Agile Demand-Supply Alignment – Part 1B

This article is an excerpt from the report Agile Demand-Supply Alignment — Part One: ADSA Elements and Examples.
A copy of the full report can be downloaded here.

In Part 1A of this series, we looked at an overview of Agile Demand Supply Alignment, including continuous planning, tactical vs. strategic adjustments, system agility, and dealing with undersupply vs. oversupply scenarios. Here in Part 1B, we look at the elements of ADSA.

Elements of Agile Demand-Supply Alignment (ADSA)

The following diagram shows the elements of effective agile demand-supply alignment capabilities:

Figure 1 – Elements of Agile Demand-Supply Alignment

Companies are at various stages of maturity in how they do all of these. As illustrated above and noted below, some of these phases run in parallel or iteratively back and forth — in particular Understand, Prioritize, and Decide.

Detect

Early detection of supply disruptions and demand volatility is extremely valuable. The earlier an issue is detected, the more options there are to address the problem, at a lower cost. Ideally, issues are predicted well before they happen, rather than being detected after they happen. Supply disruptions can be due to either production delays or to logistical delays which generally have two different sets of tools to deal with them. Our research showed that visibility into supplier’s production status is particularly poor at many companies. Some refer to this as a ‘visibility black hole.’

Figure 2 – Production Status Visibility Black Hole

Detecting Production Delays

The OEM, brand owner, or retailer often is not aware of delays in the supplier acquiring raw materials or hitting production milestones until the supplier fails to book a container by the date they were supposed to. Calling or emailing each supplier to constantly ask for updates can consume a tremendous amount of time and labor, while straining the relationship. Cultivating a trusted relationship is an important element of ADSA, particularly in Asian cultures where saving face and never saying ‘no’ is the norm and suppliers may be reluctant to deliver the bad news when things are running late. Some companies use a supplier portal for suppliers to update status daily throughout the production cycle. A portal does not magically solve the issue; it still requires training and constant reinforcement of the importance of timely updates.

Detecting Logistics Delays

For international shipments, our research found the poorest visibility was typically from the factory gate to the port. Once the shipment has arrived at the origin port, logistic status updates are usually provided by the 3PL¹ or forwarder handling the shipment. Larger firms who handle the importing themselves may get updates directly from the carriers. In either case, there is a lot of room for improvement in getting more timely status updates, and particularly in predicting delays.

For domestic shipments, there are a number of tracking networks that have increasingly broad coverage, especially for shipments by truck. These take advantage of GPS on the vehicles, driver apps, and other mechanisms for getting status updates. Many of the main tracking networks have broadened out to multi-mode tracking, trying to solve the international problem as well.

Detecting Demand Volatility

Accurately detecting demand volatility early on² can be devilishly hard. Some product categories, such as fashion apparel, have high mix-complexity, meaning the forecaster is trying to figure out exactly what mix of size-color-style will be needed at each location across a network of selling locations (retail stores and ecommerce fulfillment centers). When the mix is wrong, people come into the store (or onto the website), don’t find the size, color, and style they want, and that sale is often lost. Manufacturers (and their suppliers) who sell to distributors and retailers face the further challenge of lack of direct visibility into end customer consumption. They only see a stream of orders from their immediate customer, which can contain all kinds of distortions, such as forward buys (potentially for a promotion or for other reasons), or a slowdown in orders for reasons other than a slowdown in end consumption.

Thereby, detecting demand volatility can be broken down into two parts: 1) knowing the actual end-consumption and current inventories across the chain, accurately and in a timely manner; and 2) using that current consumption data and other casual factors to see exactly (preferably by SKU-location) how current and near-term consumption deviates from your forecast. Using modern demand sensing software, with the right data, a much more accurate picture of near-term demand can be created.

Challenges for Suppliers

Ideally, all companies in the multi-tier supply chain would like to know the actual end consumption (not just the order stream from their immediate customer), as well as how much inventory is being held at all the intermediate locations and tiers between them and the end customer. That data would enable them to net out demand and supply to understand if there is the right amount of inventory in the supply chain (already delivered, in-progress, and planned) to meet demand. For the company at the downstream end of the chain, where consumption happens (such as a retailer, hospital, construction company, dealers, etc.) this is less daunting than for upstream suppliers. A retailer needs to ensure their POS data and inventory counts across their network are accurate and timely. A hospital needs accurate and timely data on stockroom and supply room inventory and withdrawals. A construction company is generally going to take a project-driven approach to assessing demand.

