Most plants do not have a handover problem. They have a problem deciding what a handover is for, which is why the question of how to improve shift handover in manufacturing almost never turns out to be a question about meetings, forms, or discipline.
I timed a shift changeover once, out of curiosity. Four minutes. Three of them were about the weather and who was covering Saturday. The outgoing lead said “runs fine, nothing to report,” and walked to his truck. Nobody was being careless. Nobody was cutting corners. The handover took four minutes because nobody in that building had ever written down what those four minutes were supposed to carry.
It is a question about which knowledge has to survive the gap between one crew leaving and the next one starting, and who decided that.
Why does the same job produce different results on different shifts?
Because each shift is running a slightly different procedure, and none of the differences are written down.
The SOP is identical. What differs is everything layered on top of it: that the second heat ran twelve degrees cool, that someone backed the feed rate off by hand at two in the morning and never put it back, that the left spindle has been making a noise since Thursday and days shift knows to listen for it. Each of those is a real adjustment made by a competent person. None of them is in a document. All of them expire at the end of the shift.
Multiply that across three crews and you get shift-to-shift variability that no amount of retraining fixes, because the training is not what diverged. The running knowledge diverged.
How to improve shift handover in manufacturing starts with what actually goes wrong
The failure is almost never that people refused to talk. It is that nobody ever specified what had to be said.
The UK Health and Safety Executive is direct about the stakes: “Many accidents have occurred because of failure of communication at shift handover, the majority of these involved planned maintenance work.” The Cullen Report into Piper Alpha found the same thing at the center of that disaster, a failure of information transmission across a shift change.
The clearest diagnosis I have read comes from the human-factors analysis of Buncefield published by John Wilkinson and Ronny Lardner of The Keil Centre in Chemical Engineering Transactions. Reviewing the terminal’s arrangements, they found that shift changeover “was inadequately structured, variable and informal,” that “the key information required for communication at handover had not been adequately determined,” and that “the verbal passing on of information was seen as the default position,” instead of writing it in the log, whenever time or events were pressing.
Read those three findings again with your own plant in mind. There was a handover. There was a log. There were experienced people doing it face to face. What was missing was the decision about what had to cross.
Which handovers carry the most risk?
Wilkinson and Lardner name the higher-risk handovers explicitly, and the list is worth pinning to a wall: after a lengthy absence from work, such as a crew change or a long rest period; between experienced and inexperienced staff; during a plant or process upset; when maintenance activity spans a shift handover; and when safety systems are overridden or otherwise unavailable.
The second one is the one that should get your attention. Every plant running a hiring push is producing that exact handover several times a day, usually without noticing. A twenty-year operator hands to a nine-month operator. The twenty-year operator communicates at the level of detail he would want, which is thin, because he can reconstruct the rest from memory. The nine-month operator cannot. Nothing looks wrong until something is.
Why a digital logbook does not fix shift handover
Because a logbook is a container, and the problem is that nobody agreed on the contents.
The category will sell you an electronic shift report, a tablet at the line, a template with mandatory fields. Those are genuinely better than a paper book somebody left in a drawer. But if the fields were designed by an engineer at a desk rather than derived from what the crews actually need to know, you have digitized the wrong list, and you now have a permanent record of an incomplete handover.
Wilkinson and Lardner reach the same conclusion at the industry level: “it isn’t more research or guidance that is required but effective implementation of existing well-established guidance.” The knowledge of how to do this well has existed for thirty years. It does not get implemented because implementing it requires somebody to sit with the crews and extract what actually has to cross, and that work has no owner.
Run this on your floor this week: the five-handover inventory
This costs you nothing but a clipboard and about ninety minutes, spread across a week.
Step one. Stand at five handovers, across all crews. Not one, five, because the whole point is to catch the variation between them.
Step two. Write down every piece of information passed out loud that lands nowhere in writing. Just the item, not a judgment. “Second heat ran cool.” “Feed rate backed off.” “Left spindle noise.”
Step three. Beside each item, mark one letter. E if the item is about equipment condition. P if it is about process state, meaning something is set differently than the standard says. A if it is an active problem being handed to the next crew.
Step four. After five handovers, count. The items that show up more than twice are not incidents. They are your undeclared handover standard, and they belong in a written one.
Step five. Now do the harder read. Look at the items that appeared once, from one crew only, and ask whether the other crews had the same condition and simply did not mention it. That is where your variability is hiding, and that is the list that will not appear on any template you buy.
The HSE’s own principles say a handover should be face to face, two-way with both people taking joint responsibility, done in both speech and writing, “based on an analysis of the information needs of incoming staff,” and given as much time and resource as necessary. The inventory above is that analysis, done cheaply, by you.
What to do once you know what has to cross
Once you have the list, you have a documentation problem, and that is where most plants stall. The list is accurate on the day you build it and stale within a month, because the floor keeps moving.
At SenseiLab this is exactly where a 30-day SOP Sprint starts. Operators, supervisors and leads talk directly to our AI agent about what actually happens at changeover, the agent structures it into Living SOPs, and the standard updates as the process changes rather than aging quietly in a binder. The five-handover inventory is the free version. The Sprint is the version that stays true in month seven.
Your handover is four minutes. Your shift is twelve hours. The difference between them is not a communication problem, it is an undocumented-knowledge problem wearing a communication problem’s clothes. It will keep costing you a shift at a time until somebody decides what has to cross the gap.
Frequently asked questions
Why do my shifts produce different quality with the same SOP? Because the SOP is not the whole procedure being run. Each crew layers undocumented adjustments on top of it, such as a feed rate someone changed by hand or a machine condition only one crew knows to listen for. Those adjustments expire at the end of the shift, so each crew rebuilds its own version. The document is identical; the running knowledge is not.
How long should a shift handover take in manufacturing? There is no universal number, but Wilkinson and Lardner note that a handover on a 12-hour shift typically may last up to 30 minutes, and that shorter periods can be enough if the handover is well structured and the plant’s current status is simple. The failure mode is not a short handover. It is a short handover with no agreed content.
What information should be included in a shift handover? Equipment condition, process state where anything is set differently from the standard, active problems being passed on, and any safety system that is overridden or unavailable. The HSE’s guidance is that the content should be derived from an analysis of the information needs of the incoming staff, not chosen by whoever designed the form.
Will a digital shift handover tool fix shift-to-shift variability? Only if the fields in it came from the floor. A digital tool improves the record and the retrieval. It does not decide what belongs in the record, and a well-designed tool filled with the wrong list produces confident, incomplete handovers.
Which shift handovers are the riskiest? Handovers after a long absence or crew change, handovers between experienced and inexperienced staff, handovers during a process upset, handovers where maintenance work spans the changeover, and handovers where a safety system is overridden. The experienced-to-inexperienced handover is the one most plants are creating daily without tracking it.
About the author. Diego Echenique is CEO of SenseiLab and has spent more than 20 years in manufacturing operations, launching plants, leading operations, and running Lean and Six Sigma transformations across automotive, mining and heavy industry. SenseiLab captures the knowledge that runs a plant and turns it into Living SOPs. LinkedIn
