought-piece segmentation: to understand the motivations, attitudes, day-to-day challenges, needs and expectations of groups of interest to society, such as entrepreneurs, doctors, students and apprentices for public-domain reports.
- Political segmentation: enables profiling of distinct voter groups based on demographic characteristics, personality type, voting preferences and other data.