All postsJune 16, 2026

The hidden cost of changeovers

Two production plans can produce the exact same units, in the same week, on the same line — and cost wildly different amounts to run. The difference hides in the spaces between the runs: the changeovers. The cleandowns, the retooling, the sanitation, the allergen flushes, the lost minutes every time the line stops making one SKU and starts making the next. Get them under control and you hand yourself back capacity you didn't know you were spending.

Why changeovers are so easy to under-count

Changeovers don't appear on a per-unit cost line, so they stay invisible until they've quietly eaten your week. They surface indirectly — as a line that's "running all week" yet somehow always behind schedule, as overtime nobody can quite explain, as a due date that slips for no obvious reason. Every switch burns capacity you never get back, and a schedule that ping-pongs between products — a little of this SKU Monday, a little again Thursday — can spend more time changing over than producing.

The math is worth doing once, because it's startling. Say a line averages a 45-minute changeover and you run six of them a day. That's 4.5 hours — more than half a shift — gone before a single sellable unit is counted, every day, on one line. Trim that schedule to four changeovers and shave each to 30 minutes, and you've recovered 2.5 hours a day: roughly a full extra shift of capacity per week, with no new equipment and no new headcount. Nobody approved spending that time. It just leaked.

For CPG lines, the cost compounds in ways generic factories don't see. Wet-to-dry transitions, allergen sequencing, sanitation windows, label and film changes — a changeover here isn't just minutes, it's minutes at the worst possible time, standing directly between you and a ship date. Two changeovers you could have avoided is often the difference between a calm week and a weekend shift.

The discipline that fixes it has a name

Reducing changeover time is one of the oldest and most thoroughly studied problems in manufacturing, and it has a canonical method: SMED — Single-Minute Exchange of Die, developed by Shigeo Shingo while consulting for Toyota. The "single minute" is aspirational shorthand for single-digit minutes; the real insight is structural, and it comes down to one distinction.

SMED splits every setup task into two buckets: internal work that can only happen while the line is stopped, and external work that can happen while it's still running. The entire method follows from there:

Separate internal from external. List every task in a changeover and tag each one. Most teams discover that a third of what they do "during" a changeover never actually required the line to be down.
Convert internal to external. Stage the next tooling at the line before the current run ends. Pre-stage and pre-weigh materials. Pre-heat, pre-position, pre-kit. Every minute you move off the critical path is a minute of capacity back.
Streamline what's left. The internal work that genuinely remains — simplify it. Quick-release clamps instead of bolts, color-coded parts, a written sequence so nobody improvises, shadow boards so no one hunts for a tool.
Standardize and sustain. Write the new changeover down, time it, and make the time visible. What gets measured stops drifting back.

The Lean Enterprise Institute documents teams cutting changeover time by 50–80% with nothing more exotic than this separation, applied seriously. Faster changeovers don't just save the minutes on the clock — they raise the line's availability, the first lever in overall equipment effectiveness, and they make smaller, more frequent runs economical, which loosens the whole plan.

Sequencing is the planning-level lever — and the one teams skip

SMED is the floor-level fix. But there's a planning-level lever that's just as powerful and far more often ignored: sequencing. Group runs of similar product and the line changes over less. Batch a SKU's entire week into one slot instead of three and you've eliminated two changeovers outright — for free, before anyone touches the equipment.

A few heuristics that pay off immediately:

Run families together. Sequence by the attribute that drives setup cost — flavor, color, format — so each switch is a small step, not a full teardown.
Go light-to-heavy, allergen-free to allergen. Order runs so the direction of the change minimizes cleandown. Running clear-before-dark or no-allergen-before-allergen can turn a full sanitation into a quick flush.
Protect the dirtiest changeover. If one transition always triggers a sanitation window, schedule around it so you pay it once a week, not three times.

The catch is that sequencing fights against the other things you care about. Cluster too aggressively and you push a run past its due date. Spread too thin and you drown in setups. Honor every line constraint — minimum run gaps, crew availability, blackout windows — and the "obvious" sequence often isn't feasible at all. This is a genuine optimization with real tension in it, not a rule of thumb you can apply by feel.

Batch sizing is the other half of the trade-off

Sequencing decides the order; batch size decides how much per run, and it carries its own economics. Bigger batches amortize the changeover over more units — fewer setups, more output per hour of uptime — but they tie up cash in inventory and shrink your flexibility to react. Smaller batches stay nimble and cut carrying cost, but multiply the setups you just worked so hard to reduce. This is the economic order quantity trade-off in production form: setup cost pulling one way, holding cost pulling the other.

The key insight that ties the whole post together: the right batch size moves as your changeover cost moves. Drive setup time down with SMED and the economics tip toward smaller, more frequent runs — which means better service, less inventory, and more flexibility, all unlocked by the same work that cut your changeover time. Cutting changeovers doesn't just save minutes; it expands the set of good plans available to you.

A practical changeover checklist

Time your real changeovers — not the number you assume, the one on the stopwatch.
Tag each task internal vs external; convert everything you can to external.
Rank transitions by cleandown cost and sequence to pay the expensive ones least often.
Revisit batch sizes after every meaningful SMED win — the old sizes are now wrong.
Make changeover time a number you watch weekly, not a cost you discover after.

Where the software earns its keep

CPG Scheduler plans against this trade-off directly. It knows each line's changeover penalty, the minimum run gaps, the crew and blackout constraints, and the due dates — and it flags when a proposed sequence is quietly costing you capacity: a run that could be batched with another, a switch that buys you nothing, a cluster that's about to blow a date. You decide the schedule; it makes the cost of changeovers visible before you commit to them, instead of after, when the only thing left to do is pay it.

Same units, same week — but a schedule that respects changeovers can hand you back a full shift of capacity you didn't know you were spending.