Predictive Demand and Store Replenishment Optimisation
Synopsis
We were entrusted with aiding a prominent dairy and food manufacturer to execute a comprehensive improvement initiative, specifically focused on enhancing demand planning and store replenishment.
The challenge involved efficiently replenishing thousands of customer stores, each with distinct distribution frequency requirements, for over 200 active SKUs. Our focus was to implement predictive demand at the customer store level, specifically tailored for each SKU and applicable on any given day.
In light of intense regional competition and a decline in revenue and margins over the past two years, the priority was to mitigate oversupply and minimize product wastage at the customer level. At the same time, we aimed to maintain net sales and recover lost ground. Despite prior investments in forecasting software from a leading North American supplier showing no improvement, our predictive demand and replenishment optimisation approach significantly enhanced client ROI and outperformed previous software solutions.
Client Overview
🔸c. $650 million (USD) revenue, primarily supplying dairy products andÂ
🔸Several manufacturing plants and 32 Distribution depots
🔸+40,000 customer sites and 260 product lines (SKU’s)
Client Problem
“Forecasting inaccuracies not only erode profitability but also hinder customer satisfaction. With annual losses approximately $40 million USD in rebates due to expired, unsold products, our client sought a solution. They aimed to harness precise demand insights, factoring in variables such as location, customer store profiles, SKU variations, seasonal events and delivery timelines to reduce waste and enhance customer shelf replenishment”.
Â
Analytics Delivery
Brief initial engagement and methodology development, to understand business processes, products, sales history, distribution protocol and to map out framework for project delivery.
🔸SKU-level forecasts with the predictive ability to adjust to sales patterns at individual customer stores.
🔸Models deployed to client server and seamlessly relayed into daily operations after brief training.
🔸3 – level forecast; manufacturing (30 days ahead), logistics (7 days ahead) and sales (next visit)
The Challenge
The Client had enjoyed a difficult few years commercially resulting from:
⦿  Legislation capping prices on certain key products.
⦿  Raw material costs detereorating margin.
⦿  Gross revenue down 9% vs 2 years prior and net profit down 27% YOY.
⦿  Our client operated a direct to store distribution & replenishment model and liable for unsold, expired items on their customer shelves.
⦿  Product wastage from inaccurate forecasting and unsold products writing off 6 – 7% of gross revenue p.a.
⦿  Overstocking was hurting profit line and under-stocking was allowing competitors to capture sales.
⦿  Deliver a demand prediction solution at the most granular level; by SKU, by Store for any delivery day.
Trusted Data Approach
Project goal: to optimise on-shelf availability and minimise expiries and unsold products on retailer shelves.
As many products were highly perishable, we needed to:
⦿ Optimise product supply volume from plant to depot and on to customer delivery routes
⦿ ML models to increase supply into stores SKUs have increased predicted demand
⦿ Advise client on redefining customer segmentation to optimise product mix at the customer level
⦿ Ensure that wastage shrinkage did not result in adjusted revenue falling
⦿ Reconfiguring delivery frequencies by customer store to align with new demand prediction models
⦿ Integrating predictive models into the clients’ planning process via a weekly store forecast by SKU with proposed delivery days / frequencyÂ
OUR IMPACT
%
Fall in delivery miles from optimised delivery schedules
%
Reduction in product wastage
%