Rule of the week: Consistently Near Threshold

The error, and why it happens

You're reconciling the quarterly expense run and one name keeps turning up at $482, $470, $495 — never over the $500 no-receipt-review limit, never comfortably under it either. You assume it's coincidence until you scroll back three months and see the same pattern: claim after claim landing in that narrow band just below the number that would trigger a manager's second signature. Nobody has done anything you can point to on a single claim. That's exactly why it works. Approval limits exist to reduce friction on small, legitimate spend, and anyone who has watched how the limit is enforced learns, consciously or not, where the line sits.

This isn't usually invented fraud from day one. More often it starts as someone splitting a $540 dinner into two claims to avoid an awkward conversation, finds it goes through cleanly, and repeats the shape without really deciding to.

A worked example

Say the approval policy is: claims up to $500 need only the employee's own coding; above $500, a manager has to approve before reimbursement. Within 10% of that limit means anything from $450 to $500.

One employee submits 18 expense claims over a quarter:

• 11 of them sit between $452 and $498
• the remaining 7 range from $40 to $310

Eleven of eighteen is 61% of claims inside that $450–$500 band. On its own, one $485 taxi claim is unremarkable. Eleven of them, out of eighteen total, each individually under the radar, is the pattern worth a conversation.

How to find it manually

1. Get the approval limit right first. It's rarely in the accounting system — check the delegation of authority policy, and confirm whether the limit is flat or varies by role, cost centre, or category (travel vs. general expenses often differ).
2. Export the claim history per employee from your expense system (Xero expense claims, Dext, Expensify, or whatever sits upstream) for the period you're reviewing — a quarter is a reasonable window.
3. Calculate each claim as a percentage of the relevant limit, using a spreadsheet (=claim_amount/approval_limit). Flag anything between 90% and 100%.
4. Count the flagged claims against total claims per employee, and sort for anyone above 50%.

Honestly: this is tedious, not hard. You need to be comfortable with lookups across two data sources (the claims export and the delegation policy, which might be a PDF, not a spreadsheet), and if limits vary by role you'll be doing this calculation separately for each tier. For a business with thirty or forty expense-claiming staff, a proper quarterly pass is a half-day's work, done properly — most of it spent reconciling which limit applies to whom, not the arithmetic itself.

FAQ

What does "consistently near threshold" mean as an expense claim red flag? It means an employee's claims repeatedly fall just under an approval limit rather than being spread naturally across a range of amounts. One claim near the limit means nothing. A majority of someone's claims clustering there, period after period, suggests the limit itself is shaping the amount being claimed.

How do I check if someone is submitting expenses just under their approval limit? Pull their claim history, confirm the exact approval limit that applies to their role, calculate what percentage of that limit each claim represents, and see whether more than half sit within 10% of it. Do this per employee, not as a single company-wide average — the pattern is personal, not organisational.

How the automated version does this

EXP-007 will miss cases where the approval limit itself isn't recorded cleanly against the employee (shared limits, recent policy changes) and can false-positive on roles with genuinely repetitive spend near a fixed allowance — a courier with a standard $48 fuel claim against a $50 cap isn't gaming anything. Where the limit is known, it flags any employee whose claims cross the 50%-within-10% threshold continuously, without anyone running the quarterly export.

Finanomaly is currently waitlisted.