top of page
  • Writer's pictureOliver Scutt

Replenishment Planning for the new normal after COVID-19

Updated: May 1, 2020

COVID-19 has severely disrupted supply chains due to both major demand shifts and supply shortages. As a result of COVID-19, Amazon has prioritized the handling of inbound and outbound 'essential' items over other goods. To slow the intake of non-essential items, they selectively reduced and even stopped placing Purchase Orders on many types of goods. Meantime many consumers cooped up at home started shopping on-line for things they might have previously visited stores for.

The net effect is that stock at Amazon is severely unbalanced, and generally well short of targets. You can see this in the unusually extended lead time that Amazon offers on many products that would normally be ex-stock, product unavailable/stockouts messages and some surprisingly good inventory effectiveness metrics in Vendor Central. We have clients in categories that have gained from home-based work with 75% of their portfolio stocked out, and less than a week of cover on most of the remainder. This is no surprise when a system that typically has 3-5 weeks of inventory cover is starved of re-supply for 6+ weeks. The following shows a client's recent Inventory Balance distribution across markets. In normal times, inventory would ideally be distributed around 100% -- now they have a bipolar distribution that exposes stockouts and what is not selling:

There are a variety of considerations for those navigating the path to the new normal, whatever that is. The focus here is demand planning for e-commerce vendors. They will have scarce supply due to labor or material shortages, some excess on stuff they couldn't ship out and will likely see some abnormally high replenishment requests from Amazon trying to get back to their target inventories and serve pent up demand. In reality, apart from the magnitude of the problem, the scope is no different than the normal demand planning/S&OP considerations of balancing supply and demand with imperfect and fast evolving information.

Some smart analytics such as those provided by Merchant AI can automatically help transform latest market forecasts into best-thinking distribution requirements for consumption by ERP systems. We consider time-phased stock targets, inventory position, existing/Accepted POs, market forecasts, frozen horizon and economics to accelerate decision making, and then show their replenishment requirements relative to historical sales & Submitted POs for comparison. Such data is readily available in Excel for any further analysis or distribution.

On the above chart, snipped from one of our standard PowerBI dashboards a couple of weeks ago, the left hand side shows how inventory (grey) declined along with Amazon orders (blue) for one particular ASIN. On the right hand side, we have the forecasts into the future for this particular seasonal product. Having specified the forecast to chase, we compute the planned replenishment profile (red columns) that will rebuild the inventory back to target and not end up with excess at the end of season. Note the replenishment is pulled forward and peaks earlier relative to the forecast, and that projected inventories are planned to perfectly track their target. This replenishment plan will automatically update each and every week with all the latest data automatically downloaded from Amazon's marketplaces in North American, Europe and beyond. This was a tedious calculation to setup, but now that it is done it is the cleanest signal we can give the business of their planned replenishment requirements at any point in time. The forecast quantities are also readily combined with financial data to project Purchase Order value/Vendor Revenue and working capital implications either for planning, or other decision making. All the same data and measures are automatically published to vendor clients in self-serve dashboards and/or Excel format almost as soon as Amazon's makes their forecasts available.


Recent Posts

See All

Amazon Advertising -- Waste #1: Branded Search

This post is the first in a series that will use data to highlight some of the dubious, wasteful and lazy practices that we observe while monitoring, benchmarking and reporting on the work of major cl


bottom of page