The assignment: explain what the inventory can and cannot support
A fictional reporting team wants a quarterly electricity-emissions total. The activity extract contains twelve rows covering several facilities and months. It mixes kWh and MWh, repeats an invoice, includes an out-of-boundary facility and an out-of-period month, and has one missing value and one estimate. Your deliverable is an evidence register, a coverage report and a short reporting note.
The factors file is prominently labeled FICTIONAL_ARITHMETIC_ONLY. Its 0.4 and 0.6 kg CO2e/kWh values are made up for this exercise and must not be used in an actual inventory. The project tests units, boundaries, missingness and evidence quality, not selection of real emissions factors or certification of a report.
GHG Protocol’s Corporate Standard concerns corporate inventories; its Scope 2 Guidance covers purchased or acquired electricity, steam, heat and cooling. Those sources supply methodological context. They do not make this small exercise a complete Scope 2 inventory, an assurance engagement or proof of legal compliance.
Data and worked inputs
Build the expected grid before measuring completeness
The boundary file includes facilities A, B and C and excludes D. The reporting period is January through March. That creates nine expected facility-month cells. Generate this grid before joining activity records; otherwise an absent record never appears in your completeness denominator.
Remove the exact B2 retry by record ID. Exclude D’s January row because the boundary says it is out of scope, and A’s April row because it is outside the reporting period. Keep the reasons in an exclusion log. Two different record IDs covering the same facility-month would require checking whether they are distinct bills or overlapping versions; this exercise requires one record per cell.
Eight cells have quantities; C in February does not. Coverage is 8/9, about 88.89%, counting the estimated A-March cell as numerically present but separately flagged. That is not 88.89% metered evidence. Of the nine expected cells, seven are metered, one estimated and one missing. Both coverage dimensions belong in the report.
Normalize units before applying any factor
A has 12,000 kWh, 14 MWh and an estimated 13,000 kWh. Convert 14 MWh to 14,000 kWh, giving 39,000 kWh for A. B’s 8 MWh, 9,500 kWh and 10 MWh become 27,500 kWh. C has 13,000 known kWh across January and March, with February still missing.
| Measure | Result |
|---|---|
| Expected facility-months | 9 |
| Metered / estimated / missing cells | 7 / 1 / 1 |
| Known electricity, including estimate | 79,500 kWh |
| Estimated electricity included | 13,000 kWh |
| Known-input result using fictional factors | 34.4 tonnes |
| Complete-quarter result | Unknown while C-February is missing |
Known activity including the estimate is 79,500 kWh. Of that, 13,000 kWh is estimated and 66,500 is metered. Preserve both raw quantity/unit and normalized quantity. A column containing 14 beside 12,000 is not safe to sum until the unit is resolved.
Apply each facility’s fictional regional factor only after normalization. A and B use 0.4; C uses 0.6. The known-input result is 34,400 kg CO2e, or 34.4 metric tonnes. Calling that a complete quarter would hide C-February; calling it a metered-only result would hide the estimate.
Keep missing-value scenarios separate from the baseline
The case supplies two possible missing-month inputs for sensitivity: 6,000 and 9,000 kWh at C. Adding those to the known subtotal produces 38.0 and 39.8 tonnes respectively using the fictional factor. These are two stipulated scenarios, not a statistically estimated confidence interval or a verified range of actual emissions.
The A-March estimate is another uncertainty. If you replace 13,000 kWh with 10,000 or 16,000 in separate scenarios, the known subtotal moves by minus or plus 1.2 tonnes. Label this change separately from the C-February missing-month scenario so the reader knows which assumption produced each result.
Do not store either scenario as the original observation. Keep the source state missing, store scenario assumptions in a separate table and show the baseline alongside them. Later receipt of a real quantity would be a new evidence update, not confirmation that the scenario had been a measured value all along.
A reporting note that survives a careful review
An informative note does not merely attach “estimated” to the end of a precise total. It explains the boundary, coverage and consequence of what remains unresolved.
Original fictional inventory reporting note
Turn the evidence register into a useful work sample
Build columns for source record, facility, period, original quantity, unit, quality, normalized quantity, factor reference and inclusion reason. The register should let a reviewer trace the total without opening every formula. Include explicit statuses for missing and estimated values; color alone is not enough.
Test an unknown unit, a duplicate record with changed quantity and a facility without a factor mapping. None should silently contribute zero. Also test removal of the C-February row entirely: the expected grid should still reveal the missing cell. This is the difference between validating the rows you received and checking whether you received the rows you expected.
Once you have the register and coverage table, write one narrow improvement proposal. It might be a monthly facility checklist or a requirement to retain units beside each uploaded amount. State who would supply the evidence and what would count as completion. More required fields are not automatically better; choose the ones that prevent an actual ambiguity in this case.
Data and worked inputs
Describe careful measurement rather than invented climate impact
A useful project description is: “Built a synthetic facility-month inventory, normalized mixed electricity units and preserved missing/estimated evidence in a coverage-aware reporting note.” Do not claim an emissions reduction. The exercise measures stipulated activity; it does not implement a physical change.
Expect interview questions about why the full total remains unknown, why a scenario is not a confidence interval and why the boundary file matters. Explain that getting the multiplication right is necessary but does not establish the validity of an emissions factor or completeness of the inventory.
For an extension, use a properly documented public activity dataset and an appropriate real factor only after independently checking its geography, period, units and intended use. Keep that extension separate from these fictional inputs. A transparent small project is more convincing than an impressive total whose evidence nobody can trace.