In the retail clothing market, retailers often purchase clothing from a brand manufacturer with a purchase agreement that includes a buy-back clause. This clause stipulates that if the merchandise doesn’t sell, then the manufacturer must buy it back. This overstock is often then sold to off-price retailers, such as TJ Maxx, at a substantial discount. For clothing retailers that are also brand manufacturers, such as Gap, the excess inventory is heavily discounted in their brick-and-mortar locations.
Consider a scenario where a large clothing retailer is looking to transform its brick-and-mortar overstock tactics into a hybrid commerce strategy. An idea was proposed to offer overstock as a limited-time promotion where the price would automatically be optimized according to purchase popularity. As more people purchased an item, the transaction price would drop for everyone within the offer period. Initially, the promotion was considered an initiative to be integrated into the existing e-commerce solution. However, to mitigate the risks and costs associated with such an initiative, the decision was made to run a two-week experiment resulting in a functional proof of concept (POC) that would be used to measure the effectiveness of the idea within a single 12-hour shopping period.
Microsoft Azure was the primary cloud provider for the organization, so the architecture for the POC was based upon the Demand Forecasting and Pricing Optimization Azure solution. Azure App Service would be used for the web application and Azure Databricks would be used to build the pricing optimization model. Due to the desired speed of the experiment, data from the POC could not be integrated into the existing enterprise data warehouse within an acceptable timeframe. As a result, transactional data from the POC would be stored in Azure PostgreSQL.
The desired pricing optimization model wouldn’t simply adjust the price as more people made a purchase. Other factors, such as historical transactions, current inventory, seasonality, and purchase timing would need to be incorporated into the model, which presented an unforeseen challenge.
Depending upon the learnings from the experiment, the organization intended to run a pilot of the initiative during the winter holiday shopping season, meaning that the POC had to be completed quickly. A small team was assembled to run the experiment and build the POC. Although the web application was completed very quickly, building the pricing optimization model proved to be far more difficult than originally expected. The immediate availability of additional resources was limited and the timeframe for the POC to go live could not be adjusted. A short-term alternative for the pricing optimization model was needed in order to keep the initiative on track and to give the data scientists additional time to complete the model for use in the pilot.
Considering the short, 12-hour shopping period the POC would operate within, an alternative to the automatic pricing optimization model was to manually adjust the pricing of the offers based on the real-time recommendations of a group of seasoned analysts. With the proper access to purchase data and inventory counts, the analysts could provide real-time recommendations for POC product pricing.
The analysts utilized Power BI for reporting and dashboard activities. To achieve real-time connectivity to the POC transactional data and take advantage of regional inventory data from multiple sources, Conduit, Blueprint’s lightweight data virtualization accelerator, was incorporated into the solution.
The analysts were able to leverage the capabilities of Conduit to build a series of dashboards, enabling pricing decisions to be made in real-time and the POC to be completed on time and ready for its initial test. Rather than investing heavily in an unproven concept, the retailer, using Azure services and Conduit, was able to experiment to validate the idea rapidly. Once the pilot is complete and moved into production, the retailer will be able to sell through overstock inventory much faster and repurpose more clearance space in its brick-and-mortar locations for seasonal displays and other purposes.
Find out more about our accelerators and how you can access the data you need today.