Chief Human Resources Officer / Canada
Decisions that changed who does the work, and on what terms
- Colleagues
- 7,400
- Keswald Communications, June 2026; 2,450 in the craft bargaining unit
- Annual compensation and benefits expense
- C$842m
- All of Keswald in 2025, as distinct from the C$34m spent running HR
- Lost-time injuries per 200,000 hours worked
- 1.3
- 2025, down from 2.9 in 2017
Decisions & consequences
01Bringing 610 cable technicians into the craft unit with their service intact
When Keswald completed its acquisition of Bracewood Cable Systems in January 2023, we inherited 610 non-union cable technicians in Nova Scotia and Prince Edward Island. They earned an hourly rate plus per-installation bonuses averaging about C$6,800 a year, and supervisors assigned their work. Our own 1,840 craft employees were paid on hourly scales and bid for shifts by seniority. In March 2023 the union representing our craft unit applied to the Canada Industrial Relations Board to have the Bracewood technicians included in it. Bracewood's technician turnover was already 23% a year, and contractors completed 38% of its installations.
Our chief financial officer wanted to contest the application, arguing that the work and pay model were distinct; outside counsel expected a contest to last more than a year. Bracewood's former field operations head, now our Nova Scotia regional director, credited the per-installation bonus for Bracewood's same-day connections. I recommended not contesting, and using the year a contest would consume to negotiate the transition. The chief executive agreed. We gave up the per-installation model and accepted about C$7.1 million a year of additional labour cost.
The terms were settled inside the 2023 craft renewal. Bracewood technicians moved onto our scale at the step matching their service, and nobody's base pay fell. Their Bracewood service counted in full for vacation and shift bidding. For the first bid cycle Keswald technicians kept the shifts they held, which persuaded a divided union executive to vote seven to four for merging the lists by service date. For thirty-six months no technician from either company could be laid off as a result of the integration.
By 2025 voluntary turnover among former Bracewood technicians had fallen to 8%. Repeat visits within thirty days of an installation in Nova Scotia fell from 14% to 9%, and the contractor share of installations fell to 21%. Installations per technician-day dropped 4% in 2023 and were 1% below the earlier level in 2025. We now settle unit, service and scheduling questions before any acquisition closes.
What I would do differently concerns Bracewood's forty-eight field supervisors. They lost the discretion to assign work and heard the terms after their technicians did. Nine left within a year. I should have brought them into the design in February, before a single term was tabled.
Keeping truck location data out of discipline after a technician went missing
In January 2018 one of our technicians, working alone to restore a cable route in the upper Miramichi, slid off a forestry road in a blizzard. Beyond cellular coverage, with his check-in lost in the confusion of a storm night, he was found by a plough operator five hours later. He lost parts of two fingers to frostbite. With 61% of craft work done alone, lone-worker safety rested on telephone check-ins every two hours that dispatchers, swamped on storm nights, could not enforce.
I proposed satellite-capable check-in devices in all 1,300 trucks, an automatic alert when a truck stopped for forty-five minutes away from a work order without a check-in, and two-person crews for pole and aerial work after dark during declared storms. The union backed the crewing rule but opposed the devices, having seen location records used at disciplinary hearings elsewhere. Our head of field operations wanted the same data for routing and productivity reporting, which he valued at C$3.4 million a year. My argument was that technicians would leave a device they distrusted in the glovebox, so its safety value depended on the restriction.
In May 2018 we signed a letter of understanding, later carried into the 2019 craft agreement. The data may serve safety response, vehicle maintenance and, when an employee or the union asks, the answer to a customer's damage claim. Performance management and discipline are excluded. Records are kept for ninety days, and the policy health and safety committee audits every access quarterly. We gave up the productivity case and accepted storm restoration about 6% slower under the two-person rule. Interim satellite messengers were fitted within three weeks.
In the first two winters the system raised 236 alerts. Fourteen were genuine, including two rollovers found within forty minutes. Lost-time injury frequency fell from 2.9 per 200,000 hours in 2017 to 1.6 in 2021. In 2022 the union agreed to anonymised route analysis reviewed by the joint committee, which recovered about C$1.9 million of the routing gain. We now draft data restrictions with the union before buying any monitoring technology.
