Every reporting period has missing data. Assurers accept estimates. They do not accept undocumented ones.
An estimate with a stated method and assumptions passes. A number with no explanation does not.
When to estimate rather than chase
Chase the data when it exists and someone can send it. Estimate when it does not exist, or when the effort to get it is out of proportion to its size.
Decide per gap, and record which you chose.
Choosing a method
Gap | Method |
Some months missing from a full year | Annualise from the months you have |
A site with no meter data | Apply intensity from comparable sites, by floor area or headcount |
Employee commuting with partial survey responses | Extrapolate responders to total headcount |
A supplier that will not provide data | Spend-based factor |
A category with no activity data at all | Spend-based from the ledger |
A missing final month before reporting | Use the prior period average and restate once actuals arrive |
Use the same method for the same gap each year. A method that changes between years breaks comparability, and any change needs recording.
Annualising from partial data
If you hold nine months of invoices, do not simply multiply by 4/3. Check whether the missing months are seasonal.
For electricity and gas, use the same months from the prior year to weight the estimate. For steady consumption, a straight scale-up is acceptable.
State which you did.
Extrapolating survey results
Where a survey covers part of your workforce, apply the average result of respondents to non-respondents.
Record the response rate. A 60% response rate extrapolated to the whole workforce is reasonable. A 5% response rate is not, and needs a different method.
Record every estimate
For each one, record four things:
What was missing.
The method you used.
The inputs and assumptions.
How confident you are.
This goes into your Basis of Preparation, so your assurer sees the gap and the reasoning together.
Improve estimates over time
An estimate is not permanent. Each year, review which gaps can now be closed with real data.
Where you replace an estimate with actual data, record the change and its effect on the comparative.