There's a category of retrofit measure that consistently looks excellent on paper and consistently underperforms once installed. The proposal projects savings of fifteen, twenty, thirty percent. Twelve months later the meter shows two. Sometimes zero. Sometimes worse than zero.
This isn't a fraud story. Most of the time the modelling is honest. It's an interaction story — a measure that performs on a worksheet because the worksheet didn't account for how the building actually runs.
Three patterns that account for most of the gap
Measures that fight the building's existing controls. New high-efficiency equipment installed alongside legacy control logic frequently runs in a setpoint regime the controls were never re-tuned for. The equipment is more efficient. The schedule isn't. Net change at the meter: noise.
Measures with savings stacked on top of each other. Two measures targeting the same end-use, modelled separately, will each claim that end-use's full reducible portion. Installed together they share the same envelope of available savings — and the meter only sees the share. The proposals weren't wrong individually. They just weren't additive.
Measures the operations team doesn't trust. A measure that requires the building operator to change a routine — different setpoints, different alarm responses — will drift back to the old routine within six months unless training and documentation made it through with the equipment. The capital was spent. The behaviour change wasn't.
Three questions before the capital plan signs off
- What controls or schedules have to change for this measure to work as modelled — and who is going to make those changes, with what training?
- What other measures share this measure's savings envelope, and what is the combined expected reduction, not the sum of individual claims?
- What does the operational data look like for this exact measure in a building of this exact type, run by an operator like yours? Not the manufacturer's case study — the meter.
A retrofit pathway that survives those three questions shows up in the operational data twelve months later. One that doesn't is a story dressed up as a plan.
Reviewing a retrofit proposal or building a capital plan you want to defend? Get in touch to talk through scope.