Tools for Merchandise Forecasting
Moreover, such systems are based on statistical-analysis functions that are far more accurate than rule-of-thumb measurements.
The e-commerce demand model is much more opportunistic, offer-driven and instantaneous, on the one hand, and more unpredictable on the other. But since the immediacy of the Web allows you to easily remove sold-out items from your site or to engage in aggressive, short-term discounting to off-load overstocks, forecasting tools per se are somewhat less critical in the e-commerce space.
To be sure, there’s more to merchandise forecasting than predicting demand for any item or category. George Mollo, president of GJM Associates, a catalog consultancy, notes:
You can’t plan any item in the abstract. This business is all about relationships: of items to their categories, of new to carry-over items, price points offered, perhaps fabrications offered, number of appearances (catalog drops), as well as the internal and external competition each item might encounter.
If these basic factors aren’t considered in a formal planning process and, in turn, methodically applied to item forecasts, even the most sophisticated formal forecasting system won’t improve your inventory accuracy.
At the very least, you should have an “average item index” score that shows how each carry-over item performed compared to the average revenue per square inch for all items [during] the same season last year. The score permits assortment planning from a top-down perspective, while forecasting is used to revise plans from the bottom up (based on future projections).
Mollo also warns that if you project an expensive item to sell well on a unit basis, don’t let that revenue skew your planning. False averages can distort projected quantities on all other items in an assortment.
Now I think you can begin to see why doing this right requires a tool that can guide you through this thicket of variables.