The 9% That Wasn't: Estimated Meter Data and a Late Correction
A grocery retailer with 480 stores is preparing energy and refrigerant data for its first limited-assurance engagement. Electricity for 62% of stores comes from automatic meter reads; the remaining 38% is estimated by the utility from prior-year consumption and invoiced monthly. Mid-year the retailer switched energy data providers, and the new provider recalculated consumption using a different calendar convention, shifting December 2023 usage into January 2024. A regional review then found that one country's figures had been keyed in MWh where the source was kWh, overstating that country's electricity for two months before correction. Refrigerant top-ups are recorded by a facilities contractor in a free-text maintenance log, and the sustainability team converts kilograms to tCO2e using a global warming potential list last updated in 2019. The draft report shows a 9% year-on-year electricity reduction, which the CFO intends to headline in the press release accompanying publication.
1. Which of these items require a prior-year restatement, and how would you decide?
Read the worked answer
Three different things are being conflated. Separate them before deciding.
- Unit error (kWh keyed as MWh) is an error, not a methodology change. Correct it, then test significance - 5% of the relevant metric or of total emissions is a common internal threshold. Two months in one country may fall below it; calculate rather than assume. If significant, restate the prior year and disclose the restatement, the reason and the effect.
- Calendar convention change is a methodology change. It breaks comparability, so restate the prior period on the new convention, or disclose the effect if restatement is impracticable.
- The 9% headline is now unsafe. Part of it may be the December-to-January shift: moving December 2023 out of the prior year depresses that year's consumption and mechanically inflates the apparent reduction. Rebuild the comparison on a like-for-like basis before any public claim is made.
Document the correction trail, and make sure the assurance provider sees both the original and the restated figures - a restatement the practitioner learns about after fieldwork is a credibility problem, not just an accounting one. Also check whether the restatement shifts any target baseline.
2. How should the report handle the 38% of electricity that is estimated, and what should be done about the 9% claim?
Read the worked answer
Estimated data is not automatically unassurable, but it has to be handled honestly.
- Classify. 38% of stores on utility estimates is a material estimation component. Record the method (prior-year consumption, degree-day adjustment, both?), the provider, and the share of total kWh that is estimated.
- Explain the basis of preparation. The report should state that a defined percentage of electricity is estimated from billed data and describe the method. A reader told this can judge the 9%; a reader not told cannot.
- Decompose the reduction into weather (heating and cooling degree days), store count changes, known efficiency projects, metering and tariff changes, and unexplained residual. Only the efficiency and structural components support a performance claim. If the weather-normalised reduction is 2%, that is the number to state.
- Scope the assurance. Limited assurance gives "nothing has come to our attention". Agree with the practitioner which sites, months and estimates are in scope, and expect sample testing of meter reads and recalculation of estimates. The engagement letter must not describe a narrower scope than the report implies.
- Fix the controls. Set a tolerance for estimate-versus-actual variance, migrate estimated sites to automated reads, and disclose the trajectory.
3. What is wrong with the refrigerant data, and what would you require before it is assured?
Read the worked answer
A free-text log combined with a 2019 global warming potential list is both a control failure and a methodology risk.
- Capture. Replace free text with a structured record: date, equipment ID, refrigerant type (R-404A, R-134a, R-744), quantity added, reason (top-up, leak, retrofit, decommissioning), and technician. A top-up is a proxy for leakage, so the log must distinguish a leak from a planned replacement, and the recharge quantity must be traceable to a work order or invoice.
- GWP values. Applying a list last updated in 2019 is a methodological inconsistency, and GWPs for common refrigerants moved substantially between IPCC assessment reports. Confirm which assessment report the applicable framework expects, use one set consistently across sites and years, and treat the switch as a methodology change requiring recalculation of the prior year if the effect is significant.
- Disclose. State the GWP source and version in the basis of preparation, and the share of refrigerant data based on contractor records rather than measurement.
- Assure. Expect the practitioner to sample top-up records back to invoices and to re-perform the GWP conversion independently.