One name

Your team may call it a failed sync, stale record, tie-out break, or source-of-truth issue.We call it drift.

If you’ve ever closed a termination only to watch the benefits deduction keep firing, written a clawback memo, signed a restated quarter, or had your name on a control that “passed” before the auditor pulled samples — you know the pattern.

The drift patterns we've already seen

Six quiet failure modes. Five are honest error. One is the kind no one wants to sign their name to.

Drift isn't one thing. It shows up everywhere two systems are supposed to agree. Below are the patterns we keep running into. Each one costs real money. Each one has an audit consequence. And each one walks straight past the controls built to catch it.

01
PAY-GM-03Global mobility

International double-pay

Someone transfers between countries. HR updates cleanly. The new country’s payroll starts paying. The old country’s payroll never stops, because a transfer is not a termination, so no off-boarding trigger ever fires.

Slips past
  • · Country-siloed payroll reviews
  • · Bank disbursement approvals
  • · Payroll-to-GL tie-out (totals balance)
Modeled exposure $100K–$150K / employee
·
3–6 months undetected
02
PAY-RT-08Pay rate timing

Promotion pay-rate drift

The promotion takes effect on the 15th. Payroll cuts on the 10th in one country and the 20th in another. Each one reads that date boundary its own way. Some pay the raise twice. Others skip a period entirely.

Slips past
  • · Monthly payroll review (totals balance)
  • · Effective-date policy documentation
  • · HRIS change-approval workflow
Modeled exposure $5K–$15K / employee
·
Persistent until corrected
03
COMP-RV-04Comp & revenue

Terminated rep, CRM opportunities still crediting

A rep leaves. Their CRM login is disabled the same day. But 23 open deals stay assigned to them. Those deals close, flow into the comp engine, and accrue commission to someone who no longer works here. The monthly recon does catch it, usually at quarter end, after the accruals have already posted.

Slips past
  • · CRM user deactivation (opps don’t auto-reassign)
  • · Payroll roster review (doesn’t see CRM)
  • · First-pass commission calculation
Modeled exposure $50K–$300K / open pipeline
·
Often caught late in quarterly accrual review
+Three more drift patterns we've cataloged
04
PAY-CC-11GL allocation

Cost-center reorg → GL drift per employee

A reorg moves 80 people from one cost center to another. HR updates. Payroll totals still tie to the general ledger, so the close looks clean. But 7 of those people land in the wrong cost center, and their wages post to the wrong department. That department’s numbers are wrong. Nobody notices until someone asks why headcount cost looks off.

Slips past
  • · Corporate controller tie-out (totals balance)
  • · Payroll review (roster is correct)
  • · Departmental P&L (uses payroll as source)
Modeled exposure $30K–$200K / quarter
·
Quiet until flux analysis
05
PAY-GE-01Fraud-risk testing

Ghost employees & terminated-but-still-paid

Someone is terminated in HR. Off-boarding completes: the manager signs, the badge is revoked, the login is killed. Payroll was the last to know, or never knew. Direct deposit fires next cycle. Benefits deductions keep posting. The retirement match keeps accruing. It could be a process gap. It could be someone keeping a friend on the books. No local control can tell you which, and the fraud-risk walkthrough your auditor runs has to rule out both.

Slips past
  • · HRIS off-boarding workflow (completes cleanly)
  • · Payroll roster review (employee “looks active”)
  • · Bank disbursement approval (totals tie)
Modeled exposure $35K–$80K / employee
·
Often 12–24 months undetected
06
DATA-SD-01Source file structure

Schema Drift Detection

An export template changes quietly. A column gets renamed, dropped, or added. The automation keeps running against the old shape, so the check still comes back clean while the file underneath stopped saying what it used to say.

Slips past
  • · Scheduled exports
  • · Column-based automations
  • · Downstream formulas that still execute
Modeled exposure Verify+ / source file
·
Caught before value comparison runs

Exposure figures are modeled, not measured: per-event bands derived from published correction-cost and wage research, sized against the drift types we see. The full derivation and its sources are on the pricing page.

Scan my dataRun it on your own files. See which of these you have.