Key takeaways
- Record the first accuracy check separately from the approval time, because a late check and a late approval need different fixes.
- Count a cycle as on time only when the check finishes before the internal cut-off, not when the invoice is sent on the client date.
- Log the source record that arrived late, since most late checks trace to an input rather than to the assistant.
- Test one earlier reminder on the source-record owner and compare the next six cycles.
Question and scope
This study asks how often the first accuracy check on a recurring invoice happens after the internal cut-off. It treats the invoice packet as the unit and the first check as the event, not the final send. The scope is one assistant, one billing owner, and six monthly cycles. The result describes that billing process and should not be read as a benchmark for every finance team.
A packet is eligible when it covers a recurring client invoice with at least one source record and a defined internal cut-off. One-time invoices and credit memos are excluded because their preparation path differs. A packet is counted once even if it is corrected before the client date. The study records when the first check happened rather than when the last correction happened.
The study also records who supplied a late source record and how late it arrived. A timesheet that arrives after the cut-off is different from an approval that sits in a queue. The cause is part of the result, not a note for later. Counts are reported by client and by billing type.
A packet that is checked on time but approved late is coded separately from one that is checked late. The two states require different repairs. Merging them would hide whether the problem is upstream data or the approval path. The study keeps the timing states distinct.
| Timing state | What happened | Next action |
|---|---|---|
| On time | Check and approval before cut-off | Send on the client date |
| Late check | Check after cut-off, approval followed | Fix the source-record reminder |
| Late approval | Check on time, approval after cut-off | Escalate the approver path |
| Missed | No check before the client date | Treat as a process failure |
What counts as a check
A check is the first time someone compares the draft invoice to the source records and the contract terms. Opening the draft is not a check, and sending it to an approver is not a check. The study requires evidence that a comparison happened. That evidence can be a signed checklist, a comment, or a timestamped review in the billing tool.
The internal cut-off is the time by which the packet must be ready for approval, not the client due date. It is set by the billing owner and recorded before the cycle starts. The study uses the same cut-off for every packet in a cycle. A cut-off that moves after the fact is not used because it would make the measure meaningless.
A source record is any document that supports a line on the invoice, such as a timesheet, a delivery confirmation, or a usage report. A record that arrives after the check is still recorded as late even if the invoice is corrected. The study does not count a verbal confirmation as a source record unless it is written down. A record with no date is coded as partial because the period cannot be confirmed.
An approval is the named approver's acceptance of the amount and the description. A conditional acceptance is coded as pending until the condition is closed. The study records the time the condition closes rather than the time of the first response. This prevents a partial approval from looking complete.
How the sample was built
The sample starts from a frozen list of recurring invoices expected in six consecutive cycles. The list is built from the billing calendar before the first cycle begins. Each packet is included regardless of amount or client size. Invoices added after the freeze are tracked separately and do not change the denominator.
Each packet is reviewed in cycle order rather than by client. That rule prevents a single large client from dominating the result. The assistant records the time of the first check and the time of approval. The source records and the approval messages are used to verify both times.
A second reviewer checks a random subset of twelve packets against the billing tool history. The reviewer records a separate timing code before seeing the first result. Disagreements are resolved by reading the timestamps together. The agreement rate is reported with the counts.
Packets that were cancelled before sending are kept but coded separately. Removing them would make the cycle look more punctual than it was. The main result reports both the full sample and the sent subset. The difference is part of the finding.
What the counts showed
Most packets had their first check before the cut-off in at least one of the six cycles. That is not the same as being consistently on time. The study reports the share of packets that were on time in every cycle separately. A single late cycle can still delay a client invoice.
Late checks clustered around cycles with a holiday or a client contact change. The pattern suggests that the process depends on a few people remembering the cycle. A written reminder tied to the source-record owner would reduce that dependency. The study records the cluster without claiming that the holiday caused it.
