Comment: Machine Learning and the Supply Chain
Retail success is now directly related to the speed with which businesses innovate. As a result, the ‘typical’ retail employee is changing, with retail technology teams larger – and younger – than ever before. But while dedicated technologists are increasingly expert in the latest innovations, from Artificial Intelligence to Blockchain and machine learning, how can these solutions be successfully deployed to drive business value?
These disruptive technologies are phenomenally exciting but what business problems are they specifically set to solve? Where is the balance of technology expertise and retail experience required to map the innovation cycle against the level of risk and reward? Retailers may have made extraordinary changes to create technology-first business models, but when it comes to radical innovation, it is essential to understand the art of the possible, not the probable. Andy Hawkins, Product Director, Adjuno, discusses the changing face of technology deployment and technology expertise within retail.
Disruptive for Innovation
The speed with which disruptive technologies move from hype to mainstream is faster than ever; and in a retail market that has been completely transformed by technology over the past two decades, it has never been more important to be at the cutting edge of innovation. A handful of retailers are already exploring Artificial Intelligence (AI), for example, to achieve a completely automated product design to manufacture to customer model; a process that leverages social media derived insight on trends in fashion to create new designs that are produced and shipped with zero human interaction.
Clearly such innovations are in the experimental stage right now and used for just a tiny proportion of the overall product range, but they reveal what is possible. With growing numbers of retail companies now actively recruiting an entirely new type of employee – even hiving off the technologists into separate Millennial and Gen Z friendly areas – technology as an enabler is now an underpinning concept of retail operations.
And a key area of innovation is the supply chain. In the bid to deliver the perfect customer experience, there is a growing awareness across the business of both the upsides and risks associated with supply chain functioning (and misfunctioning). The technologies that support effective supply chain operations are changing both in response to escalating consumer expectation and innovation adoption cycle.
But while the technologies are keen to embrace innovation are the latest ‘disruptors’ such as AI, Blockchain or machine learning really viable options for supply chains given the vulnerability of the retail business model?
Recognising that successful and innovative deployment of technology is increasingly differentiating success from failure, retailers are creating ever larger and more youthful retail tech teams. These individuals are savvy, knowledgeable and ferociously well informed; they are also fearless and keen to push the boundaries of innovation.
This new generation creates a different challenge for the tech vendors traditionally used to coaxing retailers into investment and innovation: when retail customers have larger tech teams chock full of Gen Zs confidently exploring cutting edge solutions than many software development companies, just what can a vendor bring to the party?
In many ways the model has been turned on its head: rather than tech vendors bringing innovation to retail experts, the vendors need to leverage their market expertise, their supply chain knowledge and experience, for example, to demonstrate to gung-ho tech experts just how these disruptive technologies can be adapted to deliver value – and when.
Right now, for example, while Blockchain has legal issues to overcome regarding data storage and ownership, and AI requires both greater technical knowledge and a significant financial investment, machine learning is most definitely a mainstream solution and one that retailers can confidently embrace.
Machine Learning for Automation
Automation is the primary goal at every stage of the supply chain, to improve efficiency and consistency and drive out cost. Certainly, automation within Distribution Centres (DC) has been on the rise for some time, from the integration of voice activated picking to improve accuracy and turnaround to the use of robotics to create a completely lights out operation.
Organisations are also leveraging end to end supply chain solutions that exploit business rules to automate processes. But there are still many areas in which the system decrees escalation is required because the business rules have not been defined – or met - and managers are required to handle the exceptions. From incomplete orders and shipment authorisations, to prioritising the unloading of containers, the decision to move goods out of the factory or into the DC too often requires human intervention.
This is where machine learning is set to make a huge difference through addressing large numbers of these current exceptions by supporting further, intelligent automating of additional business rules. Essentially, the vast majority of exceptions that have not yet been converted into business rules still rely on an element of gut instinct or expertise to be applied. Machine learning will be able to track the decisions being made by expert users and begin to translate those decisions into an automated solution. Essentially the new business rules will be more complex, but they will be based on the patterns of behaviour identified by machine learning solutions and enable businesses to incrementally and confidently automate essential activity throughout the supply chain.