For suppliers the end-consumption visibility challenge is greater, encompassing multiple tiers. In theory, Collaborative Planning, Forecasting, and Replenishment (CPFR) could at least in part help solve this problem. However, in spite of being around for over 20 years, CPFR has yet to be widely adopted. There are services and software that help brand companies to consolidate POS, syndicated data, and casual factors to get a handle on end consumption. There are also software and networks that provide visibility into multi-tiered distribution channels inventory and sell-through. In general, however, lack of visibility into end consumption and downstream inventory is a big weak spot for brand owners and suppliers.

From Manual to Automated Detection

Smaller companies and those with less advanced process and systems typically rely on manual methods for detecting deviations from plan. This could include sending emails and making phone calls to suppliers and 3PLs to assess the status of orders and shipments, scanning Excel spreadsheets full of mind-numbing details on actual sales, inventory levels, shipment status, and production status. Needless to say, this is slow, labor-intensive, and error prone. Issues needing attention are missed or discovered too late in the game. Companies with more advanced approaches use various automated systems to monitor the status of their supply chain, demand consumption, and inventory levels. There are many variations and levels of sophistication in these systems, which will be discussed further in Part Two of this paper.

Understand

Types of Information Needed

When a deviation from plan (demand or supply) occurs, information is needed to understand the current situation, assess the impact, understand alternatives, and make decisions. Unfortunately, that information is usually scattered all over the place and may be stale. Different information is needed for different types of deviations and causes. Examples of the types of information that may be needed are shown below in Figure 3.

Figure 3 – Very Broad Range of Sources and Information Needed to Understand Disruptions and Deviations

Gathering all the information needed to properly understand the deviation can be very time consuming. Thankfully, there are solutions that help automate the gathering of the information. In particular, control tower solutions attempt to bring together information from many sources about issues that have been detected. Furthermore, it is important that the information is current and correct. Out-of-date or inaccurate information leads to misunderstanding of the actual situation on the ground. Near real-time situational data, as well as data accuracy tools and services, help provide a valid and current understanding of the situation.

Prioritize

Focusing Resources on the Most Consequential Disruptions and Deviations

As a business grows, it becomes increasingly difficult for them to gather comprehensive information about every disruption or deviation from plan that is detected. Furthermore, not every deviation needs to be addressed. In fact, the majority of them do not have significant enough consequences³ to justify actions or changes to the existing plans. An intelligent organization is able to prioritize the most consequential disruptions and deviations and apply their resources to them.

Sometimes a single disruption (such as congestion at a DC or a fire at a plant) impacts dozens, hundreds, or thousands of orders and shipments. In those cases, it is important to understand the root cause, and not just look at each delay separately. Hence, Prioritize and Understand phases are often intertwined and iterative. A certain amount of information gathering is needed first, to be able to prioritize and shortlist which issues to investigate further. Then more information can be gathered to do further prioritization and decide which of those issues should be focused on first. Control towers and autonomous supply chain platforms are developing software algorithms to help businesses prioritize issues as they arise.

Decide

Identifying Potential Solutions to the Disruption or Deviation

As soon as the highest priority disruptions and deviations have been identified, various options for solving the problem should be considered and the best course of action selected. Ideally these are cross-functional efforts, since different parts of the organization can bring different solutions to bear. For example, a transportation manager may reflexively turn to expediting a shipment as the solution when there may be more creative ways to deal with a pending shortage that could be discovered if inventory, production, and procurement managers put their heads together. When this happens, multiple potential solutions to the problem may be brought to the table such as A) pay for expedited production at an alternate supplier, B) pay for expedited transportation with the existing supplier, C) move inventory from another location, or D) let the item run out of stock.

Solution Impact Analysis

When the team or person is faced with multiple potential solutions, ideally they will analyze and compare the impact of each course of action, to select the best one. This may include the impact on sales/revenue, profitability, customer satisfaction and other key metrics. Some people use ad hoc analytic tools to do these what-if comparisons, but the job of pulling together all of the data and maintaining the model can be difficult. Furthermore, it is easy to introduce errors into home-grown models. Advanced control towers and autonomous supply chain platforms attempt to make the job of comparing different potential solutions easier by pulling together all of the needed data and analytics to compare the outcomes side-by-side.

