Location analytics is the process or the ability to gain insight from the location or geographic component of business data. Through location analytics, retailers can have a more detailed understanding of their businesses-costs, profits, people, customer satisfaction and loyalty etc. This apart from web analytics eventually helps the business people to spot previously hidden patterns and get answers to the above problems.

The retail market is experiencing cut-throat competition. Retailers are finding it difficult to improve their ROI’s, increase sales, reduce operational costs, and enhance customer satisfaction & loyalty.

It can seem nearly impossible to find solutions to these problems with limited budgets. But that’s the reason why so many retailers are switching to location analytics to extract valuable insights.

Location analytics in retail is a subset of predictive analytics which utilizes the power of machine learning algorithms to utilize different types of data like:

Demographic, Geographic, Infrastructure, Transportation, Organization’s operational data, CRM data, ERP data, and Enterprise asset management data.

Location analytics can be primarily used to perform the following functions:

• Determine optimal strategic location for opening new retail stores
• Optimize sales and marketing mix
• Improve asset management
• Determine factors affecting new store location selection
• Identify high density, high spend locations for new stores
• Determine customer purchasing patterns

So the company wanted to focus on reactivating dormant customers rather than acquiring new customers. But before this it was important for them to find out which customers will be suitable for reactivation.

The Company solved the problem by unleashing the power of data analytics using look-alike modeling. In the past the company had run a reactivation campaign and had identified 561 people out of the customers reactivated after the campaign who generated the highest revenue and this happened in the month of January. In the month of February the client wanted to find people like those 561 people, so by running a look-alike model on new set of 30k people they were able to generate a list of 5.2k people who were most similar to the 561 people identified earlier. After they ran the campaign on the base found by the look-alike model (5.2k) they realized 80-90% revenue from that base. So with this they were able to cut down their marketing cost by 70% and found maximum revenue generating customers among that base (5.2k).

Let’s try to understand how one of the largest pizza chains used location analytics to solve their problem.

The pizza chain had different types of stores in India like dine-in and delivery Stores. The Company wanted to expand and open new stores all over India.

The Company was struggling to find out:

Which region/state to expand in?

Given a set of cities in a selected region, which localities to expand in?

Within a selected locality, where should the store be opened (colleges/railway stations/hospitals)?

What type of store (dine-in/delivery) to open?

The company decided to use data from internal channels like

• List of all existing pizza chain outlets with location and type (dine-in/delivery/both)
• Following historical data for each outlet from the beginning of store operations till date
  • Opening date
  • Customer, sales/transactions, promotions data
  • Operating costs, number of staff etc.
• Opening date of each outlet

External data like:

• No. of households
• Population – total and gender-wise
• Literacy rates – Total, Gender-wise
• Property value index
• Number of food outlets, markets, colleges, hospitals, malls, ATMs, etc.
• Population distribution across age groups, marital status, education level, employment status
• Rural and urban population
• Number of households owning laptops, fridges, etc.
• Number of households availing telephone, mobile and internet service

 

The company followed a stepwise approach to select the most suitable localities and store type for new stores:

• Segregate existing stores into different categories based on overall sales, for example – high, average and low.
• Link the stores and their locations. This gave a list of locations having one or more stores along with their sales category and store type.
• Identify major states/cities where the pizza chain has less presence. The shortlisting was done using different parameters like population, infrastructure, tier, etc.
• Create variables/attributes at district and locality level, using external demographic data for locations with existing stores, as well as localities in shortlisted cities/states.
• Find localities that have similar attributes as the locations with high sales stores, i.e., create a lookalike model.
• Calculate similarity scores for prospective localities and prioritize them using the score.
• Number of stores that can be opened was decided using population information for locations with existing stores and comparing them to prospective localities.

The company was also able to forecast the sales volume for first 12 months for new outlets that were supposed to be opened depending on the location of the store. Thus was able to allocate the resources to every new store depending on the expected sales.

Want to know the use cases of location analytics for your business? Then feel free to reach us at info@transorg.com