While retailers are increasingly embracing technology led innovation, cultural change remains an issue. And one of the most compelling aspects of machine learning is the lack of cultural boundaries to overcome: it is essentially building on the straight through processing that has been enabled by data driven decision making over the past decade. Managers have already eradicated a raft of manual tasks as a result of fast access to trusted real-time data in conjunction with clearly defined business rules. Now they will be further released from mundane, day to day activities and empowered to focus only on those critical events that could potentially jeopardise the supply chain. In a volatile global marketplace increasingly affected by unprecedented carrier delays and potential trade wars, this ability to minimise the effort spent on business as usual will be invaluable.
Machine learning will also open the door for the next stage of fundamental change by revealing technology enabled opportunities to individuals across the business. While initially, machine learning will enable administrative savings and small efficiency gains, in the longer term machine learning and AI will eventually impact everything from product design through to analysis of demand and forecasting sales through to managing inventory flows through to engagements with the consumer.
Forward thinking retailers are no longer hamstrung by technology fear; in contrast their tech teams are keen to embrace the latest disruptive opportunities. But this remains an industry in flux – and it will be those companies that effectively manage investment at the right stage of the adoption and innovation cycle, that are able to match tech expertise and market experience, that will be successfully manage the shift to tech-first operations.
NTT DATA Services, Remodelling Supply Chains for Resilience
Joey Dean, the man with the coolest name ever and Managing Director in the healthcare consulting practice for NTT DATA and is focused on delivering workplace transformation and enabling the future workforce for healthcare providers. Dean also leads client innovation programs to enhance service delivery and business outcomes for clients.
The pandemic has shifted priorities and created opportunities to do things differently, and companies are now looking to build more resilient supply chains, none needed more urgently than those within the healthcare system. Dean shares with us how he feels they can get there.
A Multi-Vendor Sourcing Approach
“Healthcare systems cannot afford delays in the supply chain when there are lives at stake. Healthcare procurement teams are looking at multi-vendor sourcing strategies, stockpiling more inventory, and ways to use data and AI to have a predictive view into the future and drive greater efficiency.
“The priority should be to shore up procurement channels and re-evaluate inventory management norms, i.e. stockpiling for assurance. Health systems should take the opportunity to renegotiate with their current vendors and broaden the supplier channel. Through those efforts, work with suppliers that have greater geographic diversity and transparency around manufacturing data, process, and continuity plans,” says Dean.
But here ensues the never-ending battle of domestic vs global supply chains. As I see it, domestic sourcing limits the high-risk exposure related to offshore sourcing— Canada’s issue with importing the vaccine is a good example of that. So, of course, I had to ask, for lifesaving products, is building domestic capabilities an option that is being considered?
“Domestic supply chains are sparse or have a high dependence on overseas centres for parts and raw materials. There are measures being discussed from a legislative perspective to drive more domestic sourcing, and there will need to be a concerted effort by Western countries through a mix of investments and financial incentives,” Dean explains.
Wielding Big Tech for Better Outcomes
So, that’s a long way off. In the meantime, leveraging technology is another way to mitigate the risks that lie within global supply chains while decreasing costs and improving quality. Dean expands on the potential of blockchain and AI in the industry.
“Blockchain is particularly interesting in creating more transparency and visibility across all supply chain activities. Organisations can create a decentralised record of all transactions to track assets from production to delivery or use by end-user. This increased supply chain transparency provides more visibility to both buyers and suppliers to resolve disputes and build more trusting relationships. Another benefit is that the validation of data is more efficient to prioritise time on the delivery of goods and services to reduce cost and improve quality.
“Artificial Intelligence and Machine Learning (AI/ML) is another area where there’s incredible value in processing massive amounts of data to aggregate and normalise the data to produce proactive recommendations on actions to improve the speed and cost-efficiency of the supply chain.”
