Setup and adjustment loss is the production time lost to changeovers and the fine-tuning that follows. It is one of the six big losses and an Availability loss in OEE, and it runs from the last good part of the old run to the first stable good part of the new one, not from the last part to the moment the line simply restarts.
Most plants measure the changeover and forget the adjustment. They time the die swap, the format change, the recipe switch, and stop the clock when the line starts moving again. But the line that just restarted is not making good parts at rate yet. It is running slow, spitting first-piece rejects, and getting nudged toward the setpoint by an operator with a caliper. That dial-in period is the adjustment tail, it belongs to the same loss, and it is frequently longer than the changeover everyone was watching. This post defines the full loss, shows why the tail hides so well, and gives you a way to measure and shrink it.
What are setup and adjustment losses?
Setup and adjustment loss is all the time a line is not producing good parts at rate because it is being changed over from one product to the next. It has two halves that usually get treated as one: the setup the physical changeover, tooling swap, or format change, and the adjustment the dial-in, trial pieces, and speed ramp needed to get the process stable and in spec afterward. Both are Availability losses because during both the equipment is up but not making sellable product.
The classic six-loss framework, developed by Seiichi Nakajima as part of Total Productive Maintenance, groups setup and adjustment together for a reason: the adjustment is not a separate event, it is the tail of the same changeover. A setup isn't finished when the new tooling is bolted in; it is finished when the line is producing good parts at full rate. Anything before that, slow first passes, scrapped trial pieces, tweaking a temperature or a registration, is still setup loss wearing a different hat.
Where does setup loss really end?
It ends at the first stable good part produced at rate, not at the first restart. This is the single most useful correction you can make to how setup is measured. A line that finishes its mechanical changeover in twelve minutes but then spends twenty-three minutes dialing in registration, ramping to speed, and clearing first-piece rejects has a setup loss of thirty-five minutes, not twelve. If the stopwatch stops at "moving again," two-thirds of the loss vanishes from the record, and from every improvement conversation.
The adjustment tail hides in two places. Part of it looks like a run-rate problem: the line is technically running, just slowly, so it gets miscategorized as reduced-speed loss instead of setup loss. Part of it looks like a quality problem: the first-piece rejects and trial pieces get logged as scrap without anyone connecting them back to the changeover that caused them (they are startup rejects, another of the six losses, but they share the same root event). Pull those pieces back where they belong and the true cost of each changeover roughly doubles, which is exactly why changeovers that "aren't that long" quietly dominate a line's Availability.
Why is the adjustment tail often longer than the changeover itself?
Because the mechanical swap is a defined task and the adjustment is a search. Bolting in new tooling has a beginning, a middle, and an end you can practice and standardize. Dialing a process back into spec is open-ended: you make a part, measure it, adjust, make another, measure again, and repeat until it's right. If the correct settings aren't captured, every changeover re-discovers them from scratch, and the tail is as long as the operator's search takes that day. Two changeovers of the same product can differ by a factor of three purely on how fast someone finds the setpoint.
How do setup and adjustment losses map to OEE?
They hit Availability directly. OEE splits every loss across three factors, and setup and adjustment sit in Availability alongside breakdowns, the difference being that a breakdown is unplanned and a changeover is planned. Both are stopped time from OEE's point of view: the line was scheduled to run and wasn't making good parts. Because setup loss is one of the six big losses it also has a named countermeasure, and it is not "change over less often", it is to make each changeover faster and more repeatable so smaller batches stop being expensive.
That countermeasure is SMED and it works precisely on the two-part structure of the loss. SMED's core move is to separate internal setup (work that requires the line stopped) from external setup (work that can be prepped while the line still runs), then convert as much internal to external as possible and streamline what remains. Applied honestly, it also attacks the adjustment tail, by capturing the correct settings so the dial-in becomes a recall instead of a search, and by error-proofing the setup so the first piece is good instead of the fifth.
By the numbers. The reduction potential is large and well documented. Shigeo Shingo's Single-Minute Exchange of Die method was named for its target: cutting changeovers to single-digit minutes, under ten, from what were often hours. The gains come as much from standardizing the adjustment (recorded settings, first-piece checks, error-proofed fixtures) as from speeding the mechanical swap, which is why measuring the full loss, tail included, is the prerequisite to shrinking it.
How do you measure and reduce setup and adjustment loss?
Time the whole loss first, then split the changeover into work you can move, speed up, or standardize away. The sequence:
- Measure from last good part to first stable good part. Stop pretending the loss ends at restart. This one change usually doubles the recorded loss and is the honest starting point for every improvement after it.
- Separate the mechanical changeover from the adjustment tail in the record. You cannot tell whether to attack the swap or the dial-in until you know which one owns the minutes, see changeover time measurement for the timing discipline.
- Split setup work into internal and external. Anything that can be done while the previous run finishes, staging tooling, pre-heating, kitting materials, comes off the stopped-line clock immediately.
- Capture the correct settings for every product. Recorded setpoints, registration marks, and first-piece standards turn the adjustment from an open-ended search into a repeatable recall, which is where most tail time actually lives.
- Error-proof the first piece. A first-piece check and simple fixturing that only assembles the right way cut startup rejects, shrinking both the tail and the startup-reject loss that rides with it.
- Standardize and drill the changeover as a team, then re-measure the full loss. A practiced, documented changeover is the difference between a twelve-minute swap with a twenty-three-minute tail and a twelve-minute total.
What belongs to internal setup versus external setup?
The internal-versus-external split is the lever that shrinks setup loss without buying anything, so it is worth making concrete. Internal setup is work that can only happen with the line stopped, removing and installing tooling, threading new material through a machine, any task that physically requires the equipment idle. External setup is work that can happen while the previous run is still producing, staging the next tooling at the line, kitting the materials, pre-heating a die, printing the new labels, confirming the settings sheet. Every minute of work you move from the internal column to the external column comes straight off the stopped-line clock.
The reason this matters for the adjustment tail specifically is that a lot of the dial-in is really external work done late. If the correct setpoints, the first-piece gauge, and the registration reference are all staged and confirmed before the line stops, the adjustment collapses from a search into a checklist. Plants that never make the split treat the whole changeover as one undifferentiated block of stopped time, which is why their setups don't get shorter no matter how fast the crew hustles, they are sprinting through work that shouldn't have been on the stopped clock at all. Naming which tasks are external is usually the first afternoon of real changeover improvement, and it costs nothing but attention.
How does setup loss connect to the rest of the plant?
Setup and adjustment loss is the tax on flexibility, and it quietly sets your minimum economical batch size. When changeovers are long and unpredictable, planners lengthen runs to amortize them, which inflates inventory and lengthens lead time; when changeovers are short and repeatable, small batches become affordable and the whole plant gets more responsive. That makes setup loss one of the highest-leverage entries on the manufacturing KPI stack, it touches Availability, scrap, inventory, and schedule flexibility all at once.
The reason the adjustment tail stays hidden is almost always the record. A stopwatch that stops at "line moving" and a scrap log that never links first-piece rejects to their changeover will show a short, tidy setup loss that bears no relation to reality. Plants that capture changeover start and stop, run rate on the ramp, and first-piece results at the station, the way Harmony turns paper changeover and quality logs into live records feeding root-cause analysis (see the platform or the CLS field story), finally see the full loss, tail included. And seeing the tail is the whole game, because the minutes you never measured are the minutes you were never going to get back.