In this guest article, taken from the latest eDelivery Magazine, David Hogg, commerce solutions lead for Europe at IBM Commerce examines the impact of omnichannel retailing on the retail supply chain and evaluates the impact that cognitive computing will bring to the industry.
If ‘simple’ online retail is the starting point, typically with customer orders being fulfilled from a single ecommerce warehouse, then omnichannel still remains the aspiration for most retailers today. The majority are part way on the long journey to achieving omnichannel retail.
True omnichannel requires not just consistent cross-channel customer experience, but also redesigned sourcing and fulfilment processes and a radical overhaul of the stock holding real estate where these processes are executed. In addition, with the growth in drop ship and outsourced fulfilment, retailers need to ensure consistent fulfilment processes are executed in locations controlled by third parties.
Often retailers have correctly created customer fulfilment processes that offer convenience, like click and collect or returns in store, but have done so without attention to the associated supply chain costs. This was perfectly acceptable business practice in the early “Klondike gold rush” of ecommerce when year-on-year growth was 80% or higher, but today, with average year-on-year growth rates trending towards 10% the focus is understandably changing towards cost control and profit maximisation.
This can be achieved by some obvious steps. However these steps are not easy, with many falling into the ugly category of “business transformation”. For example:
1) Charging the customers to cover fulfilment costs when the product net margin is not sufficiently high to create profit from the order;
2) Changing sourcing to minimize transportation – for example pick in store for click and collect rather than picking from a warehouse and shipping to store for collection;
3) Ship to the customer from a local store rather than from a distant warehouse;
4) Pass the shipping cost on to a drop ship supplier;
5) Optimising the consolidation process where products need to be sourced from different locations but consolidated before shipping to the customer;
6) Changing warehouse location and improving the pick, pack, ship processes to deal with customer orders. Most retail warehouses were located and designed for batch/bulk picking to replenish stores and not customer orders. Hence the move by some grocers to augment their supply networks with purpose-built dark stores. However, very few companies have done a top to toe review of their supply chains, with the explicit purpose of accommodating omnichannel to maximize profit.
Demand Planning and Allocation
Retailers have spent decades building and fine tuning planning and forecasting systems to replenish warehouses and stores with the objective of maximising on shelf availability, while trying to minimise overstocks and markdowns. These processes and systems were built based upon one major source of demand – the store Point Of Sale system (POS). This POS demand signal coupled with estimated adjustments caused by increasingly frequent marketing campaigns and promotions was fed in planning and allocation systems. Now as retailers open up store inventory to online demand and marketing/promotions they are struggling to accommodate them and contemplating redesign of the planning, forecasting and allocation processes.
Different types of retailers face different challenges when adjusting planning, forecasting and allocation for omnichannel. For example, many online grocers still pick customer orders from local stores and deliver with a fleet of small vans. This means that the demand signal from ecommerce is easily incorporated, like the POS, into current planning, forecasting and replenishment. However, for slower moving goods like apparel or consumer electronics, which are sourced and fulfilled from a variety of locations – e.g. store, warehouse or drop ship supplier – the demand signal is harder to interpret and manage. Lower volumes of orders and shipments spread across multiple locations means that the peaks of activity are less predictable.
Retailers in this position need to think more carefully about how they adjust or redesign planning, forecasting and allocation systems. Merchandise and supply chain managers face the unenviable task of making stock allocations for omnichannel inventory based on what can seem to be more random demand peaks.
Most are adopting a softly, softly approach by making small, simple, incremental changes to the rules controlling sourcing and fulfilment as well as their planning, forecasting and allocation systems. They need to see what the impact is on the current processes and systems and decide whether they need to adjust or re-write planning, forecasting and allocation systems to support omnichannel.
The Impact of Cognitive Computing
Today, sourcing and fulfillment processes are controlled by rules that are manually configured in supply chain systems, based on human experience. Cognitive systems are a form of artificial intelligence and they learn from past actions and the resulting outcomes. There are cognitive solutions available now, but they are not widely deployed yet. Logistics operations managers are by nature conservative, as businesses are dependent upon smooth operations. However, the drive towards profitable commerce means that companies are testing these technologies and we expect to see growing deployment over the next few years. The concept of cognitive supply chain systems is that they will supersede these rules based systems. They will take the rules as a starting point and then “learning” from practical experience, they will start to make decisions independently.
So, over time the cognitive systems will learn and make decisions that no person has “coded/configured”. They can also make decisions in near real-time for both planning and execution. This opens up the possibility of radically more sophisticated real-time supply chain planning and execution, with constant responsive re-planning and decision making based on current supply and demand status.
The principal is simple, but how will merchandising and supply chain professionals react to this? How willing will they be to trust these new systems? What, where and when will they build in alerts to trigger human intervention? In each retail company it will take years to test and validate results that build the trust required to reduce manual interventions to a minimum.
For fast moving consumer goods retailer like grocers, cognitive supply chain applications will allow near real-time reaction to changing demand and inbound supply exceptions, like bad weather or transportation accidents. We will see cognitive applications make continual adjustments to forecasts and orders optimised to take advantage of what seems like an infinite source of supply. For slower moving goods like apparel and footwear, or fast fashion where guesstimate allocations of inventory are made and followed up with no, or minimal replenishment from suppliers, cognitive applications will make real-time decisions to balance customer demand with customer order fulfilment costs. This will allow retailers to fulfil orders from locations that were over allocated with inventory. The applications will search for inventory closest to the consumer thus minimising transportation costs. The net result should be maximised sales and profit realised by reduced markdowns at end of season or product lifecycle.
Cognitive applications offer retailers the opportunity to create a more refined vision of future omnichannel retail operations. They can deliver higher sales and profit margins by improved utilisation of inventory and better quality decision making. In an industry that thrives on intense completion and narrow margins they represent an intriguing opportunity that no chief executive, retail operations or supply chain executive should overlook.