Invent helps apparel retailer increase gross margins through more accurate replenishment decisions

DOWNLOAD
Background
The client, an apparel retailer with 250 stores, was concerned that it was not optimizing the allocation of its inventory and approached Invent to develop an improved replenishment decision-making system. The client was using simplified forecasting and replenishment methodologies that did not take into account important variables such as seasonality and promotion sensitivities. The client’s replenishment processes also had an additional layer of complexity as the client operated two warehouses that were limited in their daily capacity to process outgoing shipments. However, the replenishment methodology in use did not have a prioritization mechanism to achieve the most profitable distribution of available inventory given the warehouse capacity. The client was keen on gaining visibility into the causes of lost sales to quantify the cost of mismatches between supply and demand. Knowing whether lost sales are due to lack of inventory stemming from initial purchasing decisions or the replenishment system would give the client actionable insight into how it could prevent lost sales and improve in-stock rates in the future.

Solution
Using historical data, Invent conducted a system assessment. A retrospective demand forecast was developed to quantify lost sales and identify opportunities in the replenishment process.

Impact of calendar events, seasonality, weather and promotion sensitivities were calculated for every product and store, which were later incorporated into the demand forecasting model. Among its library of replenishment algorithms, Invent tested and selected the algorithms that yielded the highest profits on the client’s data. The best performing algorithms dynamically prioritized and profit-optimized stock replenishment for each SKU-store combination taking into account lead times and warehouse capacity constraints. The results of this optimization are synthesized into daily replenishment recommendations made available via the system’s user interface.

Results
To isolate the effects of implementing the replenishment recommendations, a six-week long controlled experiment was conducted across all categories. Stores were split into test and control groups using an algorithm for pair matching. The results were clear. By having the right stock available at the right place and time, Invent’s replenishment solution helped the client increase sales revenues by 2.5% and gross margins by 3.5%. Based on the pilot results, a full roll-out of Invent Replenishment has been completed.