Make the whole system answer for the promise

As COO, I supported a proposed operating change because a defined comparison showed chiller electricity falling from 2.40 to 2.16 GWh. The initial presentation described the 0.24 GWh reduction as a saving available to the wider plant portfolio. Operations head Sameer Vezqrani asked for pumps and other auxiliaries to be included. Their electricity had risen from 0.72 to 1.03 GWh over the compared periods. Total electricity therefore moved from 3.12 to 3.19 GWh, an increase of 0.07 GWh, while each period reported 4.2 million refrigeration ton-hours of cooling.

The comparison did not establish that every operating condition was identical. Weather, equipment loading and other factors still required professional analysis. It did establish that the chiller-only account could not support the whole-plant saving I had encouraged the team to present. I withdrew the ₹16 crore rollout recommendation and authorised a ₹62 lakh, twelve-week system review through the appropriate operating process. Qualified engineering owners examined the measurement boundary and conditions; finance kept the proposed investment separate from the review expense.

The revised recommendation required a whole-system evidence account before any wider rollout returned for approval. The original programme did not proceed through the earlier paper. No later saving is invented, and the unapproved ₹16 crore is not presented as money recovered. The 0.07 GWh difference describes the reviewed periods, not a general forecast for every plant or proof of a single cause. Operating changes remained subject to professional authority and safe procedures.

My error was allowing a useful component measure to answer a larger question than it could support. The correction needed to reach both the investment decision and the company’s language about efficiency. I now ask which energy-consuming elements are outside a headline result and whether the useful output and operating conditions have been examined at the same boundary. A CEO should not make the account more confident merely because the narrower number is easier to explain.

An early thermal-storage proposal carried a ₹28 crore capital estimate and a nominal 24,000 refrigeration ton-hour storage quantity. The intended service called for 8,000 refrigeration tons of cooling for six hours. At that rate, the nominal quantity alone represented only three hours before accounting for losses or operating constraints. Six hours required 48,000 refrigeration ton-hours delivered. The discharge-rate presentation and the storage-energy quantity had not been brought together clearly enough in the original recommendation.

I asked engineering and finance for a revised option that could be examined against the actual duration requirement. The board approved a ₹46 crore project with 62,000 nominal refrigeration ton-hours, subject to professional design, commissioning and acceptance conditions. The larger nominal quantity was not itself proof of the promised service. The programme took fourteen weeks longer than the earlier outline, and the additional capital and schedule remained in the account rather than being hidden behind the improved specification.

Qualified teams established 54,000 refrigeration ton-hours of usable energy under the defined acceptance conditions and completed three controlled six-hour runs delivering 8,000 refrigeration tons throughout each run. Each run therefore delivered 48,000 refrigeration ton-hours. The remaining usable margin applied to those reviewed conditions; it was not a claim about every future load, ambient condition or maintenance state. Safety and commissioning responsibility remained with the professional owners. The commercial commitment was written around the accepted scope, with operating limitations preserved.

The decision was to buy and verify a capability whose duration matched the promise, accepting the higher cost and later completion. The original smaller proposal was not described as defective equipment; it simply did not establish the service we wanted to offer. I wanted the board to see that distinction before authorising the capital. A large capacity label can feel reassuring, but the leadership question is what useful service remains available over the period the customer is being asked to rely on.

Nine proposed customer-continuity arrangements were reviewed before their service commitments were finalised. The initial account emphasised separate cooling equipment and described the arrangements as though each had an independently supported continuity path. Electrical and operating review found that three still shared an unresolved upstream supply dependency relevant to the proposed scenario. Six had evidence supporting a narrower, defined continuity scope after the professional assessment. The review did not establish protection against every possible outage for any of the nine.

Commercial colleagues wanted to retain one standard continuity description while the remaining engineering work proceeded. I required the customer account to distinguish the supported six from the unresolved three. We spent ₹86 lakh over fourteen weeks on qualified dependency assessment and revised proposals. The work used the authorised engineering and assurance process; this narrative does not describe interrupting a live service or changing protective arrangements. Technical owners defined the conditions the evidence could support, and legal and commercial colleagues reviewed the resulting commitment.

The board authorised a phased ₹12.8 crore redesign programme for the three unresolved arrangements, with further professional acceptance required before broader promises could be made. That approval was not completion of the works. The six supported arrangements proceeded only within their examined scope; the other three remained subject to the revised delivery and acceptance plan. No outage avoided, customer-loss amount or universal resilience result is invented from the review.

The difficult commercial choice was to give different answers where the equipment count had made a common answer attractive. Separate machines did not remove every shared dependency, and approved investment did not mean the dependency had already been resolved. My responsibility was to preserve those distinctions in the promise the company made. The board could then decide whether the cost and timing of the redesign were justified, while customers received an account tied to the evidence rather than the appearance of redundancy.