It is harder than it should be to recommend a one-week walk-through with a clipboard and a serviceable power quality logger to a client who has just sat through three pitches for cloud-based, machine-learning-augmented, real-time energy intelligence platforms. The audit is the older, slower, less visually compelling option. It is also, on our project records, the one that consistently delivers more recovered cost per cedi of consultant spend than any alternative.
Three reasons — none of them a criticism of the platforms, which are genuinely useful applied to plants that have already had the boring audit done.
One — the platform does not know what is broken.
A real-time platform monitors what its sensors are wired to. It does not know about the condensate drain stuck open, because there is no sensor on the drain. It does not know the boiler air-fuel ratio has drifted, because that data lives on the boiler controller and was never federated. It does not know the chilled-water bypass has been wired open by an operator tired of resetting it, because there is no sensor within ten metres of the valve. An engineer walking the plant finds all three in an afternoon. A platform makes a well-instrumented plant smarter; it does not, by itself, make an uninspected plant inspected.
Two — the highest-yield findings are nearly always physical.
On the audits we have closed, recovered cost is dominated, year after year, by findings of roughly this kind:
- A steam trap failed open, venting [PERCENTAGE] of boiler output to atmosphere.
- A cold-room evaporator carrying [MM] of frost because the defrost cycle was never recalibrated.
- A compressor at 8.2 bar serving a network whose worst case needs 5.5 bar.
- An exhaust fan running continuously because its production interlock was lost in a control upgrade.
- A transformer at [PERCENTAGE] load with no consolidation plan, paying full no-load losses every hour.
None require artificial intelligence. They require somebody to look. Cumulative recovery from a serious walk-through on a mid-sized facility is, in our experience, in the [PERCENTAGE] band of annual energy spend, payback inside [TIMEFRAME] on the fixes and roughly [MULTIPLE] return on the audit fee — a one-off recovery, after which the platform (or disciplined sub-metering) earns its place keeping the gains.
Three — the audit teaches the plant something the platform does not.
The most underrated output is not the report. It is the two days the auditor spends walking with the plant's own engineers, asking why the bypass is open, what the receiver pressure has been, when the boiler was last combustion-tested. They know most answers, or can find them by lunchtime. What they have rarely been asked is to think about their plant as an energy system rather than a sequence of fires. The boring audit transfers a way of looking; the plant keeps it after the consultant leaves. The platform transfers a dashboard.
So when is the platform the right answer?
After the boring audit, not instead of it:
- Year 1 — walk-through, closure of physical findings, cost-centre sub-metering.
- Year 2 — weekly accountability against the energy-per-unit numbers.
- Year 3 onward — analytics layered on top, for the specific cases it is genuinely good at: anomaly detection, demand prediction, predictive maintenance.
The principle. In that sequence every layer earns its capital, and the order of the layers is everything. The boring audit is not an outdated method. It is a foundational one. Skip it, and the more impressive engineering above it has nothing to stand on.