My practice starts where the employment offer meets the working week.

In late 2021 our recruitment proposal called for 120 apprentices to join the following year. The number reflected anticipated staffing demand across operating assets. It did not yet reflect how many experienced employees could supervise practical learning. Operations director Gao Nianru could release twelve mentors without leaving existing assignments dependent on informal cover. The commercial team wanted the larger intake to demonstrate that workforce preparation was keeping pace with the asset pipeline.

I asked us to cost both versions honestly. A larger cohort would require additional experienced hires or a different schedule, neither of which had been secured. We agreed on seventy-two apprentices and a RMB 4.6 million programme covering learning support, mentor release and coordination. I owned the employment and programme arrangements. Operations retained responsibility for technical assessment and any permission to perform particular work. We explained that distinction to recruits before they accepted their offers.

The decision meant forty-eight fewer places than the initial proposal. It also meant that future staffing needs would require further action rather than being treated as solved by recruitment. Some managers saw the smaller intake as a retreat. I regarded it as a commitment we could actually resource, while recognising that the remaining workforce requirement had not disappeared.

By December 2023, sixty-three apprentices had completed the programme, five had left and four required additional development time. Those categories describe the programme record, not sixty-three newly authorised technicians. Operations reviewed readiness for work separately. We also found that two mentors needed more support with giving feedback, which changed the preparation for the next intake. I would not attribute subsequent site performance to this programme alone. The defensible outcome was a funded learning arrangement with an explicit account of who completed it, who did not and what remained to be assessed. It was smaller than the launch proposal and more useful than an unstaffed promise.

At four remote operating locations, voluntary departures among 186 employees reached twenty-nine in the twelve months preceding our 2023 pilot. Managers proposed a general retention allowance. Employee conversations suggested a more mixed picture: inconsistent travel connections, unresolved accommodation repairs and uncertainty about whom to approach when a promised arrangement failed. Pay mattered, but the evidence did not support treating every departure as a salary problem.

I proposed a RMB 3.8 million package covering repairs, published travel arrangements and a named follow-up process with site supervisors. Finance preferred a smaller allowance that could be administered centrally. Operations worried that publishing travel arrangements would create expectations that weather and urgent work might disrupt. We agreed to describe both the normal arrangements and the process for handling an exception, rather than promise an uninterrupted schedule.

The trade-off was visible. Physical repairs took time, and colleagues elsewhere could reasonably ask why these locations received investment first. We selected the pilot because the problems had been documented and the operating teams were willing to examine the results. I explained that it was a local experiment, not a judgement that other employees had fewer needs. Site managers remained responsible for day-to-day delivery; my team checked unresolved requests and reported recurring issues.

Voluntary departures fell to nineteen during the following twelve months. That change was encouraging but did not isolate the effect of repairs, travel or the wider labour market. Several employees still left for family reasons or work nearer home. Two repairs also ran beyond their planned dates, requiring temporary arrangements. We retained those facts in the review rather than present the pilot as a complete answer to retention.

The lesson I carried forward was about the employment offer itself. A remote role includes the practical experience of living away, and that experience deserves a budget and an accountable owner. An allowance can be useful, but it cannot repair a room or make an unreliable explanation more credible.

Our 2024 career review identified ninety-six technical employees eligible to apply for a new senior specialist track. Previously, a significant pay step often depended on taking a supervisory position. Some strong specialists wanted deeper technical responsibility without managing a team. Meanwhile, existing managers were concerned that a parallel route would diminish the standing of their roles or create a title without comparable accountability.

I worked with operations to separate the two contributions. Technical assessors defined the evidence required for specialist appointments, including the scope of problems handled and the contribution to developing other practitioners. My function designed the application process, reward ranges and review arrangements. We did not substitute a generic leadership interview for technical judgement, and we did not present the track as an automatic reward for long service.

