CRM and billing · Controlled comparison

Subscription reconciliation.

Compare normalized CRM and billing records, inspect discrepancies and keep unresolved cases visible. This example verifies the report against known expected outcomes.

This is a separate reconciliation example. It does not measure Lead Agent preparation quality or customer savings.

Reconciliation / Verify the report

A checked discrepancy report.

This controlled example compares 3,000 CRM rows and 3,000 billing rows by explicit rules against a manifest of expected cases. It is not a customer deployment.

Agreed records

2,730

Matching amount, currency, period and status. Ready for report review.

Confirmed differences

240

180 amount differences and 60 status differences under the declared comparison rules.

Explicit uncertainty

30

USD pairs require review. No assumed FX conversion or automatic correction.

Flagged amount delta

€900

Billing minus CRM for supported flagged pairs. A difference to investigate, not recovered revenue.

Checked against the expected result

240 of 240 planted discrepancies detected · 0 false positives · 0 missed discrepancies · 0 cents error in the supported amount total.

All 3,000 subscription outcomes, per-pair amount deltas and reason codes match the expected manifest, including the 30 review cases. These results apply to this controlled dataset and rule contract. They do not establish error-free performance on future customer data.

SUB-00001 · discrepancy

CRM: EUR 49.00 · active

Billing: EUR 54.00 · active

amount mismatch

SUB-00181 · discrepancy

CRM: EUR 49.00 · active

Billing: EUR 49.00 · past_due

status mismatch

SUB-00241 · review

CRM: EUR 49.00 · active

Billing: USD 49.00 · active

unsupported currency

SUB-00271 · match

CRM: EUR 49.00 · active

Billing: EUR 49.00 · active

All compared fields agree.

Inspect all 30 pending review cases

These pairs remain unresolved. A reviewer must establish the currency treatment before any monetary comparison; the report makes no FX assumption.

Reconciliation review queue · recorded pending cases
SubscriptionCRM / BillingDecision
SUB-00241EUR / USDPending
SUB-00242EUR / USDPending
SUB-00243EUR / USDPending
SUB-00244EUR / USDPending
SUB-00245EUR / USDPending
SUB-00246EUR / USDPending
SUB-00247EUR / USDPending
SUB-00248EUR / USDPending
SUB-00249EUR / USDPending
SUB-00250EUR / USDPending
SUB-00251EUR / USDPending
SUB-00252EUR / USDPending
SUB-00253EUR / USDPending
SUB-00254EUR / USDPending
SUB-00255EUR / USDPending
SUB-00256EUR / USDPending
SUB-00257EUR / USDPending
SUB-00258EUR / USDPending
SUB-00259EUR / USDPending
SUB-00260EUR / USDPending
SUB-00261EUR / USDPending
SUB-00262EUR / USDPending
SUB-00263EUR / USDPending
SUB-00264EUR / USDPending
SUB-00265EUR / USDPending
SUB-00266EUR / USDPending
SUB-00267EUR / USDPending
SUB-00268EUR / USDPending
SUB-00269EUR / USDPending
SUB-00270EUR / USDPending
Download the unresolved cases →
Rules, evidence and limits

Exact stable subscription ID, same monthly period, EUR integer cents and explicit normalized statuses. No fuzzy matching. Duplicate IDs, unsupported currencies, unknown statuses, invalid amounts and incompatible periods require review. Missing counterpart records are flagged without inventing an amount.

This reconciliation uses deterministic code and makes no model calls. Financial totals are checked against known expected values.

The input rows, expected manifest and all outcomes are downloadable. Input and expected-result digests are saved with the report. No live HubSpot/Stripe connection or write-back occurs. Tax, discounts, proration and currency conversion require an agreed extension to the rules. Investigation time, reviewer acceptance and customer savings have not been measured.

Inspect the recorded verification report →Download inputs, expected cases and all outcomes →

A customer pilot would agree these rules first, run against an approved reference set, then measure report acceptance and remaining investigation work. That is how we would establish correctness and economic value for the actual process.

Back to all demos →