Two teams can produce different figures for the same apparent question while each relies on a source it considers authoritative. One may be reviewing a settlement record, while the other is working from an internal calculation supported by meter or operational data. The disagreement can persist even after both teams repeat their calculations because the figures may reflect different periods, revisions, populations or calculation rules. Before deciding which value to use, the organization needs to establish what each one represents.
ReconIntel is being developed for electricity-market teams that need to review settlement outcomes, charges and operational variances. Its proposed workflow connects authorized settlement records with expected values and supporting evidence, allowing differences to be investigated by period, account or charge category. This pre-launch case study uses a fictional reconciliation scenario to explain that approach. The amounts and findings are illustrative; they do not describe a client engagement, a live ReconIntel deployment or an actual IESO settlement calculation.
The challenge: A variance appears before its cause is understood
In the scenario, an analyst’s internal calculation produces an expected charge of CAD 100,000, while an authorized settlement record shows CAD 112,000 for what initially appears to be the same scope. Both values come from controlled processes, and each has an identifiable owner. The resulting CAD 12,000 difference is sufficient to require investigation, although the difference itself does not establish that the settlement charge is incorrect.
The first difficulty is determining whether the comparison is valid. A shared account name and period label are useful starting points, but they do not confirm that the same charge categories, interval boundaries or revisions are included. An internal estimate might have been prepared before a source correction, while the settlement record may incorporate that correction. Alternatively, the two sides may use different allocation logic or include different components within an apparently common total.
This is where a reconciliation process can lose its explanatory value. If the internal result is overwritten with the settlement value, the visible difference disappears without establishing why it occurred. Selecting whichever number came from the more recent report can create a similar problem because a later report may describe a revised scope. The review needs to retain both original values and build an explanation that can be assessed independently of the final total.
Establishing a common basis for comparison
ReconIntel’s proposed approach begins by defining the comparison and retaining the supporting records. The reviewer needs to identify the relevant account or participant, period, charge category, units and source versions before deciding whether the values can be compared. Where the source records use different time boundaries or levels of detail, any alignment must be documented so that the reconciled view remains traceable to the originals.
For electricity-market work, the calculation also needs to use the rules and market-version coverage applicable to the period under review. An internally expected value depends on its source inputs and agreed logic, so its authority cannot be established solely by the fact that it came from a controlled spreadsheet or system. The same scrutiny applies to the settlement side: the review must confirm which record and revision are being examined and whether they match the intended comparison.
Once this basis is established, the initial variance can be retained as a review item with its supporting evidence. In the fictional case, CAD 12,000 is the opening difference between the settlement record and the internal expectation. Keeping that opening position allows subsequent findings to be assessed against a stable reference, even if corrected inputs or rules produce a new expected value later in the investigation.
Investigating the difference through evidence
The first finding in the illustrative review concerns scope. A charge component included in the settlement record was absent from the internal expectation, accounting for CAD 7,000 of the opening difference. Before accepting that explanation, the reviewer would need to confirm that the component belongs to the agreed comparison and that it has not already been included elsewhere. Once verified, the adjustment explains why the initial internal total was lower without requiring either source record to be silently rewritten.
The second finding concerns timing and revision. The internal expectation used an earlier version of supporting data, while the settlement record reflected a later revision. Recalculating the affected component with the aligned source version accounts for a further CAD 3,000. In this scenario, the amount represents the effect of a verified revision under the agreed calculation logic; it is not an assumed relationship between meter consumption and every settlement charge.
The third finding concerns an internal allocation rule. Reviewing the affected calculation identifies a CAD 1,000 difference attributable to how a component was assigned within the internal model. Correcting that rule explains another part of the opening variance, with the correction documented separately from the source revision. This separation matters because a change to an internal calculation and a change to its input evidence require different explanations and potentially different follow-up actions.
The remaining CAD 1,000 has not been explained. It stays open for further review rather than being absorbed into one of the confirmed categories. In the illustrative bridge below, the explanations are mutually exclusive and each contributes once to the opening variance, preventing the same adjustment from being used to explain more than one part of the difference.
| Reconciliation item | Illustrative amount, CAD | Treatment in the review |
|---|---|---|
| Original internal expectation | 100,000 | Retained as the opening baseline |
| Verified scope difference | +7,000 | Missing component added to the comparable expectation |
| Verified source revision | +3,000 | Inputs aligned to the relevant revision |
| Verified internal allocation correction | +1,000 | Calculation corrected with supporting explanation |
| Revised comparable expectation | 111,000 | Original expectation plus verified adjustments |
| Settlement record under review | 112,000 | Retained as the comparison value |
| Remaining unexplained variance | 1,000 | Left open for investigation |
The review has therefore explained CAD 11,000 of the initial CAD 12,000 difference, while preserving a CAD 1,000 unresolved balance. That is a more useful result than a total forced into agreement because the record distinguishes established causes from outstanding questions. It also shows how apparently trusted values can disagree for several reasons at once, with each reason requiring its own evidence.
The controls needed to make the explanation defensible
A reconciliation becomes reviewable when its decisions are retained alongside its arithmetic. ReconIntel’s proposed workflow includes keeping supporting evidence, attaching explanations and tracking unresolved exceptions through review. For the fictional case, that would mean preserving the original comparison, the records used to verify each adjustment, the revised expectation and the remaining open balance. A reviewer could then follow how the conclusion was reached without relying on an analyst’s recollection.
Source-data gaps, timing differences and calculation issues also need separate treatment. A missing supporting record may prevent a conclusion, while a verified timing difference may explain a variance without indicating that a source is wrong. A confirmed internal calculation issue, meanwhile, may require a correction to the model and a check of other affected periods. Grouping all three under a generic “data issue” would obscure the action needed to resolve each one.
The workflow should also distinguish explanation from resolution. A reviewer may understand a timing difference while still awaiting the appropriate revised record, just as a model correction may be identified before it has been approved and applied. An exception’s status should communicate that progress clearly. Closing a review therefore requires the evidence and action appropriate to its cause, rather than merely a smaller difference on the screen.
Review thresholds belong within an agreed process as well. They can help prioritize work by amount, recurrence or significance to the business, but a value falling below a threshold does not establish its cause. In the example, the remaining CAD 1,000 would still need a recorded disposition under the organization’s review policy. Any accepted difference should retain the basis for acceptance so that later reviewers can distinguish an approved treatment from an overlooked exception.
What the pre-launch case demonstrates
The scenario illustrates the scope of ReconIntel’s proposed reconciliation workflow: compare two values on an agreed basis, surface the variance, retain evidence and show the path towards resolution. The initial number identifies where a review should begin, while the investigation determines whether the difference arises from scope, source timing, calculation logic or an unresolved issue. The usefulness of the product depends on keeping that reasoning connected to the records being compared.
ReconIntel remains in pre-launch development, and this case does not claim recovered charges, reduced review time or proven customer savings. Its intended value is to make settlement and operational variance reviews more consistent and easier to inspect, using authorized records, customer-specific inputs and verified calculation rules. Demonstrating that value in practice will require the relevant data, configured comparison logic and evidence from actual use.
For teams handling reconciliation today, the central question is whether the process preserves what has been learned when a difference is investigated. A total that agrees can be useful, but the explanation behind that agreement determines whether the work can be reviewed, repeated and applied to the next period. Where does reconciliation consume the most time for your team: assembling the records, aligning their definitions, explaining the variance or securing the evidence needed to close the review?