The first round made fourteen appointments. Nine further applications required additional evidence and were given a defined opportunity to return for review. Other eligible employees chose not to apply in that round. The appointments added RMB 1.1 million to annual pay costs. Finance had asked whether the change could be funded entirely by reducing management positions. I opposed linking the two before we understood the operating requirement. A specialist route had to be justified by contribution, not by an assumed saving elsewhere.

We accepted the additional cost and some uncomfortable comparisons. Two line managers requested a review of their own role scope, which was reasonable and handled separately. One initially unsuccessful applicant challenged the clarity of the evidence requested; the panel revised its explanation without changing the assessment outcome. Those conversations showed where the framework needed better communication.

The immediate result was a credible alternative route for fourteen people, not a claim that the whole technical workforce had been transformed. We scheduled a review of the responsibilities attached to the appointments after a year. My responsibility was to keep the employment architecture intelligible while leaving technical standing with the people qualified to judge it.

In 2018 Lanshuan Power acquired operating assets whose workforce included 214 employees. The central integration plan proposed moving them onto our payroll timetable in the first month. That would simplify administration, but it would also alter when some employees received their regular pay. The transaction timetable treated this as a systems matter. Conversations with employees showed that household commitments had been organised around the existing dates.

As people director for operating assets, I recommended retaining the existing timetable for six months while records were checked and the proposed arrangements discussed. Legal colleagues reviewed the employment requirements, and finance assessed the cash and administration implications. I did not offer legal assurances or promise that every existing arrangement would remain indefinitely. I wanted the transition to be understood before a central process changed something people relied on.

The integration lead objected to maintaining two administrative cycles. Finance estimated RMB 420,000 of additional processing and reconciliation cost over the six months. We accepted that cost and delayed the payroll consolidation. In return, each unresolved record had an owner, employees received individual explanations where their circumstances differed, and the eventual change was communicated with enough time for questions to be addressed.

By the end of the transition, 207 employee records had been reconciled and seven remained under individual review because supporting documentation was incomplete. Those seven were handled separately; they were not forced through the change to make the project appear finished. We did not experience a missed regular payroll run during the transition. That outcome depended on payroll, finance and local administrators, not on my decision alone.

The consequence was six months of extra work and a more deliberate integration. It reinforced my view that an asset transaction does not erase the practical history of employment. A central team can close a project while the people affected still lack an answer. I prefer to leave the open cases visible until someone has resolved them.

At Hexuan Grid Services in 2011, I supported an autumn development schedule that looked efficient on paper. We would bring 148 technical employees through practical sessions while several customer maintenance assignments were also under way. I led workforce development and believed that careful scheduling could accommodate both. Regional operations managers warned that the same experienced colleagues were being counted as instructors and field supervisors.

I did not give that warning enough weight. Within the first six weeks, eighteen planned sessions had been cancelled or postponed. Employees had rearranged travel, and instructors were repeatedly pulled back into operating work. The attendance figures did not describe a functioning learning programme. I took the problem to the operations director and recommended reducing the autumn schedule rather than continuing to circulate revised calendars that depended on the same unavailable people.

We moved practical assessment for fifty-two participants into the following year and spent RMB 680,000 on relief cover and revised travel. Finance challenged the extra cost because the original programme had already been approved. I acknowledged that my team had helped create the overcommitment. The remedy needed a realistic operating calendar, not an argument that the approved budget should make the plan possible.

The remaining ninety-six participants completed their scheduled learning that year, while the deferred group received revised dates and an explanation of the delay. Completion still did not represent technical authorisation. We separately checked that no assignment depended on treating a delayed participant as already qualified. Operations made those decisions; my team corrected the employment and training records.

The failure changed how I prepare a development proposal. I now ask for named teaching capacity and the work from which instructors will be released before agreeing an intake or timetable. That is more demanding at the planning stage and less wasteful for the people who would otherwise discover the conflict after they have travelled. It is also a reminder that a people leader must be willing to revise her own attractive plan.