By Stephen Gardeen
Principal Data Scientist at Cognira
If we were to ask a retailer how much revenue they lose in lost sales per year, many would scratch their heads, or turn to unreliable data to crunch the numbers. The reason? They don’t have the proper tools to calculate it.
To put it simply, most retailers have a very unsophisticated (if any) methodology to account for lost sales and it is typically inconsistent across the business. However, by sweeping this problem under the rug, retailers don’t have a clear idea of how it is damaging their bottom line.
In order to better assess business performance, retailers need to invest in understanding their historical selling performance (which involves product sales that are inventory constrained).
Throughout this blog, gain insight into how to accurately analyze lost sales, and the importance of investing in advanced technology to make stronger and more accurate predictions.
5 tips on how to analyze lost sales (accurately)
1. Operate at a granular scale
In order to track lost sales, retailers need to look at sku/store/day for both sales and inventory data. For most tier 1 retailers, this is at least 1 billion rows of data if you are looking over an extended period of time. Keep in mind that this amount of data can get overwhelming, and near impossible to process without the help of advanced technology.
2. Avoid calculating lost sales with out of season items
When calculating sales performance, it is important to establish the item’s sales period to avoid including a sale outside of a sale period. Today, many retailers have a generous return policy, allowing customers to return items outside of the sales period. For example, a customer may buy a long sleeve shirt in December (during the sales period), and later return it in June (way outside of the sales period). The retailer then puts the item back on the floor, and it sells during the June sales period. On paper, this can look like the long sleeve shirt had a very long selling period, which could result in an inaccurate lost sales calculation.
3. Look into substitution across products
While a preferred item might not be available, it is important to not jump to conclusions and assume a negative impact to your overall selling performance. Retailers should identify if there is another similar item that satisfies the demand.
4. Consider items as attributes
Analyzing demand is hard, and can be even more challenging for items that were never on the floor. However, one workaround is to look at items as attributes, and identify what is trending. By doing so, retailers can get a better idea of what items to put in their store based on demand.
5. Analyze the in-stock sales patterns to estimate out-of-stock situations
An initial simplification could be to look at how an out-of-stock item is currently selling in other stores, or how similar products are selling at the same store. However, defining similar products can get complicated, requiring a need for advanced technology to ensure accuracy.
Analyzing lost sales with advanced technology
To better adopt the 5 tips presented above, we recommend utilizing advanced technology. With the power of advanced technology, retailers are able to train AI/ML models to get predictions on data they do not know, creating a solution for managing lost sales that is both scalable and accurate.
We understand that this is not a simple task, and requires expertise both in retail and advanced technology. At Cognira, we work within this domain, pairing our retail expertise with the latest technology to solve complex retail problems such as dealing with lost sales.
Interested in learning more about problems we can solve with advanced technology? Contact us here.
