Using spatial aggregation and consumer predictions to analyze markets

A fundamental part of location intelligence, spatial aggregation combines individuals into groups and generalizes information about them.

Using spatial aggregation and consumer predictions to analyze markets

This article is part of Faraday's Out of the Lab series, which highlights initiatives our Data Science team undertakes and challenges they solve.

Smart location-based growth strategies require a combination of predictive and geospatial analysis capabilities. Consumer predictions analyzed with a geospatial lens lets brands compute realistic revenue potential in specific geographies.

There's tremendous value in predicting consumer behaviors on the individual level, but you can learn a lot about a market by aggregating those predictions using geographic boundaries.

What is spatial aggregation?

Spatial aggregation — a fundamental part of geospatial analytics — combines individuals into groups and generalizes information about them. These groups are based on known boundaries, with which most of us are familiar: states, counties, metropolitan areas, and others.

The Atlanta metropolitan area shown five different ways, with the population grouped by counties, cities/towns, post codes, census block groups, and individuals.
The people of Atlanta – at five levels. (Labels ©OpenStreetmap contributors, 2020)

The Atlanta metropolitan area serves as a good example; an individual living here fits within a set of distinct boundaries they might not even be aware of. Folks in Hopkins Mill are part of Census block group 131350503142, ZIP code 30096, in the city of Norcross, in Gwinnett county, in the greater Atlanta metro – and each of those aggregate levels can offer unique insights if you're a brand with market expansion in mind.

Consumer predictions at the macro level

There’s key context to be gained in these larger units. Many organizations build their strategies at a national level, choosing new markets to enter and others to deprioritize.

At the highest levels – that of state or Direct Marketing Area (DMA) – the combined modeled propensities of individuals form a picture of concentration and dispersion. This understanding can answer many questions for brands planning for expansion: Does the Atlanta DMA seem like a safer bet than the Las Vegas DMA from an ROI perspective? Which should get a bigger share of the media buy? Which is more likely to benefit from a new branch?

A map of the United States color-coded by opportunity index, with Atlanta showing more promise than Las Vegas in this example.
In this spatial aggregation example, Atlanta shows greater opportunity than Las Vegas.

These questions may be answered with a hunch, but areas of opportunity can be further investigated and substantiated with an aggregation of predictive model scores. (Ideally, both strategies are involved: statistical projections always benefit when paired with local and situational knowledge.)

Consumer predictions at the micro level

On a smaller scale, spatial aggregates within the larger units can be used to prioritize resources. In the Atlanta metro, individual propensity aggregated to the census block group level can be used in combination with geographically weighted models to guide the placement of out-of-home media like billboards, or even storefronts.

A map of the Atlanta area, broken down into smaller color-coded chunks.
"So that's the spot for our next pop-up..."

Mind the gaps

Some important details are lost when people are rolled up into spatial units like this. Geographers remind us that there is no solution to the modifiable areal unit problem (an issue of statistical bias), and the outliers supplanted with a single median value may actually tell key parts of the story. Beyond these is the essential truth that boundaries are created to serve a purpose — any aggregation encodes the politics of its border, often to the detriment of those within it.

Whatever your strategic use case, be it trade zone analysis or geo-targeted ad campaigns, individual scores from a well-trained predictive model are just a starting point. Use them — wisely! — as the building blocks of market expansion, and you’ll be able to operate intelligently at the scale of a city or state as well as at the level of a mailbox or computer screen.

Learn how Burrow optimized their Facebook spend and showroom placement with AI

Wondering how you can leverage spatial aggregation to improve your marketing and sales strategies? Check out Faraday's Location & Market Intelligence solution.