Site selection is an important consideration for most retailers big or small. Demographic and geographic data is widely used to determine optimum sites. Location analytics answers question like:

• How many stores can a particular market take?
• How will a store perform in this location?
• Which stores are underperforming and why?
• Where are the most profitable customers?

While newer approaches use GIS systems with the traditional methods inbuilt into them, the traditional approaches to location analytics are:

• Analog methods- are crude models used by smaller retailers. They rate potential sites on a numeric scale and generate sales predictions by identifying best matches to existing stores. The rating is then compared to existing sites with similar scores — the sales at the new site should be analogous to those at the existing sites.
• Statistical models-use multiple regression techniques to identify best sites based on what will drive sales volumes. These models are often used by apparel and clothing, book, convenience store, furniture and office supply retailers.
• Gravity interaction models-are spatial analyses models which use the number, location, and drawing power of competitors in the region.

Some important variables used in these models are store area socio-economic and demographics, store and site attributes (e.g. square feet), situational attributes (e.g. nearby traffic generators), competition (e.g. amount of competition within a certain distance) and in-store management quality type variables.

As retailers look to customize their stores to meet local customer needs, they want to understand what drives sales at a store level to be able to analyze store results, predict key operational variables, and optimize store-level resources.

Retail analytics provides the option of building models that tell retailers what these store level sales drivers are. Research in this area has hypothesized and analyzed conceptual models based on four primary factors that affect store performance i.e. store related, market related, customer related and competition. Some examples of variables used for each of the factors are:

• Store related-store selling area, format, location, age of store, discounts etc
• Market related-no. of households or population density around store, households with higher family and children size etc
• Customer related- income surrogates, consumer durables ownership, car ownership, education, age etc
• Competition related-no. of competing stores in the area etc

The store performance variable can be measured as sales (sales or sales per sq foot), store traffic, market share or store profit. The driving factors or variables are modelled using regression techniques or structural equation models. Data required to execute analysis for store performance primarily comes from internal store performance data (sales by category, brand and sku), geo demographic data, census or other consumer demographic data and trade area data.

Analytical models built on this data need to start from studying importance of each of these factors individually and then incorporating the extent to which these factors interact with each other to drive store sales. Retailers need to build and look at multiple measures of store performance based on different retailing actions to be able to fully understand store success.

While most retailers are familiar with customer segmentation, the last decade has seen them looking to localize their knowledge about their stores. In-depth store level knowledge allows retailers to narrow down store level issues to distribution, price of sales related and address them. Many retail stores are as different as the customers who visit them. Stores can be segmented and differentiated based on a lot of important metrics like-sales, size, location, local customer demographics, formats, competitive density and age. Analytical techniques like cluster and latent class analysis can then be applied to derive similar store segments. Based on these store segments, the retailer can optimize their assortment mix. Assortment can be customized for each store segment by focusing on SKUs that are in high demand in the store segment in question. This exercise is called assortment planning and optimization

The key benefits for retailers who develop a marketing strategy based on store segmentation are:

• More effective targeting via promotions to customers who visit a particular set of stores based on understanding of their shopping trips to these stores
• Better testing of promotions and new product launches by using test and control store designs
• Benchmarking ‘high’ and ‘low performance stores and taking action against low performers
• Development of optimum pricing strategies by analyzing impact of price on volume at store level aggregation of data

One of the major business decisions a retailer has to face on a day to day basis is, what size of store they should sign up for when opening a new outlet? A variant on the same is, are their current stores optimally sized?

A smaller than ideal store could mean, lost opportunity for sales, as a lower variety of stock is stocked & displayed, compromising the assortment range. A larger than ideal store, would result in lower Sales / Square Foot as compared to internal and external benchmarks.

TEG creates a 2 phase solution for store size optimization

Phase I – Determination of ‘Key Drivers’ of store sales:

• To validate hypothesis, that Store size does indeed have an impact on sales

Phase II – Quantify the impact of the store size on sales & determine optimal store size:

• If size of store is determined to be a ‘significant driver’ of sales, in Phase I
• Basically, we wanted to catch the Apex of the curve shown below, if possible and determine, if there is a point of diminishing returns for store size