Evolving Procurement Models
From asking more of suppliers to beefing up stocks, Dean believes procurement models should be remodelled to favour resilience, mitigate risk and ensure the needs of the customer are kept in view.
“The bottom line is that healthcare systems are expecting more from their suppliers. While transactional approaches focused solely on price and transactions have been the norm, collaborative relationships, where the buyer and supplier establish mutual objectives and outcomes, drives a trusting and transparent relationship. Healthcare systems are also looking to multi-vendor strategies to mitigate risk, so it is imperative for suppliers to stand out and embrace evolving procurement models.
“Healthcare systems are looking at partners that can establish domestic centres for supplies to mitigate the risks of having ‘all of their eggs’ in overseas locations. Suppliers should look to perform a strategic evaluation review that includes a distribution network analysis and distribution footprint review to understand cost, service, flexibility, and risks. Included in that strategy should be a “voice of the customer” assessment to understand current pain points and needs of customers.”
“Healthcare supply chain leaders are re-evaluating the Just In Time (JIT) model with supplies delivered on a regular basis. The approach does not require an investment in infrastructure but leaves organisations open to risk of disruption. Having domestic centres and warehousing from suppliers gives healthcare systems the ability to have inventory on hand without having to invest in their own infrastructure. Also, in the spirit of transparency, having predictive views into inventory levels can help enable better decision making from both sides.”
But, again, I had to ask, what about the risks and associated costs that come with higher inventory levels, such as expired product if there isn’t fast enough turnover, tying up cash flow, warehousing and inventory management costs?
“In the current supply chain environment, it is advisable for buyers to carry an in-house inventory on a just-in-time basis, while suppliers take a just-in-case approach, preserving capacity for surges, retaining safety stock, and building rapid replenishment channels for restock. But the risk of expired product is very real. This could be curbed with better data intelligence and improved technology that could forecast surges and predictively automate future supply needs. In this way, ordering would be more data-driven and rationalised to align with anticipated surges. Further adoption of data and intelligence and will be crucial for modernised buying in the new normal.
These are tough tasks, so I asked Dean to speak to some of the challenges. Luckily, he’s a patient guy with a lot to say.
On managing stakeholders and ensuring alignment on priorities and objectives, Dean says, “In order for managing stakeholders to stay aligned on priorities, they’ll need more transparency and collaborative win-win business relationships in which both healthcare systems and medical device manufacturers are equally committed to each other’s success. On the healthcare side, they need to understand where parts and products are manufactured to perform more predictive data and analytics for forecasting and planning efforts. And the manufacturers should offer more data transparency which will result in better planning and forecasting to navigate the ebbs and flows and enable better decision-making by healthcare systems.
Due to the sensitive nature of the information being requested, the effort to increase visibility is typically met with a lot of reluctance and push back. Dean essentially puts the onus back on suppliers to get with the times. “Traditionally, the relationships between buyers and suppliers are transactional, based only on the transaction between the two parties: what is the supplier providing, at what cost, and for what length of time. The relationship begins and ends there. The tide is shifting, and buyers expect more from their suppliers, especially given what the pandemic exposed around the fragility of the supply chain. The suppliers that get ahead of this will not only reap the benefits of improved relationships, but they will be able to take action on insights derived from greater visibility to manage risks more effectively.”
He offers a final tip. “A first step in enabling a supply chain data exchange is to make sure partners and buyers are aware of the conditions throughout the supply chain based on real-time data to enable predictive views into delays and disruptions. With well understand data sets, both parties can respond more effectively and work together when disruptions occur.”
As for where supply chain is heading, Dean says, “Moving forward, we’ll continue to see a shift toward Robotic Process Automation (RPA), Artificial Intelligence (AI), and advanced analytics to optimise the supply chain. The pandemic, as it has done in many other industries, will accelerate the move to digital, with the benefits of improving efficiency, visibility, and error rate. AI can consume enormous amounts of data to drive real-time pattern detection and mitigate risk from global disruptive events.”