Control towers and autonomous supply chain platform providers are also investing heavily in creating AI/ML⁴ engines that learn by observing the actions taken under various circumstances and recording the outcomes of those various actions. These engines can then start to build internal models, looking for patterns, and recommending actions based on the current situation. In theory, they will get better and better at prescribing the best course of action over time.

Optimizing Globally

Ideally, the global impact of different resolutions is considered. Most planning engines optimize within some local domain or functional area, such as the transportation or inventory needs of a single firm, without regard for the impact outside of its domain. However, a resolution to an issue that is optimal for transportation may overrun the capacity of the DC (such as having too many arrivals at or near the same time). Similarly, a fix that optimizes inventory for one company may place an even greater inventory burden on a trading partner. The various issues, resolutions, and supply chain elements are all interconnected. Decisions about which solution is best should be made within a unified global framework. Global network-wide optimization generates greater value and performance for the supply chain as a whole, compared with locally optimal solutions.

Act

Executing the Selected ‘Best’ Solution

Once a course of action has been decided, it needs to be executed. This may involve changes to production plans, expediting a shipment, transferring inventory, or other actions. Control towers may package together all of the information needed to ensure precise instructions are given to the person or service provider taking those actions. More advanced control towers and supply chain platforms are attempting to use direct API connections to execution systems to drive those changes without human intervention. In a few cases, the control tower solution provider also provides the executing applications, which usually results in tighter, more robust integration of decision making and execution.

Monitor

Ensuring Execution Stays on Track

Once action has been initiated, progress should be monitored. In many cases, this is simply part of the normal process of monitoring execution of production and logistics. In other cases, the action may involve a mini-project to fix the problem. For example, if the disruption is caused by a quality issue, then there may be a mini-project created, with milestones such as diagnose root cause, propose fix, test fix, implement fix/rework, and release the quality hold. As with the other phases, automating the monitoring of progress allows skilled supply chain professionals to spend their time on higher value problem-solving activities.

ADSA Maturity Progression

Companies become better at agile demand-supply alignment by investing in their underlying processes and systems. Automation and intelligence can help employees become more productive and effective. Many if not most companies still do things manually. Just automating the grunt work of collecting data and monitoring supply chain activities can free up skilled human capital and shorten time-to-detect and time-to-act. The ability to automatically gather all the right information together helps teams make quicker, more informed decisions.

For example, consider potential steps on the maturity journey for detecting production delays at suppliers’ factories:

1. Manual — sending emails and making phone calls to the supplier, asking the 3PL or booking agent whether a container request has been made yet.
2. Supplier self-service — suppliers update their production status on a supplier portal.
3. Auto-monitoring — alerts are generated when the supplier reports a problem, or schedule deviates beyond a threshold.
4. Instrumented factory — supplier’s factory is instrumented, so production steps are automatically and continuously monitored and communicated machine-to-machine.
5. Predictive – I/ML software monitors supplier portal data, factory instrumentation data, and other data sources⁵ to detect patterns and predict upcoming delays or disruptions.

Each of the ADSA phases has analogous levels of maturity or capabilities that companies acquire.

In the Next article of this series, Part 1C, we examine examples of specific ADSA capabilities including In-Season Re-ordering, Agile Compliance with Customer Mandates, Channel and Market Flexibility, and Leadtime Reduction.

________________________________________

¹ 3PL = Third Party Logistics provider. — Return to article text above
² i.e. detecting deviations from forecast well before they happen. — Return to article text above
³ If a shipment leaves the factory late but will still make it to the port in time for the sailing, no action is needed. If production is behind schedule or demand is surging, but there is enough safety stock to cover the shortfall, there is no need to take action. In other cases, there may be a shortfall, but the consequences of that shortfall are small enough that taking corrective action doesn’t make sense. — Return to article text above
⁴ AI/ML = Artificial Intelligence/Machine Learning. — Return to article text above
⁵ Other data could include on-time delivery rate for the supplier, industry-wide material shortage indicators, and other data that might be useful in forecasting delays and disruptions. — Return to article text above

To view other articles from this issue of the brief, click here.