On a contract packaging line, changeover time is the interval between the last good unit of one job and the first good unit of the next, measured at rate. That definition matters, because plants that stop the clock when the line starts moving hide the ramp, and the ramp is often the longest part. Serious changeover reduction on contract packaging lines starts with measuring that full interval honestly, then separating the work that truly requires a stopped line from the work that only happens on a stopped line out of habit.
Why contract packagers change over more than anyone
A brand-owned plant may run two or three SKUs for a week at a time. A contract packager runs someone else's portfolio, and the order book is built from short runs, promotional packs, seasonal artwork, and trial quantities. It is common to see five to fifteen changeovers a shift on a filling or labeling line, each one carrying its own cleaning requirement, its own label and insert reconciliation, and its own customer specification. The math is unforgiving: if a line changes over ten times a shift and each changeover runs forty minutes, more than half the shift is not producing. Cutting ten minutes off each one buys back a run's worth of capacity without buying a machine.
There is a second reason changeovers matter more here. Contract packaging is priced per unit against a quoted rate. When changeovers run long, the margin on the job erodes quietly, and it usually erodes on the jobs that were already thin. Plants that do this work at volume tend to find that their least profitable customers are not the ones with the lowest price, they are the ones with the most format changes per case shipped. You cannot see that pattern without changeover data tied to job numbers.
Where the minutes actually go
When plants time changeovers with a stopwatch and a notebook for two weeks, the breakdown is usually consistent, and usually surprising to the front office. The physical work of swapping change parts is rarely the biggest bucket.
- Waiting on materials. The next job's cartons, labels, caps, or inserts are not staged at the line. Someone walks to the warehouse, finds a partial pallet, and waits for a lift truck. This can be twenty minutes of a forty minute changeover.
- Waiting on release. Sanitation sign-off, line clearance, or a QA check that requires a person who is currently at another line. The line is clean and ready and nobody is allowed to start it.
- Hunting for tooling and settings. Change parts live in a cart that may or may not be where it was left. Machine settings for that SKU live in an operator's memory, a laminated card, or nowhere.
- Dialing it in. The line runs, makes scrap, gets adjusted, makes less scrap, and eventually holds. On labelers and case packers this ramp can easily exceed the mechanical changeover itself.
- Documentation. Artwork verification, lot code entry, first-article checks, and the paperwork that proves all of it happened. Necessary work, often done serially when it could be done in parallel.
Note what is missing from the top of that list. The wrench time, the actual swapping of format parts, is often the smallest and most improved-upon piece, because it is the visible part. Teams optimize what they can see.
Internal versus external work, applied honestly
The classic SMED distinction is between internal work, which requires the line stopped, and external work, which does not. It is a sixty year old idea and it still does most of the heavy lifting, because most plants have never applied it rigorously. Staging the next job's change parts, pre-kitting labels and cartons, pre-heating a sealer, verifying artwork against the customer specification, and printing lot codes are all external in principle. In practice they happen after the line stops, because nobody was assigned to do them before.
The honest version of this exercise takes a video of one changeover, watches it with the operators, and marks every minute as internal, external, or waiting. It is uncomfortable, and it works. Expect to find that a meaningful share of a stopped line's time is external work that drifted inside, plus waiting that belongs to no one. Converting even part of that requires a person or a role who owns pre-staging, which is an organizational change more than a technical one.
One caution: sanitation and allergen changeovers in food and beverage packaging have real regulatory floors. Wet cleans, allergen validation, and swab results cannot be compressed by scheduling cleverness. What you can change is whether the rest of the line's work happens during the clean instead of after it, and whether the release step waits on a person who is somewhere else.
Sequencing beats speed
The cheapest changeover is the one that does not happen. On a line running many formats, the order of the jobs determines how much change work exists in a shift. Running all the 12 ounce bottles together before moving to 16 ounce, grouping jobs that share the same closure or the same carton size, and moving from light to dark or from allergen-free to allergen-containing all reduce the total change burden without touching the machines.
Most contract packagers know this and still cannot do it, because the schedule lives in a spreadsheet that is rebuilt by hand every morning, and rush orders scramble it by nine. Sequencing for changeover similarity requires the scheduler to see, for every candidate job, which change parts and which cleaning level it implies. That is a data problem before it is a scheduling problem. The pillar guide to manufacturing scheduling software covers how plants get from a spreadsheet to a schedule that accounts for sequence-dependent setup, and why constraint-aware sequencing is the piece most ERP scheduling modules leave out.
There is a real tension here. Grouping by format lowers changeover time but can raise finished goods inventory and push out due dates for customers whose jobs got bumped down the sequence. The right answer is rarely pure grouping; it is grouping within the window that due dates allow. That tradeoff is exactly the kind of thing a person should be deciding with the numbers in front of them.
What to measure, and how to get the data without a clipboard
Three measurements carry most of the value. First, changeover duration from last good unit to first good unit at rate, per line, per changeover. Second, the reason code for the largest block of waiting inside it. Third, the from-SKU and to-SKU pair, so that sequence-dependent patterns become visible over a few hundred changeovers.
Getting this by hand is possible and it is how most plants should start. Sustaining it by hand is where it falls apart, because the person with the stopwatch gets pulled to a line issue and the data stops in week three. The durable version reads the state directly from the equipment. Harmony connects at the PLC, whether that is Allen-Bradley, Siemens, Omron, or Mitsubishi, over OPC UA or whatever protocol the machine speaks, so run state, rate, and reject counts are captured without an operator entering anything. The operator's contribution is the reason code, which takes a tap on a screen and is worth asking for, because a duration without a cause is not actionable. Harmony is software and hardware agnostic, so this works on a mixed floor where the filler, the labeler, and the case packer come from three different decades and three different vendors.
Where Harmony fits
Harmony's work on packaging lines is usually the unglamorous version of this. Forward-deployed engineers come on-site, connect to the line, and make the real changeover interval visible with its reasons attached, then use that history to propose job sequences that reduce total change burden inside the due date window. The AI proposes the sequence and a person approves it, because the scheduler knows things the data does not, like which customer will accept a day of slip and which will not.
The published pilot is $15 to $20K one time over 4 to 6 weeks, with working software by week three. That scope is deliberately one line or one value stream, not a plant. Customers in adjacent work include Mossberg, MoonPie, and Chattanooga Labeling Systems. If you want the industry-specific version of this, the contract packaging industry page covers the operational patterns we see most often in co-packing and co-manufacturing environments. We would rather show you a measured changeover on your own line than argue about the number in a slide.