Collaboration and machine learning key for successful inventory

Why does it take so many systems and spreadsheets to order inventory and place it in stores and fulfillment centres? Spreadsheets, software and even people can get in the way of improving inventory placement processes. Software companies promise that machine learning is the answer, but machine learning without good business processes is a recipe for wasted capital investment and expense dollars.

The good news is that there is a better way. But it requires real work—not shortcuts—and revolves around five important ideas:

  1. Tear down the silos: Work with one plan, not three

Inventory placement today is controlled by three, often disconnected plans. There are plans created by merchants when curating the assortment, by planners when creating the bulk orders, and by allocators/replenishers when restocking the stores. This disconnection leads to lost opportunity with regards to inventory levels, turn, in-stocks, and end-of-life markdowns. A single plan available to the whole team reduces the opportunity to drop the ball.

  1. Collaborate: Merchants, planners, and allocators/replenishers

Having one plan requires all three teams to collaborate on that plan and requires tools (whether they’re spreadsheets or something more expensive) that function like Google Docs and facilitate real-time collaboration. Some retailers may find it beneficial to merge two of the roles, considering that a collaborative team is a more efficient team.

  1. The plan should exist at multiple levels simultaneously

The reason why silos exist in the first place is because each of the three roles makes inventory decisions in different times and places—from full-lifecycle and the entire chain for merchants, to the day/store level for allocators and replenishers, with planners somewhere in the middle. The business processes and tools must align on a way to manage—creating a plan to account for top-down, middle-out, and bottom-up inventory, ordering, assortment, and promotion decisions simultaneously and in real time.

  1. Promotions and assortment changes should be a living plan (real time)

Another reason for silos and multiple plans is the merchant must make real time decisions with regards to pricing, promotions, and assortment without waiting for overnight batches. The business process must be built around this need to make decisions for the chain and for the store in real time. Historically, tool capability has constrained this ability, but no more. You should demand real time promotion and assortment analysis from your business processes and from your tools.

  1. Machine learning is important, but so is the intelligence of the team

Just like the calculators that we used to lug around from meeting to meeting, machine learning (ML) gives us a more detailed way to plan inventory and evaluate changes to the plan. However, some describe ML as if the “M” stood for Magic. Machine learning without sound business processes (not just data) will waste your money. IDG Retail Insights refers to this combination of machine learning and the wisdom of the team as contextual collective intelligence (CCI). Machine learning-driven tools give the merchants, planners, and allocators/replenishers a more powerful calculator to plan at multiple levels in real time, evaluating prices, promotions, and assortment changes to ensure inventory is bought and placed for maximum service levels and profitability. But it is just that—a calculator—and one that needs to be placed in the hands of a business-savvy user and supported by a collaborative planning process.

The best way to place inventory actually has more to do with business process than it does machine learning. Collaboration, tearing down silos, and thinking in multiple dimensions with a real-time business process are the keys to better inventory placement in stores and fulfillment centres.

Matt Jones is Vice President – Retail Strategy at Infor

This article was first published by RetailBiz

TOP