To ship or not to ship, that is the question. Should you drive inventory levels down to increase shareholder value with great ROI but run the risk of being out of stock? Or, do you stack inventory high with hopes of selling more but risk having to mark it down at the end of the season? And how do you decide that?
Efficiency vs Effectiveness
As an example, a policy not to replenish on the weekend due to higher freight expenses may be prudent from a cost perspective, low shelf-availability on your best sellers on a Saturday, a day that often generates 60% of the weekly sales, may not be the wisest as the high cost of replenishing on the weekend might only be a fraction of the incremental sales you make because people can find their size. And remember that happy customers return and unhappy ones do not.
100% conversion - impossible
The problem is that, especially in the fashion and sporting goods industry, typical rates of sale (the number of units sold per retail location each week) are often not higher than 1, making the decision ‘to ship or not to ship’ a critical one. Highly important and easy to get wrong.
If we distil this decision to its ROI impact for each SKU, then selling an item from 2 units of inventory on hand delivers 33% more ROI than selling the same unit with 3 units on hand. Logically, having only one unit on hand is even more profitable but the risk of stockouts increases significantly.
Why you should apply machine learning
For example, it is common knowledge that 80% of the store’s revenue is typically generated from only 20% of the store’s assortment, the so-called Fast Movers. Therefore, making the decision ‘to ship’ (an extra unit) to the Fast Movers would seem common sense. Shipping and extra unit significantly improves the chances of selling more, while the focus on Fast Movers reduces the risk of getting stuck with redundant stock that needs to be marked down to clear.
While this may be common sense, how do we know what the fast-movers will be in each retail location? This is where data-science and machine learning comes in. Distilling fast-movers to their success factors (style, colour, price, fabric or fit) enables ‘the machine’ to find out why certain items are Fast Movers and to predict locally fast-moving merchandise even in a network of hundreds of stores and hundreds of thousands of stock keeping units.
Leave it to the machine
While averages may serve as indicators of the financial health of a retailer, they fail to drive operational decision making, as each SKU is an opportunity to generate cash and improve the ROI.
Ensuring high shelf-availability and a healthy ROI for each SKU requires that the decision ‘to ship or not to ship’ is made at the level of SKU and that in turn is best left to a machine that hosts advanced technology and artificial intelligence and learns all the time.
Have you enjoyed this article? Please share it on your favourite social channel.