1094-C Count Reconciliation: What Software Should Show

Use the 2025 Form 1094-C and its instructions to establish which count the operator is reviewing. Line 18 concerns accompanying Forms 1095-C. Line 20 concerns the ALE Member's total across its transmittals. Part III contains monthly employee counts with different definitions.

For the ordinary column (b) task in this demonstration, the reviewed count excludes employees in a Limited Non-Assessment Period. Column (c) includes all employees and uses a permitted counting day consistently through the year. The instructions also provide an exception from completing column (b) for an eligible employer using the 98% Offer Method. Do not make a product populate a field merely because an evaluation spreadsheet expects a number.

Have the evaluator name the applicable rule and source report before opening a product dashboard. This prevents a visually prominent total from becoming the presumed answer for every field.

The product may satisfy these tasks through several reports rather than one screen. Judge whether the operator can follow the evidence reliably. A manually assembled reconciliation can be evaluated too, provided its workload and risk of using the wrong version remain visible.

The following wireframe describes information relationships, not a screenshot of an existing product. Ask the vendor to locate equivalent information in its workflow. Missing fields become concrete follow-up questions instead of a vague impression that the dashboard is incomplete.

A link labelled details is useful only if it opens the right employer, month and version. Test whether changing a filter silently changes the context. Check that downloaded evidence retains enough context to make sense after the browser is closed.

Fictional example: Harbor Tool has two disjoint original batches containing 110 and 70 Forms 1095-C. The selected batch's line 18 count is 110, while the employer's authoritative line 20 total is 180. Neither value is intended to be the employer's June workforce count.

For June, the reporting reviewer establishes 94 full-time employees under the applicable method, including seven in a Limited Non-Assessment Period. The demonstrated column (b) result is 94 - 7 = 87. The employer's consistently applied last-day count produces 109 total employees for column (c). These populations use the distinct instructions described above.

Ask the evaluator to retrieve the 110-record batch list, the two-batch employer summary, the reviewed full-time population with its exclusions, and the June last-day roster. A warning that all four totals differ would be unhelpful. A useful exception identifies a violated definition or an unexplained source discrepancy.

Continue the fictional case by having the reviewer correct the classification of one of the seven excluded employees. Assume the reviewed fact now establishes that only six of the 94 are in a Limited Non-Assessment Period for the relevant count. The expected column (b) value becomes 88. This controlled change does not add an employee or alter the original batch population, so the expected values of 110, 180 and 109 remain unchanged.

Observe whether the product updates the intended month and field, identifies the changed source record, and regenerates the affected draft. If a downstream report remains at 87, ask whether it is a preserved historical snapshot or an unexpectedly stale current report. Both can look similar without version labels.

Do not let the demonstration imply that software can decide the classification from this arithmetic. The reviewer supplies the corrected fact; the task tests its traceable effect.

Give the exported packet to someone who did not watch the demonstration. Ask them to identify the employer and year, find the selected count, locate its source population, and explain the change from 87 to 88. If they need a vendor narrator, record which context is missing from the export.

Check exception handling with a deliberately incomplete source snapshot. The product should make it possible to distinguish a true zero, a field legitimately not required, and missing data awaiting review. Record exactly how the workflow communicates each condition; these are evaluation goals, not a claim that the IRS requires specific software labels.

Bring findings into the 1094-C evaluation scorecard. Preserve the dataset, expected results and observed limitations so an implementation team can repeat the test after configuration or an upgrade.