By the time I have finished writing this article, and certainly before it is published, the situation will have changed. This year has been one hell of a rollercoaster ride, with changes to the global trading environment every day. One is reminded of the Chinese proverb, “Better to be a dog in times of tranquility than a human in times of chaos.”
I recently hosted a webinar in which our star speaker was Lora Cecere, Founder at Supply Chain Insight. And insight is what we got, in spades. Here are some of the lessons I took away from the webinar.
Supply chain management typically does not fit very well with procurement, which is a challenge at the best of times, and can be a disaster in difficult times. As late as the 1990s, supply chains were very local. Then they became regional: for example, a North American supply chain and a European one. And then, in the 2000s, we saw the start of truly global and interconnected supply chains, with growing volumes of goods in transit.
The success of this globalized model rested on three assumptions, the first of which was that governments would act in a rational manner to ensure frictionless trade. Well, now we are seeing unprecedented irrationality, with trade wars being fought over tariffs and compliance issues. The second assumption was that variability would be low, and IT systems have been implemented in the belief that the global economy would remain on an even keel. But with wild fluctuations in prices, currency conversion rates, lead times and so on, IT systems have had difficulty coping in recent years. We need to rewrite these systems and introduce new taxonomies to manage hyper-variability. The third assumption was that the logistics infrastructure would always be available. But during Covid, and other constraints like the Suez Canal closure, we saw what happens when suddenly logistics are not there when needed.
Consequently, we must now deal with a world in which these three assumptions no longer apply.
Many put forward artificial intelligence as the magic potion to solve these challenges. But as Lora said, there is little point in applying AI to give us better optimizers and taxonomies when the basic assumptions behind globalization have been fundamentally undermined.
She suggested that we need to “unlearn” what we know before we can “learn” how to fix this situation. The facts have changed, and knowledge is not necessarily understanding.
Demand latency & the bullwhip effect
Over the past six years, we have witnessed a sharp increase in demand latency, defined as the time lag between customer purchase behavior (at the point of sale) and the corresponding signal received by those responsible for replenishment or production. Moreover, behavior can be changed based on sales incentives or things that are happening within the company that don’t truly represent demand. For example, the sales team could be pushing products into the market despite customers’ lack of interest, or conversely there may be unfulfilled demand. These shifts in consumer behavior may be small but the bullwhip exaggerates their effect the further you go up the supply chain, because each vendor along the chain from end-consumer to raw materials supplier, has greater observed variation in demand, and thus greater need for safety stock.
In short, demand latency has increased because supply chains have become more fragmented, buffered, and risk-averse — while real consumer demand has grown more volatile and less visible to upstream partners.
It is now a matter of planning the procurement system and the supply management system outside-in and then look at the true demand to minimize the latency. The future of decision support is looking at the end-to-end supply chain in terms of market-to-market, looking at what is happening in the channel to drive bidirectional orchestration. This takes different forms in different industries. In heavy industry, the channel signal is typically an internet of things signal, telling managers how components are performing. Instead of buying parts the traditional way, they use predictive maintenance to buy based on the health of the equipment. Food manufacturers need to look at signals such as the price of commodities and patterns of consumer demand to plan their buying based on outside-in bidirectional orchestration.
Procurement must adjust accordingly. We have to think about how we measure value in the supply chain; according to Lori’s research, only 29 percent of companies can easily calculate the total cost of the commodities they are buying. And for a number of reasons. Total cost of ownership (TCO) is cross-functional, time-based, often qualitative, and poorly supported by existing systems. It’s like trying to judge the cost of owning a car without factoring in fuel, insurance, breakdowns, or resale — although in a supply chain, the multidimensionality is far greater.
She found that the best combination of metrics to evaluate total cost is a balanced scorecard of revenue per employee, operating margin, inventory returns, and return on capital employed. What this for procurement is that instead of looking at functional cost or purchase price variance we must look at total cost as it translates to margin and work more seamlessly across operations. That’s difficult in today’s organizations where our systems automate the functions while ignoring the collaboration needed to make better decisions together across source, manufacture and deliver.
To ensure better market-to-market decision making, a design layer needs to be added to sales and operations planning to determine what are the best levers to pull from market to market. These levers could be alternate sourcing, alternate routing, changes of push-pull decoupling points, or even changes to the product portfolio. Our thinking is still siloed. For example, category management in procurement means something completely different to category management in the channel. Only by getting a market-to-market overview can we begin to speak the same language and identify where there is scope for optimization.
The role of technology
Technology can help fix these issues but not if we simply slap technology on old processes—on old, siloed thinking. You need to start by getting all the right people in the room at the same time and think about what has happened in recent times. What are the challenges and where do they arise from sourcing through manufacturing and delivery? These cross-functional discussions can identify where they succeeded and where they failed. More often than not the solution lies in this bidirectional orchestration between procurement and operations. Procurement systems need to align better with what is happening in the demand streams—and an organization may have several (eCommerce, spare parts, retail partners, wholesale, B2B marketplaces etc.) Demand streams have become increasingly difficult to measure and we’re not managing the bullwhip effect, which has been magnified by recent developments and events, and can be unleashed by many factors, such as lack of visibility across the supply chain, forecasting errors, batch ordering (instead of smaller, regular replenishment), price promotions or discounts that distort true demand, lead-time delays that cause over-ordering “just in case”, and order amplification due to fear of stockouts.
To deal with this, procurement systems need to ingest downstream sales data (POS, channel activity, customer orders) in near real-time — not just historical forecasts. At the same time, sales teams need visibility into upstream constraints such as supplier lead times, inventory positions, and risk exposures. This implies a need to build integrated platforms or middleware that enables bidirectional orchestration: sales informs supply, and supply informs sales.
This is also important because of our social need to be more sustainable. If we don’t manage demand streams effectively, we make the wrong goods at the wrong time and they end up in landfill or sitting for years in a warehouse we have to heat and cool.
Vast amounts of information are generated across the value chain — from customer orders and inventory levels to supplier performance and logistics timelines — but much of it remains siloed, outdated, or inaccessible to decision-makers. Without timely, accurate, and shared data — including unstructured information and external feeds such as weather patterns, crop yields, or transportation disruptions — procurement teams can’t anticipate demand shifts, and sales teams operate blind to upstream constraints. This disconnect not only amplifies volatility but also prevents the kind of coordinated, demand-responsive planning needed to mitigate the bullwhip effect.
Schema-on read
Because traditional supply chain and procurement systems often rely on structured, internal data — such as forecasts, orders, and supplier records — they tend to be inherently rigid. You can only analyze what you’ve already defined as important enough to capture in structured form. By contrast, modern procurement platforms are increasingly adopting schema-on-read approaches, which allow organizations to ingest a wide variety of data sources — including social media, weather feeds, IoT sensor data, drone footage, and third-party risk ratings — without needing to predefine how each field fits into a fixed schema. This makes it far easier to respond to new and unexpected signals, such as sudden shipping disruptions or shifts in consumer sentiment. Instead of relying solely on preconfigured reports, procurement and sales teams can query data on demand, according to current priorities — enabling far more agile and responsive decision-making. Additionally, schema-on-read architectures are well suited to support machine learning and advanced analytics, which depend on diverse and expansive data inputs.
In short, emerging artificial intelligence technologies offer more than just opportunities to optimize existing processes — they invite us to rethink how we orchestrate information flows bidirectionally, from sourcing and manufacturing through to logistics and last-mile delivery. By reducing demand latency and improving responsiveness across the value chain, we can reshape business models to be not only more financially resilient, but also more environmentally sustainable.
To be continued…
