Build A Clicks-to-Bricks Strategy Using Spatial Data Science


For online-native retailers eyeing growth and expansion, a clicks-to-bricks strategy using geospatial analytics can mitigate risk in site planning

This post may describe functionality for an old version of CARTO. Find out about the latest and cloud-native version here.
Build A Clicks-to-Bricks Strategy Using Spatial Data Science

McKinsey estimates that in-store sales will still make up 75% to 85% of all retail sales in 2025. For online-native retailers eyeing growth and expansion this makes an omnichannel retail strategy critical.

Embracing a Clicks-to-Bricks' mentality can help these retailers offer hyper-convenient online engagements at the same time as creating bespoke and experiential brick-and-mortar interactions such as letting shoppers try the product and boost their likelihood to buy. In this way the forward thinking retailer can engage a wider audience lower inventory costs and invigorate their brand. But for the aspiring Clicks-to-Bricks retailer one of the earliest and most challenging questions is 'How do I turn my online customers into brick-and-mortar customers?'

Picking the Perfect Site for Brick-and-Mortar Expansion

The number one most illuminating tool for an online retailer looking to establish themselves in the physical world is their own sales data. When informing site planning decisions that sales data when properly analyzed using geospatial analytics can ensure maximum profitability.

To show you how this is done we will use the example of EternalRoad a fictional online retailer eager to step into the real world. EternalRoad a seller of high quality Women's fashion wants to open its first stores in New York and San Francisco two cities where it already is doing significant online sales for shoppers to come in and try-on their clothes.

Clustering to Zero-In on Retail Zones

To start the process of picking the location in these cities our data scientists used a clustering analysis on point-of-interest data related to fashion retail in NYC and SF. This clustering analysis allows us to designate a number of 'retail zones' in each city that we might target for these location.

From clusters around Soho Midtown and the Upper East Side in NYC and Downtown and along Fillmore street in San Francisco we can see that both of these cities have a number of retail zones for us to choose from.

Assigning predicted value to each cluster

The next step for deciding which of these retail zones to target for EternalRoad's new brick-and-mortar locations is to approximate which of the clusters will yield the highest sales revenue both online and offline. In order to do this EternalRoad maps out their online sales in these cities.

The choropleth layers on these maps represents the sales density (or monthly sales per person in that tract) in each of our target cities broken down by census tract.

With our online sales data mapped alongside our retail clusters we can now begin to estimate and assign values to determine which cluster would be the most lucrative. To create this estimate we created 2-mile walk routes from each cluster. We took the sales data from each census tract within that distance and then weighted those sales values based on distance to the cluster itself.

The routing analysis coupled with the sales data allows us to create filters on these maps to look at the top 5 top 3 and top overall estimated most lucrative retail clusters to target for EternalRoad brick-and-mortar locations.

Building a Model to Estimate Retail Success in New Markets

Making predictions based on EternalRoad's existing sales data allows us a degree of certainty as we assess brick-and-mortar locations but EternalRoad has dreams of expansion into new markets where they lack significant online sales data. In order to provide a Location Intelligent site selection recommendation we built a model to predict those sales figures. For this example we looked at Los Angeles.

We built our model off of our San Francisco sales data as that was the closer market to our proposed LA location and demographic data to determine feature importance. The below graph shows feature importance in our model and notes that we adjusted travel time for the LA model to account for longer commute times in the Los Angeles area.

LA Model Feature Importance

Based on our model we are able to create sales projections by census tract for Los Angeles like we had available for NY and SF.

Once again we can visualize this alongside clustered retail point of interest data for Los Angeles. Combined once more with the two mile routing we are able to create a list of retail location targets in Los Angeles based on projected future sales.

Learn how major brands like Starbucks and Dunkin can use modern data streams to inform investment and expansion strategies. Check Out the Recorded Webinar!

Going from Clicks to Bricks with Location Intelligence

For online retailers looking to expand their business more directly engage their customers and make sure they have a slice of the 75%+ of all retail that will be happening in-store over the next 7 years going from Clicks-to-Bricks is a must. But making that change requires a Location Intelligent site planning process. Integrating existing data census and demographic data and even modern data streams like human mobility and credit card transaction data and then performing powerful geospatial analyses to find location candidates can help mitigate risks in any site planning process.

Want to chat about CARTO VL?

Our location intelligence specialists are available to talk about your next project.

Request a Demo