CDMO production scheduling: the challenges of shared capacity and multiple clients

Run a scheduling meeting at almost any CDMO and you'll hear some version of the same sentence: "we can hit the client's date, or we can hit the other client's date, but not both, and nobody told us that until this week."

That is not a scheduling failure in the usual sense. It is what happens when the core math of a CDMO's business, running multiple clients' products through shared capacity, meets a scheduling approach that was never built for that math.

A plant that makes its own products for its own demand has one master plan to worry about. A CDMO doesn't have that luxury. Every line, every piece of shared equipment, every changeover slot is pulled in several directions at once, by clients who don't talk to each other and don't particularly care whose capacity commitment gets squeezed to make room for theirs.

Five challenges show up again and again.

Client A's batch was promised for the 15th. Client B's urgent request came in yesterday and needs the same line. Both are contractually reasonable asks.

Somebody has to decide. That decision usually happens under time pressure, informally, by whoever picks up the phone first, rather than as part of a considered plan.

Running a longer campaign of one product reduces changeover losses and is the more efficient choice on paper. But a long campaign also means every other client waiting on that shared line waits longer.

The efficient schedule and the schedule that keeps everyone happy are frequently different schedules. Reconciling them by hand, batch by batch, doesn't scale past a handful of clients.

Sales and commercial teams commit to capacity slots well in advance, often before the detailed production schedule for that period has been built. By the time scheduling happens, those commitments are treated as fixed constraints. Reality (a delayed raw material, a failed batch, an equipment issue) can make honoring all of them simultaneously mathematically impossible.

In a CDMO handling multiple clients' products, potentially with different potency levels, allergen profiles, or cross-contamination sensitivities, the changeover sequence between two client batches isn't just a time cost. It is a hard compliance constraint.

Worst-case bracketing, MACO limits, the cleaning validation matrix, and the split between dedicated and campaigned equipment all narrow which sequences are permitted at all. A scheduler that treats changeovers as roughly interchangeable time blocks will eventually propose a sequence that isn't allowed, and somebody has to catch it by eye.

Clients increasingly want visibility into where their batch sits in the queue, not just a promised ship date. Providing that transparency across a shared, constantly shifting production plan is a different problem from running the plant efficiently.

Add a third client, and scheduling complexity doesn't grow by a third. It grows by something closer to the number of ways those clients' priorities can now interact, which is a much steeper curve.

A CDMO running five plants with a dozen active clients faces an order of magnitude more potential conflicts than one running two plants with three clients, even when the amount of production is only somewhat higher.

This is the point where informal coordination breaks down: a good planner who knows the plant, a shared spreadsheet, a weekly call. Not because anyone is bad at their job. The number of things that have to be held in someone's head has quietly outgrown what one person or one spreadsheet can track.

A few patterns show up in CDMOs that handle this well, whatever tools they use.

Capacity commitments and the detailed schedule stay connected. When sales-side capacity promises are made with visibility into what the schedule can absorb, rather than as an independent negotiation, far fewer conflicts surface later as surprises.

Changeover and compliance rules live inside the scheduling logic, not in a manual check afterward. If a sequence would violate a validated changeover requirement, that should be caught before it is proposed, not by someone reviewing the plan the night before.

Re-planning happens in minutes, not days. The plants that handle multi-client pressure best aren't the ones that never have conflicts. They're the ones where a conflict gets resolved and the downstream schedule updates quickly, instead of sitting unresolved for a week while people argue about whose client gets bumped.

Client-facing status is a byproduct of the real schedule, not a separate manual update. If someone has to translate the internal schedule into a client-facing update by hand, that update will lag reality and eventually be wrong.

The payoff is measurable. At a multi-line API site running commercial and CDMO production on shared equipment, scheduling both streams against one live model of the plant improved commercial lead time by 2.5%. The site reallocated that freed capacity to five extra CDMO batches a year. Replanning after a shop-floor resource issue moved from one to two hours of manual work to immediate. Those are customer-validated figures from an anonymized deployment.

Pick your two largest clients by volume and ask: if both had an urgent request land on the same day, for the same shared equipment, how would that get resolved today?

If the honest answer involves "someone would have to manually check three schedules and make a judgment call," that is not a people problem. It is a sign the scheduling layer isn't built for the multi-client reality the business runs on.

This is the specific challenge Bodhee Production Scheduling is built around for CDMOs: keeping Production, Quality Control, and Maintenance schedules synchronized in real time across shared capacity and multiple clients. It deploys alongside the ERP, MES, and LIMS systems already in place rather than replacing them.

If reconciling client priorities on shared lines currently depends on someone remembering every open commitment, that is usually the sign this layer is missing. The cost of that gap has a name and an economics of its own, which we covered in the silo tax.

A plant making its own products optimizes one master plan against one demand signal. A CDMO runs several clients' products through the same lines and vessels, with separate contractual commitments, separate due dates, and no shared view between clients. Every scheduling decision is a trade-off between parties who never negotiate with each other directly.

Commercial teams sell capacity slots months before the detailed schedule for that period exists. Those commitments then arrive at the planner as fixed constraints. When a raw material slips or a batch fails, honoring all of them at once can become mathematically impossible, and the conflict surfaces only when the schedule is built.

The decision itself is commercial, not technical. What scheduling can change is the quality of the information behind it: what each option costs in changeover time, downstream due-date risk, and knock-on effects for other clients. Making that trade-off visible in minutes, rather than after a manual check of three schedules, is the part software can own.

Only reliably when client-facing status is generated from the live schedule rather than compiled by hand. A manually maintained status report lags the plant by however long it has been since someone last updated it, which is why transparency commitments tend to erode as client count grows.

No. Scheduling sits between the ERP's order and materials view and the MES's execution record, reading the event stream from the floor and re-solving the sequence when reality changes. It complements those systems rather than replacing them.

Bodhee: Dynamic Manufacturing Scheduling

Bodhee delivers dynamic, constraint-aware scheduling for Production, Quality Control, and Maintenance on one live model of your plant.