Forecasting, Reserved Capacity, and Safety Stock for Recurring Supply

Caught between stockout and expiry

Recurring supply is a balancing act with a failure on each side. Order too little or too late and you stock out, stalling the work that depends on the material. Order too much or too early and you carry inventory that ties up cash, fills cold storage, and may reach its retest or use-by date before it is consumed. Peptides make the second failure sharper, because material with a finite stability window cannot simply be held indefinitely as a buffer.

There is no single reorder number that resolves this for every material. Demand variability, lead time, shelf life, and the cost of a stockout differ from one peptide to the next, and a formula that fits a fast-moving, forgiving material will misfire on a slow-moving, short-dated one. The useful move is not to find the universal equation; it is to reason about the levers, forecasting, reserved capacity, and safety stock, in the specific conditions each material presents.

Forecasts you can act on

A forecast is only useful if it is expressed in the units and horizon that drive an order. Translate expected consumption into quantity per period over a horizon that comfortably exceeds your supplier lead time, so an order can be placed before demand becomes urgent. A forecast whose horizon is shorter than the lead time cannot prevent a stockout no matter how accurate it is.

Carry a sense of the forecast's uncertainty, not just its central estimate. A steady, predictable draw and a lumpy, campaign-driven draw call for different buffers even at the same average. Research on pharmaceutical demand forecasting shows that incorporating supply-chain signals and understanding variability improves inventory decisions more than chasing a single point estimate, and the same logic applies at program scale.

Revisit the forecast on a rhythm rather than treating it as fixed once set. Early in a program the estimate is largely assumption; as real consumption accrues, replace assumption with observation and let the buffer shrink or grow accordingly. The discipline is to change the plan on evidence at a scheduled review, not to react to every individual order, which only adds noise to a system you are trying to stabilize.

Freeze the near-term window

Constant re-planning is its own source of instability. A frozen window, a near-term period during which the plan does not change, gives both you and the supplier something firm to execute against. Inside the window, quantities and dates are committed; outside it, the forecast can flex as new information arrives.

Set the length of the frozen window to the supplier's lead time plus the time you need to react to a problem. Too short a window and the supplier cannot rely on it; too long and you lose the ability to correct a forecast that has clearly moved. The window is a negotiated balance between stability for the supplier and flexibility for you, and it should be written down rather than assumed.

Reserve capacity, not just inventory

For material that is slow to make, holding inventory is not the only buffer available; you can also reserve the supplier's capacity to make more. A capacity reservation commits a slot in the supplier's schedule so that when you place an order, the lead time reflects an agreed queue position rather than whatever backlog happens to exist that month.

Reserved capacity trades a commitment for predictability. You may pay for or commit to a minimum against the reservation whether or not you fully use it, so it suits materials where lead-time certainty matters more than avoiding a modest carrying commitment. For short-dated material, reserving capacity can be more attractive than holding a large safety stock, because it lets you make material closer to when you need it instead of watching a buffer age on the shelf.

The reservation is only as good as its terms, so make them explicit. Define how much notice converts a reserved slot into a firm order, what happens if you under-use the reservation, and how far ahead the arrangement extends. A vague understanding that the supplier will fit you in is not a reservation; it is a hope, and it will fail in exactly the busy period when you needed the certainty most.

Reorder logic without a fake universal formula

Reorder logic answers two questions: when to reorder and how much. The when is governed by lead time and demand during that lead time, plus whatever safety margin you have chosen; the how much is governed by lot economics, shelf life, and storage. Rather than plug numbers into a borrowed equation, reason about each material's own constraints and set a reorder point and quantity that respect them.

Safety stock is the explicit buffer against variability in demand and lead time, and it should be sized to the specific risk, not applied as a flat percentage across everything. A material with volatile demand or an unreliable lead time earns a larger buffer; a stable, reliably supplied material earns a smaller one. Operations-research treatments of safety stock make the drivers explicit, and the discipline they encourage is stating why a buffer is the size it is.

• Reorder point: demand across the lead time plus a buffer sized to that material's variability.
• Reorder quantity: balanced against lot economics, shelf life, and storage capacity.
• Safety stock: proportional to demand and lead-time variability, not a flat percentage.
• Shelf life: a hard constraint that can cap how much it is sensible to hold at once.

Lot cadence and shelf-life reality

The rhythm of lots matters as much as their size. Many small lots keep material fresh and reduce expiry risk but raise per-lot cost and multiply the release and logistics effort. Fewer large lots are cheaper to make and handle but concentrate shelf-life risk and a single quality failure into a bigger event. Choose a cadence that fits both the consumption rate and the stability window.

Let shelf life act as a genuine ceiling. There is no point ordering a quantity you cannot consume before it dates out, however attractive the unit economics of a large lot look. Where demand is uncertain and shelf life is short, a smaller, more frequent cadence backed by reserved capacity is usually the more robust choice than a large buffer that risks aging out.

Exceptions and governance

Plans meet reality in the form of exceptions: a demand spike, a delayed lot, a failed release, an expedite request. Decide in advance how these are handled and who decides, so an exception becomes a defined process rather than a scramble. An exception path that names the trigger, the options, and the approver keeps a single disruption from cascading.

Wrap the whole arrangement in light governance: a regular review where forecasts, buffer levels, reservations, and exceptions are examined together, and where the parameters are adjusted as the program's demand pattern becomes clearer. A pharmaceutical quality system treats supply changes as controlled, reviewed events, and a recurring supply relationship benefits from the same habit, so the numbers evolve deliberately rather than drifting.

Keep the governance proportionate to the risk. A high-volume, forgiving material may need little more than a periodic glance at the numbers, while a short-dated, hard-to-make, critical material warrants a closer and more frequent look. The point is not to build a bureaucracy around every purchase; it is to make sure the materials that could actually stop the work are the ones getting the attention, and that the reasoning behind each parameter is written down where the next person can find it.