Capital choices with continuing obligations attached

In 2022, we were preparing a C$42 million extension intended to serve a new mixed-use precinct. The development case included seven expected anchor buildings. Three had reached a firm service commitment; the others remained dependent on property decisions and construction timing. The development team believed that starting the network would encourage the remaining customers to commit. Operations and finance were concerned that the company would carry the cost of capacity long before the associated demand arrived.

I asked for the investment case to be separated into committed, probable and exploratory connections. We examined the cash requirement if the four uncommitted buildings arrived later, in a smaller form or not at all. The exercise did not establish that the precinct would fail. It showed that the proposed construction sequence relied on demand for which the enterprise had no comparable commitment. I recommended an eighteen-month pause in major construction while preserving a limited set of site options and development work.

The board accepted the recommendation. The pause cost C$700,000 in site and option expenditure, and we lost the original construction timetable. I did not present the full C$42 million as a saving: it was expenditure deferred, not a completed project delivered more cheaply. The commercial team continued discussions with the prospective customers, using a clearer account of the conditions required before the next capital decision.

Some owners found the slower timetable frustrating because the extension had featured prominently in the growth discussion. I kept the distinction between interest and commitment in subsequent reports rather than quietly replacing the old date with another optimistic one. The project remained an opportunity under review. The result I can claim is a better-defined exposure and an avoided premature construction commitment, with the cost of preserving the option still visible. It is not proof that waiting will ultimately produce a more profitable development.

For another extension, I would establish the minimum customer commitment before commissioning the detailed schedule, reducing the pressure to treat engineering preparation as a reason to proceed.

Our 2023 capital review included a set of buried pipe sections with recurring service interruptions. The engineering team proposed C$9.6 million of renewal work. At the same time, the commercial plan contained attractive new connections that would have made the investment programme easier to describe as growth. The available funding could not support both proposals on their original schedule. I asked for a comparison that placed the existing service obligation beside the expected contribution from expansion.

The condition evidence was incomplete in some locations, and I did not want the renewal case strengthened through false precision about when another failure would occur. Engineers described the observed condition, interruption history and access constraints. Commercial colleagues set out the consequence of postponing selected connections. I recommended the renewal programme with a reduced new-connection allocation, accepting that the immediate revenue story would be less favourable.

In successive twelve-month periods, unplanned supply interruptions on the reviewed sections fell from seventeen to five. The weather and operating circumstances differed, so the comparison was not a controlled estimate of the programme's entire effect. It did provide evidence consistent with the purpose of the work. We kept the period, locations and definition of an interruption unchanged in the report and avoided turning the result into a claim about every part of the network.

The consequence was also organisational. Network teams saw that renewal could compete for capital through a clear service case, rather than only after a highly visible failure. The postponed connections remained a cost of the choice and were reconsidered in the next plan. I would not argue that every maintenance request should outrank expansion. The decision showed that an operating obligation deserved a comparable place in the capital discussion, even when its successful delivery would be largely invisible to customers.

I would ask for comparable maintenance evidence before the first capital ranking, rather than allow visible new connections to set the terms of the investment debate.

A proposed electrification programme raised a question about how much of the network's future operating behaviour we understood. The full concept would have required a much larger commitment than the first site could justify on its own. In 2024, I supported a C$6.2 million first stage using a heat pump at one network, with existing service arrangements retained during the transition. Qualified engineering and operating teams defined the technical scope and the conditions for commissioning.

The development team wanted the first stage described as the beginning of a settled rollout. I asked for a different approval: an operating stage whose evidence could alter the subsequent programme. We specified the measurements required, the financial boundary and the questions the first review should answer. The board could then see that proceeding with the initial work did not approve every later site or resolve every remaining assumption.

During the 2025 review period, the installation delivered 3.1 GWh of metered heat and used 1.2 GWh of electricity within the stated measurement boundary. Additional electricity expenditure was C$260,000, while the estimated reduction in gas cost was C$190,000. Those figures excluded capital recovery and some shared operating costs. They therefore did not establish a complete cost comparison, but they plainly did not support a claim that substitution had automatically lowered the energy bill. We also made no verified emissions claim from those figures alone.

I reported the learning and the unfavourable cost difference together. The next proposal was revised to examine operating periods, customer demand and the cost assumptions more closely before another site was approved. That response disappointed colleagues who wanted a simple success announcement. I regarded it as the purpose of staging the investment: the first season should change the next decision when the evidence warrants it. The programme remained a transition effort with unresolved economics, not a completed proof replicated across the portfolio.

I would define the full seasonal cost account before authorising another electrification trial, so the first review can examine the original proposition without reconstructing its boundary.

In 2023, building managers on four networks repeatedly challenged invoices whose calculations were consistent with the underlying agreements but difficult to follow. Our service team treated many queries as a communication problem, while finance was concerned that additional explanation would create manual work. I asked for a joint review of the invoice journey, including meter periods, estimated readings, later adjustments and the distinction between fixed and usage-related charges.

We selected a stable group of customer accounts and redesigned the explanatory schedule accompanying the invoice. The work also identified genuine corrections worth C$180,000 across the reviewed period. Those corrections were handled separately from the presentation change. I insisted on that separation because describing every dispute as customer confusion would have obscured errors for which the company was responsible. Contract terms and approved charging arrangements were not rewritten through a communications exercise.

Over six comparable billing cycles, disputes requiring a formal service investigation fell from seventy-two to nineteen per thousand invoices in the cohort. The finance team spent more time preparing the initial explanatory material and checking the adjustment history. We had not isolated that effort well enough to claim a net administrative saving. What we could show was a clearer account for customers and fewer escalated disputes under the same recording definition.

Some colleagues wanted to describe the new format as proof that the original bills had always been correct. I rejected that framing because the review had found both explanation problems and actual corrections. We retained a route for customers to challenge the calculation and used recurring questions to improve the schedule. The case changed how I viewed billing governance: accuracy and intelligibility are separate obligations, and meeting one does not excuse a failure in the other. A service relationship lasting many years needs both.

The step I would move forward is testing the bill with customers before introduction, including those who cannot rely on a service colleague to explain the calculation.

As COO in 2018, I reviewed spare equipment held across seven networks. Individual sites had accumulated items over time, sometimes with incomplete records of compatibility or replacement arrangements. Finance saw an opportunity to release working capital through a shared inventory. Operating managers worried that a central store would look efficient until a site needed a part that could not arrive or be fitted within the required service window.

I asked the engineering and logistics teams to test the proposed sharing arrangements before removing local holdings. They examined compatibility, access, transport and the circumstances in which two sites might need the same item. This was a planning and commercial review, not permission to bypass technical acceptance or operating procedures. We kept some duplicate holdings where the consequence of relying on a single shared spare was not acceptable.

The resulting programme retained C$2.7 million of identified critical spares and released C$1.1 million from redundant or genuinely duplicative stock over eighteen months. The release was recorded as a reduction in inventory, not recurring profit. We did not claim that the programme prevented a particular future outage, because that outcome could not be established from the period reviewed. Network managers had a clearer record of what was available and how an authorised request would be fulfilled.

The decision required me to disappoint both a larger savings expectation and a preference to keep every historical local item. I learnt that a shared resource is only real when its practical dependencies have been examined. A line in a central register does not establish that the resource can support a site at the time it is needed. That lesson later influenced my approach to portfolio investment: I ask which assumptions connect assets across the group and whether those assumptions have been tested before treating the connection as an economic benefit.

I would rehearse moving a critical spare between locations before accepting the pooled economics, making the physical response part of the initial investment evidence.