It shouldn’t be surprising that better order, inventory, and shipment visibility tops the priority list for 60% to 80% of companies in supply chain surveys. 

Where manufacturers and retailers once developed, stocked, and pushed out large volumes of goods to regional markets based on predictable historical and seasonal patterns, D2C e-commerce is accessible to much wider audiences via the internet, on a pull basis. An aggregate, near-continuous flow of smaller orders shipped on demand, along with rising overall freight demand, has swamped terminal, warehouse, equipment and vehicle capacity in a tight labor market. 

Mutable customer expectations compound the difficulties. Last-mile pressures and costs are vastly different for palletized freight held in a distribution center for gradual release to factories or stores at the shipper’s direction, versus time-definite orders with multiple delivery time and location options and a baseline expectation of on-time and in-full delivery.  

Whether it’s a new pandemic variant, weather event, or containership blocking the Suez Canal, unforeseen circumstances can easily provide a tipping point that throws demand, supply, and capacity out of alignment overnight. 

Visibility’s Many Moving Parts

Most supply chains still lack adequate visibility on the downstream demand side at the point of sale (POS), upstream in supplier sourcing and production, and in transit during shipment. Sensing demand early is especially critical given ongoing market volatility due to steady D2C growth, amplified by the pandemic, climate, the war in Ukraine, global inflation and other external pressures.  

Demand signals, more than any other single influence, drive the supply chain. They dictate what to produce, in what quantities, and where to ship — in short, everything from sourcing to asset and resource allocation to workflow. It seems counter-intuitive, then, that most conventional hierarchical supply chain models still don’t connect factories and suppliers directly to retailers and customers in a virtuous feedback loop.

Instead, most communication flows from the center outward, and partner input rarely extends beyond one level up or down, trapping critical data inside organizational silos. Third-party aggregator data languishes in marketing, customer relationship management (CRM) data in sales, production data in operations, and in the C-suite. This poses a significant risk of higher costs and lost business in the event of a disruption. 

Supply chain complexity compounds the problem, with more than 60% of global consumers now using e-commerce, more than 25 million global retail outlets open, a ten-fold increase in new products coming to market each year over the past decade, and 10% of merchandise experiencing stockouts.

“In emerging markets, global manufacturers ship through distributors, and their visibility stops at that point,” explains Suresh Prahlad Bharadwaj, platform head for TradeEdge at EdgeVerve Systems, a wholly owned subsidiary of Infosys. “They don’t know who their customers are, mostly small mom-and-pop stores. Even in modern trade, where manufacturers are selling through a wholesaler or directly to a big-box store like Walmart or Target, they are not equipped to process that point-of-sale visibility coming back to them.” 

In a decentralized e-commerce environment, Suresh says, points of sale can be dispersed among hundreds or thousands of distributors, retailer, and websites, all with different levels of maturity in collecting and sharing data, and different ways of formatting data and communicating. 

“Who are my customers, where are they located, what are they ordering?” asks Suresh. “To know that, I need to collaborate with retailers to get that aggregate point-of-sale and store inventory information back to manufacturers quickly, so they can make adjustments.” Right now, he adds, that process can take three to four weeks, relying on third-party data syndicators like Nielsen or IRI to collect and harmonize data from a panel of stores, and then prepare custom reports for particular clients. “In today’s world,” he says, “that’s too late.”

As cloud-based data processing power has increased and costs have come down, Suresh explains, more retailers and intermediaries are cutting direct data-sharing deals with customer companies to disperse primary-source sales data back up the chain. But that’s only the beginning.

Finding Needles in Haystacks

Software-based demand-sensing tools, aided by artificial intelligence and machine learning, are gaining attention for their ability to predict near-future demand. These tools model aggregate real-time POS data against internal and external supply chain anomalies such as climate events, port congestion, a rail strike, fuel price moves, interest-rate rises and high unemployment rates — all of which influence purchasing decisions. 

