Site reporting reconciliation workbook
A central reporting total can agree with the sum of site returns while individual sites carry the wrong quantities. This workbook compares each final site return with its transferred central-pack value. It keeps the net difference, gross absolute difference and individual exceptions visible together, so equal and opposite errors cannot disappear behind a matching total.
The editable XLSX contains a clean 16-site Reconciliation sheet, a separate Fictional example and a field-by-field Read me. The matching PDF preserves those three sheets as a static reference; it does not recalculate. Use a separate copy for each metric, unit, period and approved site population. All case names, records and quantities are invented.
Download the site reporting reconciliation workbook (XLSX).
Download the static site reporting reconciliation workbook (PDF).
Start at the transfer point
Use the pack after the reporting team has selected the site quantities and decided which operations belong in the collection. Its specific question is whether those selected values reached the central pack at the right site. It does not build a meter inventory, select an organisational boundary or decide which source records should be added. Those earlier decisions need their own supporting schedules.
Keep both versions accessible: the final site return and the exact central extract being checked. Record the file version and a precise row, cell or record locator for each. If the central extract changes during review, create a new comparison copy or clearly version the affected records. Otherwise, one person may be investigating an error in an extract that another person has already replaced.
GHG Protocol Corporate Standard, chapter 7, table 4, discusses checks on transcription, electronic-file versions and aggregation. This supplies general quality-control context for the original transfer-check workflow here.
Give the population a separate reference
Enter the reporting pack, metric, common unit, period label and reference to the independently approved site list. The expected count can be from one to sixteen. Enter every expected site in the working rows, including a site whose quantity is still missing. Do not construct the expected list merely by copying the files that happened to arrive; that would make an omitted return disappear from both sides of the check.
Compare the actual IDs with the retained list before choosing Confirmed for the population state. The count is a second check, not a substitute for that comparison. Three entered sites and an expected count of three can still conceal one omitted site and one unintended replacement. The workbook cannot read the external list or establish that the human confirmation is correct. Leave Pending until that comparison has been completed.
Choose a common absolute tolerance and record why it is suitable for this transfer check. The workbook treats equality with the tolerance as within tolerance. The choice concerns working precision only. It is not a regulatory materiality threshold, a policy for correcting a report or permission to leave a known attribution error unresolved. Retain small differences and explain them even when the numerical flag is within tolerance.
Complete each pair before comparing it
A row contains the stable site ID and name, references for both versions, the two quantities, a same-coverage confirmation, an explanation or action reference, a responsible role and an Open or Reviewed state. Confirm same coverage only after checking the metric, unit, period and physical coverage on both sides. A common label at the top of the sheet does not prove that an individual return follows it.
The site quantity belongs in column E and the central quantity in F. The calculated signed difference is central minus site. A positive result means the central figure is higher; a negative result means it is lower. The absolute-difference column removes the sign only for the separate gross check. It does not turn a decrease into an increase or identify a physical change in operations.
The input check requires the fields before calculating a row difference. It also catches duplicate site IDs without regard to letter case. IDs must be text, with no leading or trailing spaces, control characters or the wildcard characters *, ? and ~. Use a stable ID rather than a changing display name. A duplicate rejection does not resolve which record should be retained; the owner must trace the versions and correct the pair.
Replay the offsetting-error case
The fictional Orchid pack P1 requests grid electricity for Q2 2026 in MWh. Population POP1 contains S01 Orchid North, S02 Orchid South and S03 Orchid Office. Its miniature site evidence is SR1, 240 MWh; SR2, 360 MWh; and SR3, 100 MWh. Central extract CP1 contains 247, 353 and 100 MWh against those respective IDs. The case stipulates matching periods, units and physical coverage; no conversion or estimate is required.
The site total is 240 + 360 + 100 = 700 MWh. The central total is 247 + 353 + 100 = 700 MWh. The signed row differences are +7, −7 and 0, so the net difference is zero. The gross absolute difference is 7 + 7 + 0 = 14 MWh. At the fictional tolerance of 0.1 MWh, two sites are outside tolerance. Two review states remain Open.
The invented migration explanation says CP1 assigned a seven-MWh sub-record from S02 to S01. In a separate repair copy, change F17 to 240 and F18 to 360. The central total stays 700; gross difference becomes zero. Preserve CP1, identify corrected extract CP2 and link the reassignment action to both affected sites. Changing the arithmetic does not automatically set their review states to Reviewed or supply the evidence for that decision.
Treat missing information differently from zero
Clear F19 in a copy of the example. S03 then lacks its central quantity, its row calculation is withheld and the aggregate results disappear. This is intentional: a total of only the available pairs could look complete when it is not. The row remains in the expected population and needs a specific action. Retain the previous complete example separately so the test cannot be mistaken for a reporting update.
Now enter a supported numeric zero instead of a blank. Under that different assumption, the central total is 600, the net difference is −100 and the gross difference is 114 MWh. Zero is an actual assertion about the quantity; it does not mean that the file is unavailable. The workbook distinguishes those states but cannot determine whether the zero has evidence. That remains the responsibility of the people preparing and reviewing the pack.
Preserve the supported working contract
Numeric inputs accept literal numbers only. Paste values and retain the General input formats. Numeric formulas, logical values, text numbers and errors are rejected. Quantities can be zero or from 0.001 to 1,000,000,000 in the declared unit; tolerance can be zero or from 0.001 to 1,000,000. Both allow at most three decimal places. Differences and totals are rounded to three decimal places. These are software bounds, not measurement guidance. A date-formatted numeric serial is still a number; the checks cannot identify a mistaken date or a wrong semantic unit in a quantity field. If your scale requires another range, redesign and test a separate tool rather than silently rounding the source.
The period is a text label. The workbook performs no date-overlap, unit-conversion or source-measurement test. Rows 17–32 are the full capacity. Do not insert rows, overwrite formulas or extend the table without extending and retesting every dependent range. There are no macros or external data connections. Input cells are not password-protected; preserve a clean master and check that the expected formulas remain in place.
The arithmetic gate requires valid setup, complete active rows, a confirmed population and agreement with the expected count. Open review count is displayed separately. This lets the team inspect a fully entered mismatch while keeping the human follow-up visible. A status of Inputs complete describes the entry checks only, and Reviewed records the supplied human state. Neither is a verification opinion.
Carry the exceptions into the handover
Send the labelled totals with the site-level differences, population reference, two version references and unresolved actions. Ask the recipient to reproduce one positive difference, one negative difference and the gross calculation before relying on the file. Check the saved copy in the intended spreadsheet application. Exact-file headless LibreOffice recalculation and export checks are documented for this candidate; Microsoft Excel and desktop interaction have not been tested.
The site-consolidation guide addresses the broader expected-versus-received collection and aggregation workflow. Use the calculation change register when a correction needs a separate version and release trail. This workbook supplies the narrower paired-site transfer check between those activities.
For help organising a reporting handover, contact Nitium Advisory with a non-confidential description of the pack, population and unexplained differences. Keep restricted source records in your approved storage.
Version 1.0. Prepared 15 October 2026. Original Nitium Advisory workbook, diagram and fictional case. This optional records-preparation aid supplies no emissions calculation, regulatory determination, professional sign-off or independent assurance. Source context was checked on 15 October 2026. No source organisation endorses the workbook.
Comments