Automotive suppliers and navigating fickle customer demand

Don’t lose the assumptions and errors you make along the way.

Don’t lose the assumptions and errors you make along the way. A lot goes into every plan adjustment. The reasoning, assumptions, and documentation are valuable intel for the next time you’re faced with a decision. Use the knowledge from each adjustment to continually fine-tune confidence intervals and help minimize your forecast errors going forward.

Anyone can document assumptions in Excel, PowerPoint, or email, but an enterprise tool will come in handy to properly disseminate and control information if you want to be serious about mining, sharing, and learning from previous assumptions. A systemized workflow can also help you avoid sharing disinformation and inefficient processes.

Reinvigorate supplier collaboration

Reinvigorate supplier collaboration and advocate for increased visibility into BOMs for critical or new components. Obtaining visibility across your supply chain, from finished goods to raw materials and everything in between, is difficult. On top of that, collaborating with suppliers to consistently share forecasts and capacity reservations is a well-known leading practice but does not occur consistently in the automotive supply chain. To gain some traction, focus on areas that you know are problematic—those 10 SKUs that have a lead time of six months or more, for example, or the brand-new category of parts to support BEV production. The most challenging spaces often belong to suppliers, however, as a lot of the constrained materials for autos are constrained everywhere. To combat this, identify shared benefits for your suppliers and partner with them to access BOM data. Then, extend your modeling tool to be on alert for supply risks that have the highest impact to your business.

Shared access benefits both parties. The more reliable your forecasts are to your suppliers, the better they’ll be able to manage their own production flow and inventory costs. You may even be in a position to finance shared commodities between you and your supplier if you have more volume leverage. This insight into BOMs for constrained materials can be a win-win.

Establish templates and develop response playbooks to common what-ifs.

Establish templates and develop response playbooks to common what-ifs. Swings in customer demand happen all the time. Be as proactive as you can be by planning for frequent course corrections. If you asked your planners directly about the most common customer demand changes, shipment delays, or raw material shortages, for example, they would likely tell you that it’s been a recurring issue. Take those scenarios, build out the option analysis, and quantify the trade-offs (inventory, capacity, etc.) and the previous management decisions that were made for them. Save these as a playbook so that, the next time it happens (and it will), you can pull the playbook, refresh the data, and have a solid starting point around what action to take.

Debunk common myths with tech-enabled analytics.

Debunk common myths with tech-enabled analytics. Rely on quantitative methods to establish inventory policies. It’s not possible to set service levels at 100 percent consistently; just because a product is shipped “on-time and in-full” doesn’t mean you should aim for 100 percent either. There are exceptions to why it happens, when it happens—for example, you might have fulfilled every customer demand on-time and in-full last month, but only because you received approval to spend three times the typical rates to expedite the shipment by three days. Many companies, including auto suppliers, have a difficult time separating emotional reaction from statistical proof when setting 100 percent service level targets. Under a normally distributed (bell curve) demand, three standard deviations on both sides of the average will still only give 99.99 percent coverage under the curve. In other words, to achieve 100 percent service level statistically, you would need an infinite (and unrealistic) amount of inventory.

Parameterize wherever you can so machines can actually learn.

Parameterize wherever you can so machines can actually learn (and not just process an input). It’s likely your leaders incorporate IHS Markit (S&P Global) data or other external market intelligence into your planning process to enhance accuracy. But do you still struggle with maintaining and optimizing pro-rata factors, vehicle models to SKU allocations, or other variables to externally forecast the components you produce? When was the last time your conversion factor was updated, and do you believe it’s an accurate representation of the latest market trends?

Planning parameters (such as forecast disaggregation factors, replenishment lead time, and more) are dynamic in nature but often used as static inputs to formulate a plan. With artificial intelligence/machine learning (ML) on the tip of leading organizations’ tongues, it’s time to “put the robots to work” and get your return on investment (ROI). If you’re looking for somewhere to start, then begin with ML. Employing a simple ML program can help you analyze and predict the most accurate lead time based on past deliveries, current port congestion, and other dynamic variables. Your planners won’t need to waste time manually forcing a static measure to act more dynamically.