Increase Margins With Predictive Localized Analytics

As I go through the discovery process with prospective SPI customers, there's a moment in almost every engagement when one of the planners or allocators shows their current multitab, multidimension, multipivot table, holographic (OK, I exaggerated on the last point) homegrown Excel spreadsheet. It always impresses. I'm genuinely in awe of the complex usage of Excel spreadsheets — both the creativity that inspired the retailer to build it and the commitment to detail administration to then maintain it.

Setting aside my awe and respect for the spreadsheets, predictive analytics has now arrived and it's time for retailers to let go of some of their spreadsheets.

A great example and opportunity is the use of localized analytics — e.g., incorporation of SKU-store sales and margin — in the store allocation process.

Excel files are inherently limited to compiling data in pre-defined aggregates, such as item averages across all stores or stores in pre-defined ABC clusters. In contrast, analytics tools are now available, such as SPI’s allocation solution, that are capable of using daily sales data by SKU and store location in the allocation algorithms to identify localized performance differences and accordingly adjust the allocation quantity.

For example, the size distribution for a dress may show 25 percent size large when aggregated across all A-ranked stores, but regional variations will show some A-ranked stores with only 20 percent to 22 percent distribution of size large in the same dress, while others show 28 percent to 30 percent. That variation mathematically causes stock-outs and markdowns for individual sizes in all stores that vary from the average. Analytics-based allocation algorithms solve that issue.

Another example, weather, has a significant impact on individual store-level sales. Allocation algorithms that can incorporate adjustments to reflect storm impact on actual sales results as well as predictive sales impact on stores anticipating an upcoming storm are highly valuable to the allocation team.

It’s not just theoretical. Empirical evidence from SPI customers has shown margin improvements of one to three basis points by applying predictive localized analytics to allocation and assortment planning processes.

If you're still planning and allocating on spreadsheets, it’s time to consider technology alternatives with relevant predictive analytics.