In short, understanding in a granular way the conditions under which goods were sold yesterday offers short-term insights into how and where the same goods are likely to sell tomorrow, under the same or different conditions. As more granular data is collected over time, artificial intelligence and machine learning sense patterns and insights that would be missed by a traditional manual operation running on an enterprise resource planning (ERP) suite. More frequent reporting intervals shorten response time when sudden, more pronounced events occur.

Given the near demise of traditional long-term strategic and demand planning since the onset of COVID, constructing near real-time data in this way can yield important benefits. Suddenly companies are working off yesterday’s POS store-SKU sales and inventory data, versus weeks-old summary reporting. Sales data also tends to deliver more accurate demand forecasting results than comparable shipment data, since goods may be shipped for a variety of reasons — exchanges or sample merchandise, for example.

Using defined business rules and standards as benchmarks, AI and machine learning map retailer SKU, product, UPC and other coding against manufacturer codes as part of the onboarding process. They can also differentiate among standard and promotional SKUs with, say, small content changes for the same product. An important benefit is the ability of AI and machine learning to analyze and eliminate phantom inventory and display voids in order to predict and reduce stockouts. Using analytics, companies can validate sales trend data within hours.

“One of the things we know about forecasting is that it’s not going to be accurate,” Suresh argues. “So the question becomes how we plug the gaps. We do it through execution of short-term replenishment decisions across the entire network.”   

Building the Supply Chain Value Network

Downstream visibility into how markets and customers interact to influence sales, generating valuable demand signals in the process, sets the table for a larger rethinking of the entire supply chain. 

Visibility both upstream and downstream, from order to payment in a non-hierarchical, “many-to-many” network model, presents an opportunity for end-to-end, real-time data reporting and sharing, and for collaboration by all parties in the network. 

The process begins with building a single, trusted, shareable source for information across the network. Partners are onboarded with appropriate permissions to access specific types of data for specific uses. Data, including relevant forms, documentation, and communications, are standardized, harmonized and structured in a common database format for ease of use. 

So what happens when the demand signals begin flashing? Can production be quickly scaled up or down, or the product mix and sequencing modified to ensure orders are filled on time? Do Tier 2 suppliers have the materials and parts to surge production as needed? If not, can existing inventory in the system be located, redirected, and replenished? If not, should operations and planning teams be rethinking safety stocks, supplier diversification, or product portfolio alternatives? What would be the cost impacts? Time is critical in getting answers to these questions and taking the optimal corrective action.

The important difference with the network model is that suppliers, manufacturers, and retailers can not only sense demand shifts, but also collaborate directly and proactively, in real time, to solve problems, rather than each having separate, siloed communications through the principal company where crucial details can get lost in translation. In addition, AI and machine learning-enabled analytics can run hundreds or thousands of scenarios in minutes, gaming each out based on current and historic shipment and inventory data to formulate an optimal solution.

But as the old tech adage goes: garbage in, garbage out. Network performance is only as good as partner buy-in and an accurate dataset. “It’s not just about technology in the cloud,” Suresh insists, “it’s about driving the compliance of partners in reporting, the volume and timeliness of data, granularity of the information and the frequency with which it’s shared.”

Suresh acknowledges that, up to now, it has been mainly very large companies, in the $6 billion and above range, that have driven this level of digital transformation, in part because of their leverage to force and manage change with smaller suppliers, vendors, and customers. But he sees an opportunity in recruiting clients in the $1 billion to $5 billion range. 

Where is all of this headed? It will become imperative over time for businesses of all sizes to undertake digital transformation, leading to the interconnection and consolidation of supply chains over time. Look for more operations and processes to be automated, further shortening response times, eliminating error, and compressing the order-to-pay cycle, while freeing up people and resources for more productive, rewarding work. Onboarding and data harmonization will likely become almost plug-and-play for small and mid-sized suppliers and vendors, with network capability emerging as a key differentiator on the way to becoming ubiquitous. 

The bottom line: After a brief, at times difficult, period of adjustment, the supply chain is about to get much faster, simpler, and more resilient. 

Resource links: 

EdgeVerve, http://www.edgeverve.com 

TradeEdge, www.edgeverve.com/tradeedge