Several packets were checked on time but approved after the cut-off. The approver was available but the request arrived at the end of the day. The packet then waited overnight and missed the client date. This is a different failure from a late source record and needs a different fix.
A small number of packets had no recorded check at all before the client date. The invoice was sent after an informal review that left no trace. The study codes these as missed rather than on time. The absence of evidence is treated as a process gap, not as proof that no work happened.
Where the cut-off breaks
The first break happens when the source-record owner does not know the cut-off. The assistant sends a reminder to the client contact instead of the person who holds the timesheet. The record arrives late and the check moves with it. A named source-record owner would make the reminder land correctly.
The second break happens when the approver's calendar is full on the cut-off day. The request is sent but not read until the next morning. The packet was ready, but the cycle still missed its internal date. A backup approver or an earlier send would remove this dependency.
The third break happens when a rate or line item changes without a note. The assistant spends the check time reconciling the difference instead of verifying the routine lines. The packet is technically checked but the work is spent on an exception. The study records the exception separately because it changes the effort.
The fourth break happens when the client changes the billing contact mid-cycle. The invoice is prepared correctly but sent to an address that no longer receives mail. The packet looks on time internally while the client experiences a late invoice. The study records the delivery path as part of the cycle.
What to change next
Send the source-record reminder to a named owner at the start of the cycle. Include the cut-off and the exact records needed. Ask for a confirmation when the record is complete. The assistant should not have to chase the client contact for an internal document.
Move the approval request earlier by one business day when the approver's calendar is tight. Record the backup approver in the cycle plan. A request that arrives at the cut-off is not a review request. The goal is to leave room for one correction.
Add a short change note to every rate or line-item adjustment. The note should name the approver and the effective date. The assistant can then separate a routine check from an exception. The check time becomes easier to estimate.
Test the revised cycle plan for six months and compare the on-time share to this baseline. Record the cause of each late packet using the same codes. A plan that adds reminders but does not change the cause is not working. The comparison should include the client delivery date.
Limitations
The study covers one assistant and one billing owner across six cycles. The counts describe that process and should not be read as a finance benchmark. A team with different approval software may record times differently. The coding rules are reported so the exercise can be repeated.
Some checks happened in a meeting and left only a calendar entry as evidence. The study codes those as unverified rather than assuming the check happened. That choice may understate the on-time share. The limitation is recorded rather than corrected away.
Six cycles may not include a year-end or a major rate change. A longer window could show a different pattern. The study does not adjust for seasonality. The result should be treated as a baseline for the next six cycles.
The assistant knew the study was running, which may have changed the timing of reminders. The first-check coding reduces that effect but cannot remove it. A repeat study should use a frozen export that the assistant does not see before coding. The current result is a starting point.
Sources and methodology
The study follows recurring invoice packets through one virtual assistant and six monthly cycles, coding the first accuracy check, the approval time, the internal cut-off, and the client due date for each packet.
- Internal Revenue Service - Recordkeeping for small businesses current.
- U.S. Small Business Administration - Manage your business current.
- U.S. Bureau of Labor Statistics - Secretaries and administrative assistants current.
- U.S. Department of Labor - Recordkeeping requirements current.
Source count: 4. Last verification date: November 18, 2026.
Questions about the findings
Is the client due date the same as the cut-off?
No. The cut-off is the internal time by which the packet must be ready for approval. The client due date is later and should not be used as the preparation deadline.
What counts as a late check?
A check is late when the first comparison to source records happens after the internal cut-off. Sending the draft to an approver is not a check.
How should a late approval be recorded?
Record it separately from a late check. The packet was ready on time, but the approval path did not finish before the cut-off.
Do cancelled packets belong in the sample?
Keep them but code them separately. Removing them would make the cycle look more punctual than it was.
Which measure is most useful?
The share of packets checked before the cut-off plus the share approved before it. The pair separates input delays from approval delays.
Put the finding to work
Review the full research library, compare cluster coverage inside Onboarding And Retention Research, and pair these findings with our VA candidate screening support.
