Developing an Open, National-Level Geodemographic Classification of Consumer Behaviour

The Consumer Insight Dilemma: Why We Needed an Open Classification

Consumer behaviour in the UK has undergone seismic shifts over the past decade. The acceleration of e-commerce, hybrid working models, rising environmental consciousness, and the persistent pressures of the cost-of-living crisis have reshaped how, when, and where people spend their money. Yet for urban planners, local authorities, and academic researchers seeking to understand these dynamics at the local level, existing analytical tools present major shortcomings:

Our paper fills this critical void by introducing the first open-access, national-level, small-area geodemographic classification of consumer behaviour covering England and Wales at the Lower Layer Super Output Area (LSOA) scale.

The Conceptual Framework: 8 Domains of Contemporary Consumption

Rather than treating demographics as a simple proxy for consumption, our framework posits that contemporary retail behaviour is co-determined by social, spatial, economic, and digital realities. We established eight analytical domains:

Methodology: Spatial Microsimulation & Clustering

A primary methodological challenge in consumer analytics is that comprehensive behavioral surveys (such as the PDV Consumer Lifestyle Survey) lack sufficient sample density to be mapped directly to small geographic areas like LSOAs without severe sampling bias.

To overcome this, we employed spatial microsimulation. By constraining individual survey records against official 2021 UK Census small-area margins (age profiles, household types, tenure, economic activity) alongside physical accessibility metrics, we synthesized small-area populations that reflect realistic consumer profiles for every neighborhood in England and Wales.

Following variable selection and rigorous robustness checks across multiple algorithms (evaluating K-Means, Partitioning Around Medoids [PAM], Fuzzy Geographically Weighted Clustering [FGWC], and GMMs), an optimal clustering architecture emerged.

The Classification: 4 Primary Clusters & 9 Granular Subclusters

The classification reveals four distinct consumer archetypes, each further decomposed into nine subclusters capturing the rich nuances of modern British consumption:

Applications for Policy, High Streets, and Digital Equity

By making this classification completely open, our aim is to empower planners, policymakers, and retailers with actionable evidence:

• Town Centre Regeneration: Identifying whether a struggling high street needs experiential/hospitality reinvention (to cater to Urban Digital Millennials) or accessible value provision (for Budget-Conscious Urbanites).
• Digital Inclusion: Mapping areas where rapid online transitions leave vulnerable, less digitally-connected populations isolated from basic banking and retail services.
• Transport & 15-Minute Cities: Evaluating how mobility constraints directly dictate retail access and consumer resilience under volatile fuel and living costs.

Looking Ahead: Feeding Micro-level Demand into Agent-Based Models

This geodemographic classification is not just a static descriptive output. In our ongoing research at the University of Liverpool, it acts as the foundational empirical demand layer for dynamic spatial simulations.

In our agent-based model (ABM) of metropolitan retail centers (such as RC-ABM), synthetic consumer agents are initialized using these specific subcluster behavioral priors. This allows us to simulate how localized consumer populations choose retail destinations, how they react to retail unit closures, and how external shocks (e.g. transport disruptions, store vacancies) propagate through urban retail hierarchies.