Just-in-Time Is Not Dead. It Needs a Buffer Policy.

Another Shock, Another Obituary

Russia's invasion of Ukraine on February 24 has added a new kind of disruption to a supply chain that had barely begun to recover from the last one.

Ukraine has been a major source of semiconductor-grade neon: Reuters estimated that two companies in Odesa and Mariupol supply 45% to 54% of the world's output. On March 8 the London Metal Exchange suspended nickel trading after prices more than doubled in a matter of hours, briefly passing $100,000 a ton. Energy, wheat, aluminum and freight are all moving. For the third time in two years, manufacturers are being told that just-in-time is finished.

Just-in-time is not finished, but the last two years have shown that lean inventory without a deliberate policy for where to hold buffers is fragile.

Just-in-time is not finished, but the last two years have shown that lean inventory without a deliberate policy for where to hold buffers is fragile. Most plants never wrote that policy down. They ran lean everywhere because it looked efficient, and then, when disruption hit, they stockpiled everywhere because it felt safe. Neither is a strategy.

Lean Everywhere Was Never the Idea

Just-in-time was always about removing waste, and inventory that exists only to hide problems is waste. But the original thinking never said every part should have the same minimal buffer regardless of risk. A fastener available from a dozen distributors down the road and a controller with a single source on the other side of the world do not deserve the same treatment.

The problem is that in most ERP systems they get the same treatment by default. Safety stock is set by a blanket rule, or left at zero, or set once at implementation and never changed. The system has no idea that one part is a commodity and the other is a single point of failure.

Segment the Catalog by Risk

The work of a buffer policy is mostly sorting. Take your purchased parts and classify them on two dimensions: how much they matter and how exposed their supply is.

Importance usually comes from value and from what the part stops. A low-cost part that halts a line if it is missing matters more than its price suggests. Exposure comes from the number of qualified sources, the geography of supply, the lead time and its variability, and any concentration in raw materials, like the neon in chips or the nickel in stainless and batteries.

The parts that score high on both are where buffers belong: strategic safety stock, a second qualified source, longer purchase commitments, or all three. The parts that score low on both can stay as lean as ever. Most of the catalog sits in between, and that is where judgment and data both matter.

Put the Policy in the System

A buffer policy that lives in a slide deck changes nothing. It has to show up in the ERP as parameters that MRP and replenishment actually use: safety stock levels by item, reorder points, minimum order quantities, planning time fences and lead times that reflect reality.

That also makes the policy auditable. When someone asks why the plant is holding six weeks of a particular part, the answer is a documented decision with a named owner and a review date, not a buyer's instinct from a bad month. And when conditions improve, the review date is what brings the buffer back down, so emergency stock does not quietly become permanent.

Price the Buffer Honestly

Inventory is not free, and this year it is getting more expensive to hold. Carrying cost includes the capital tied up, the storage, the handling, the obsolescence risk and the insurance. The Federal Reserve raised interest rates in March for the first time since 2018 and has signaled more increases, so the capital part is going up.

That cost should be weighed against the cost of a stockout on the same part: lost margin, expediting fees, overtime to recover, penalties and the harder-to-measure damage to a customer relationship. For a critical part the comparison is usually easy. For the middle of the catalog it is a real calculation, and it deserves to be done with numbers rather than with the fear of the last shortage.

The Bottom Line

Just-in-time did not fail. Running every part at the same thin buffer, without regard to risk, failed, and so will stockpiling everything in response. What survives repeated shocks is an explicit buffer policy: parts segmented by importance and supply exposure, buffers concentrated where a stockout stops the line, the policy encoded as real parameters in the ERP, and every buffer priced and reviewed on a schedule.

At Cherry Street, this is the work we do. We help manufacturers segment their supply base by risk, turn the result into planning parameters their ERP can enforce, and model what each buffer costs against what it protects.