What I would do differently: contractors' crews stayed outside the system until a near miss in February 2019. Equivalent check-in should have been a contract condition from the first day.
Keeping repair calls in Atlantic Canada and sending the night queue out
Early in 2024 Keswald's four contact centres, in Moncton, Saint John, Sydney and Corner Brook, employed 1,180 agents at an annual cost of C$96 million. Volumes were falling about 6% a year as customers moved to self-service, and attrition was 41%. Our chief financial officer and our owners' operating partner proposed moving 45% of voice volume to an offshore outsourcer for an estimated saving of C$18.6 million a year. Our clerical agreement restricted contracting out that caused layoffs, so the proposal assumed a negotiated exit.
Having opened an offshore centre myself, I did not argue that the work could not be done well abroad. I argued that the proposal priced calls and ignored what the centres produced. Of the 318 technicians we hired between 2018 and 2023, 116 had started as agents, and their first-year retention was 94% against 81% for external hires. I proposed sending the overnight queue and simple billing chats to an outside provider, keeping daytime voice and all repair and technical support at home, and cutting attrition through twelve-week schedule blocks and a paid route into technician apprenticeships. Quillon, my former employer, bid for the overnight work. I declared the relationship and stepped out of the selection, and another provider was chosen.
We gave up about C$8.5 million of the projected saving. The clerical union opposed any offshore work at first. In the October 2024 renewal it agreed to amend the contracting-out clause for overnight and chat work, in exchange for a commitment that no one would be laid off because of the change and that reductions would come through attrition and voluntary moves.
By June 2026 the centres employed 960 agents and nobody had been laid off. Attrition had fallen to 26%. Seventy-four agents had entered technician apprenticeships and thirty-one had moved into network operations roles. Repeat repair calls within seven days fell from 19% to 14%. The annual saving reached C$10.1 million, including lower training costs. Every sourcing proposal now carries a workforce page showing internal career routes as well as unit costs.
What I would do differently is the handover between the overnight provider and the morning shift. For four months, overnight callers often had to explain their fault again after eight o'clock. A shared case note, tested before go-live, would have prevented it.
Training fibre splicers in the communities our network reaches last
Between 2021 and 2026 Keswald committed to bring fibre to 186,000 rural premises under co-funded federal and provincial programmes, many of them in northern New Brunswick, Cape Breton and Labrador. In 2021 we employed ninety-four splicers with an average age of fifty-one and spent C$21.4 million a year on contract splicing, much of it for crews flown into Labrador. When a fibre cut isolated a coastal Labrador community, restoration averaged fifty-eight hours, mostly spent waiting for people and weather.
I proposed a twenty-four-month paid apprenticeship with a curriculum written jointly with the craft union, and community-based positions posted first to residents of the area they would serve. That second element cut across seniority. Under our agreement a senior technician in Moncton could bid for a Labrador posting and then work it on rotation. Senior members argued, fairly, that we were asking them to surrender a right they had earned. Our head of network construction preferred contractors, who were quicker to deploy and carried no training cost; each apprentice cost about C$52,000 beyond the value of their output in the first two years. Several Indigenous community organisations told us they had watched companies start programmes and abandon them within two years.
The craft union's members ratified a letter creating forty-one community-based positions, in which residents bid by seniority among themselves and outside bids are closed for five years. The build ran two months late in two regions in 2023. With community-run employment and training organisations we agreed how candidates would be recruited, where training would happen and what we would report back to them each year.
Four cohorts have started 128 apprentices since 2022, and ninety-six have certified or are on track. Thirty-four live in First Nations communities or Indigenous communities in Labrador. Contract splicing spend fell to C$11.8 million in 2025, and average restoration for the coastal Labrador communities fell to fourteen hours. Apprenticeship is now our default route into splicing and network trades.
What I would do differently is where the classroom blocks were held. The first cohort trained in Moncton, with poor housing arrangements for people a long way from home, and nine apprentices left. We moved classroom blocks to regional hubs in 2024. We should have started there.