This scorecard is part of the guide to reward hacking in a month-end close, where a bank reconciliation ties only because of a $9,600 plug. That guide shows what a gamed pass looks like. This page helps you judge, on a call with a controller or CFO, whether a client kept enough to show the difference between a real pass and a gamed one, using answers only.
Three gates to clear first
A gate is pass or fail. If one fails, stop scoring.
| Gate | Passes when | Question that settles it |
|---|---|---|
| Size | 50 or more full-time employees at peak, contractors excluded | "What was your full-time headcount at its highest?" |
| Authority | A named executive could authorize a discussion about licensing finance operations records | "Who would sign off on anything involving your accounting records?" |
| Location and English share | The company is in a covered country; outside the U.S., Canada and the EU, at least 70% of workpapers, notes and entries are in English | "Where is the company based, and are reconciliation notes and journal descriptions written in English?" |
Companies in the U.S., Canada and the EU qualify, as do companies in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia when at least 70% of the data is in English. The company also needs documented operating history and rights to license; criteria 5 and 6 test those.
Six criteria for close records, scored 0 to 2
AI developers want examples of systems that pass a check without doing the work. Google DeepMind's name for this is specification gaming, and OpenAI reports running a monitor model to catch it in reasoning models on coding tasks. For a finance client, the question is simpler: could an outsider tell, from the archive alone, a close that was finished from one that was only made to look finished? Each criterion scores one part of that.
| Criterion | 2 points: strong | 1 point: borderline | 0 points: weak |
|---|---|---|---|
| 1. Line-level detail | The matching worksheet is archived every month, item by item | Detail kept for some accounts or some years | Only the signed summary balance survives |
| 2. Review notes | A reviewer's reasons for sending work back sit on the workpaper | Notes exist but live in email or chat | Feedback is verbal |
| 3. Corrections as reversals | Corrections are separate reversing entries with their own user and date | Mixed: some reversed, some edited in place | Entries are edited directly, leaving no trace |
| 4. Open item tracing | Carried items keep their reference from the month they appear to the month they close | Traceable with manual searching | Cannot be traced |
| 5. Years of close files | Five or more years in one system, or migrated with detail | Two to four years | Under two years |
| 6. Ownership | The company owns its records; client agreements permit use | Client or owner agreements not reviewed | Records belong to clients or a parent |
Criterion 4 earns its place because software lets work roll forward. Microsoft's documentation for Dynamics 365 bank reconciliation says a worksheet can be finished with lines left open, which then carry into the next one. A team that can follow each carried line to the month it cleared can prove nothing was buried.
Criterion 6 matters most for property managers, fund administrators and outsourced accounting firms, because some records belong to the owners or clients they serve. Introducers who work with trustees or assignees should also read whether a referral reward belongs to a bankruptcy estate, since their own role can affect the reward.
What strong, borderline and weak fits look like
| Total | Band | Typical profile | Next action |
|---|---|---|---|
| 10 to 12 | Strong | Item-level worksheets, review notes and reversals across five or more years | Introduce the CFO |
| 6 to 9 | Borderline | Good detail, but notes in email or a recent system change | Ask one follow-up question, then decide |
| 0 to 5 | Weak | Summary balances only, edits in place, or client-owned records | Note the company for later |
A borderline company is often one discovery fact away from strong: where review notes were kept before the current system, or whether older worksheets were archived on a shared drive.
A borderline score, worked through (Illustrative)
Illustrative and fictional. A 310-person wholesale food distributor in the UK keeps its records in English. The controller answers:
| Criterion | Answer given | Score |
|---|---|---|
| 1. Line-level detail | Bank and intercompany worksheets archived monthly since 2023 | 2 |
| 2. Review notes | Review comments go by email, not on the workpaper | 1 |
| 3. Corrections | Reversals are separate entries with user and date | 2 |
| 4. Open item tracing | Carried items lose their reference after a quarter | 1 |
| 5. Years of close files | About three years since the 2023 ERP change; older files in the legacy system | 1 |
| 6. Ownership | Records unreviewed against a licensing clause in two large customer contracts | 0 |
| Total | 7 of 12, borderline |
The one zero is the deciding answer. If counsel confirms the customer contracts do not restrict the company's own finance records, criterion 6 moves to 2 and the total reaches 9: still borderline, but worth introducing with the email-notes gap named. If the contracts do restrict use, the opportunity may be narrower. SourceX's rights review settles it, not the introducer.
Five follow-up questions in plain finance language
- "Do you keep line-level reconciliation detail, or only the signed summary?"
- "When a manager sends a reconciliation back, is the reason written down, and where?"
- "Are corrections posted as reversals, or are entries edited in place?"
- "Can you trace an open item from the month it appeared to the month it closed?"
- "Who could approve a conversation about licensing finance records, and do any client agreements limit it?"
Keep the conversation at metadata level: which accounting system, how many years, roughly how many accounts reconciled monthly, who owns the files. Never ask for workpapers or exports. Partners who advise finance teams can see how assistants already query ledgers in the CFO dashboard guide and the general ledger reconciliation guide. If the owner asks how your reward relates to their payout, the guide on referral rewards versus the company's licensing proceeds explains it. The introducer guide to operational data licensing covers the wider process.
After scoring, the reward hacking evidence gaps page shows what is likely missing and the review packet shows what the company would assemble later. More record families are in data types.
Next step
Refer a company. No cap on rewards. Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.
If a finance client clears the gates and scores strong, or borderline with a gap you can name, register as a partner, confirm the basics on who qualifies, and read the program terms before introducing the CFO.