Small-business outsourcing research
Where does a weekly cash-flow check lose its source evidence?
A protocol for finding where a small business's weekly cash-flow check drops the source link behind a balance or payment.
Research question and scope
This protocol asks where a weekly cash-flow check loses the source evidence behind a reported balance or payment. The unit is one line item in a weekly check, identified by the week, the account, and the item. The study measures whether the line has a source record, whether the source was available at the cut-off, and whether the amount matches the source. It does not evaluate whether the owner made a good spending decision.
Set the scope to one small business, one set of accounts, and one weekly template. Record the bank feeds, the payment systems, and the cut-off time at the start. If the template changes during the window, keep the earlier definition for the full sample or analyze the change as a separate period. A new column can look like better evidence when it is only a new field.
Population and unit of analysis
Define the population as every line item in the weekly checks produced during the window. The unit is the line item, not the report, because one report can contain several items with different evidence. A line that is revised the following week remains one unit with a revision history. A new line for a different payment is a new unit.
Record the account, the item type, the amount band, and whether the item was committed or planned. These strata matter because a fixed rent payment and a one-off purchase carry different evidence needs. Include items with no amount, such as a missing invoice, because the gap is part of the finding. Test the unit definition on two ordinary weeks before collection begins.
Definitions that make the measure reproducible
Define a source record as the bank transaction, invoice, contract, or payment confirmation that supports the line. Define a linked line as one that names the source or provides a direct reference. Define a matching amount as agreement within a stated tolerance. Define a lost source as a line that cannot be traced to a record at the cut-off.
Do not count a remembered balance as a source. Do not count a total that matches only because two errors cancel out. A line that is estimated should be labeled as estimated and should not be counted as sourced. Each definition should include one positive example and one boundary example.
A bank feed can show a pending transaction that later changes amount or disappears. Record whether the line was based on a pending or posted item. A pending item is evidence of an expectation, not of settled cash. That distinction prevents a forecast from being treated as a balance.
Evidence standard and data dictionary
For each line item, capture the week, the account, the item type, the amount, the source reference, and whether the amount matches. Add whether the item was committed or planned and whether it was revised later. Record the person who prepared the check and the time it was sent. Store the packet with the weekly check.
Also capture whether the source was available before the cut-off. A source created after the check is not evidence for that week. If the line combines several transactions, record the list and the total. A second reviewer should be able to rebuild the line from the packet.
A second view is the share of lines where the amount matches but the source is missing. A matching total can hide an unsourced line when other lines are correct. Record the source status separately from the amount match. That distinction prevents a clean total from masking a weak check.
Sampling and collection
Draw the sample from the weekly checks rather than from a list of difficult weeks. Record the export date, the filters, the row count, and the selection rule. Include weeks with different volumes and different preparers. If the population is ordered by week, a fixed interval spreads the sample across the window.
Collect the check and the source as they existed at the cut-off, not as they look after later corrections. If the check was revised, record the revision time and analyze the original separately. Do not infer a source from the eventual payment. Note the collection date separately from the week.
The review should also separate a source that was unavailable at the cut-off from one that was never created. The first is a timing problem, while the second is a process gap. Show the counts separately and describe the fix. The owner can then decide whether to move the cut-off or change the source process.
Analysis and reporting
Report the number of line items, the share with a source, the share with a matching amount, and the share with an estimated value. Show the counts by item type and preparer. Present the share of lines that were revised the following week. Use the same denominator for every rate.
Inspect the lost-source items and describe the pattern without naming the preparer. A cluster in one item type may point to a missing feed, while a cluster in one account may point to a manual process. Compare the result with the amount band, because a large item may receive more attention. Report the estimated items separately.
A useful additional view is the time between the source and the cut-off. A source that arrives after the cut-off may be correct but unavailable for the check. Show the gap and describe the consequence. That view helps the owner decide whether to change the cut-off or the data feed.
A useful check is whether the preparer was the same person who reviewed the check. A self-review may miss a missing source. Record the preparer and the reviewer. That evidence helps the owner decide whether a second review is needed for the largest items.
Limitations and sources of error
The record cannot show whether the owner understood the cash position or whether the decision was wise. It can show only whether the line had a source. A sourced line is not proof of a good forecast, and an unsourced line is not proof of an error. The protocol also cannot capture a verbal balance the owner checked on a banking app.
Bank feeds can lag, duplicate transactions, and omit pending items. A line that looks unsourced may be a timing difference. Record those cases and their treatment instead of dropping them silently. If the preparer and the reviewer are the same person, note the independence limitation and analyze that subgroup separately.
The final view is the share of lines that changed after the check was sent. A report that is revised every week is hard to compare. Show the revision count and describe the cause. That view helps the owner decide whether the template or the cut-off needs to change.
Review cadence and next action
Give the result to the owner and the preparer and ask them to choose one observable change. The change might be a required source link, a later cut-off, or a rule that estimated items are listed separately. Record the owner, the effective date, and the measure that will show whether the change worked. Avoid changing the template and the cut-off at the same time.
Repeat the protocol with the same definitions after the change and compare the same fields. If sourcing improved but the check arrived later, the cut-off may be the real constraint. If estimated items fell but the owner still did not use the check, the report may be too detailed. Close the review when the owner accepts the finding and the next sample is scheduled.
Keep the weekly check and the source links together so the next review can repeat the sample. A finding without the underlying records is hard to verify. Record the reviewer and the date. The packet should be easy to find. Record the review date with the sample.
Sources
- IRS, recordkeeping for small businesses, IRS; checked 2026-11-05
- U.S. Small Business Administration, manage your business, U.S. Small Business Administration; checked 2026-11-05
- U.S. Government Accountability Office, internal control standards, U.S. Government Accountability Office; checked 2026-11-05
- Federal Trade Commission, business guidance, Federal Trade Commission; checked 2026-11-05
Frequently asked questions
Does a favorable sample prove the routine caused the result?
No. The design is observational and supports a bounded operating decision, not a universal causal claim.
What should an owner review first?
Start with high-consequence defects, missing denominators, exceptions, and cases that required the owner to reconstruct the evidence.
Related research
How often do small-business invoices arrive without a matching purchase record?
A protocol for measuring invoice-to-purchase-record matching in a small business, with a clear unit, evidence standard, and dated review.
What share of vendor renewals are approved after the notice deadline?
A protocol for measuring late vendor renewals and the notice process that lets a small business decide before the deadline.
How complete is the refund record for a small-business support desk?
A protocol for measuring whether a small business can reconstruct why a refund was given, who approved it, and what the customer was told.