The same plant on every shift

In 2020 our maintenance review found that nineteen of sixty-four technicians could retire within five years, including six of the eight electricians who understood the controls on Erie's folder-gluers. Outside contractors already cost us US$2.6 million a year, and the experienced technicians we might hire were being offered sign-on bonuses by every manufacturer within thirty miles.

The chief financial officer proposed contracting more maintenance and paying our own bonuses to attract experienced hires. That would have been quicker. My objection was that it bought capacity without passing on knowledge: a retiring technician's understanding of a thirty-year-old machine would leave with them. I proposed a four-year registered apprenticeship in industrial maintenance with Tobermane Community College, paying apprentices their full wage for one classroom day a week. Alongside it I proposed a mentor role for technicians within eighteen months of retirement: a four-day week at their full hourly rate plus a US$3 an hour mentoring premium.

The unions at Mansfield and Erie wanted apprentice places filled by seniority, while supervisors wanted to choose. We settled on seniority among applicants who passed the college's entry assessment, which union officers helped to review. The board approved US$3.4 million over four years. We gave up the faster option and accepted a period in which mentors worked fewer hours than before.

Twenty-four apprentices started in 2021 and 2022. By mid-2025 seventeen had completed, three were still progressing after repeating a module and four had left. Eleven technicians took mentor roles before retiring. Contract maintenance spending fell to US$1.1 million in 2025, and operations now schedules a qualified technician on every shift at all four plants.

Maintenance succession has since become part of the annual workforce plan the board reviews. If I began again, I would start two years earlier and include Erie's electricians in the first intake. Two of them retired before the programme reached that plant, and we paid contractors for their knowledge for another year.

In 2021 first-year turnover on Mansfield's third shift reached 61 percent, against 24 percent on first shift, and its recordable injury rate was double the day rate. The cause was not mysterious. Under the seniority bidding rules in our agreement, new hires started on nights and moved to days as soon as their seniority allowed, so the least experienced operators worked the hardest hours with the thinnest supervision.

Our vice president of operations wanted to propose four-week rotating shifts in the 2022 negotiations, spreading experience across the clock. I drafted the proposal, then attended two meetings the union arranged for members to explain their objections. Rotation would have disrupted childcare, second jobs and sleep for several hundred people to solve a problem concentrated among about ninety. Senior operators had earned their day shifts and saw rotation as taking them away.

I recommended withdrawing rotation before it was tabled. The vice president disagreed, and the chief executive asked each of us to present the costs. We proposed instead to raise the third-shift premium from US$0.90 to US$2.40 an hour, pay a further US$1.50 an hour to senior setup operators volunteering for twelve-month night assignments, and move two of our most experienced supervisors to nights. The union's bargaining committee agreed to direct part of the first-year general increase towards the premium. The package cost about US$1.3 million a year, and we gave up the operating simplicity of one rotating pattern.

Fourteen senior operators volunteered in the first year. By 2024 first-year turnover on third shift was 33 percent and its recordable rate was close to the plant average. Overtime used to cover night vacancies fell by roughly a third.

What I would change is the sequence. I extended equivalent terms to Lima, our non-union plant, nine months later. Employees there asked, reasonably, why they had to wait, and the delay contradicted the standard I say I hold.

We bought a folding-carton plant in Bowling Green, Kentucky, from its founding family in March 2023. Its 300 employees were not represented. Pay averaged about fourteen percent below our Lima rates, and in 2022 the seller had moved the plant to a rotating twelve-hour schedule that many employees disliked. At closing we announced a two-step wage increase, with the second step due in February 2024.

In November 2023 employees petitioned for a representation election, which a National Labor Relations Board regional office then ran. The plant manager, who had run the site for the seller for fifteen years, wanted outside campaign consultants and compulsory meetings, and our chief executive initially agreed with him. I argued for a different course. Managers could explain the company's view and answer questions factually, but we would hire no consultants, and every supervisor would be trained on what the law prohibits, including threats, questioning employees about union support, promises of benefits and surveillance. On counsel's advice, the February wage step would be paid exactly as announced, neither brought forward nor withheld.

The chief executive accepted that approach after counsel reviewed it. We gave up a harder campaign and, probably, some chance of a different result. The plant manager felt we had tied his hands, and he left the company later that year.

The employees voted 141 to 118 for representation in January 2024. No unfair labour practice charge was filed against the company. We bargained a first contract, ratified in December 2024, which replaced the rotating twelve-hour schedule with fixed eight-hour shifts. Annual turnover at Bowling Green fell from 44 percent in 2023 to 27 percent in 2025.

Corrinwood now enters every acquisition with a stated position on employee choice. What I would do differently is earlier and simpler: meet employees on every shift in the first month and ask about the schedule. The petition grew from a grievance we had inherited and never explained.