Small-business outsourcing research

How long do supplier bank-detail changes wait for independent verification?

A protocol for measuring the delay between a supplier bank-detail change and the independent check that confirms it.

How long do supplier bank-detail changes wait for independent verification?

Research question and scope

This protocol asks how long a supplier bank-detail change waits between the request and an independent verification. The unit is one change request for one supplier record, identified by the supplier's legal name and the account being replaced. The study measures the time to first contact, the time to a completed check, and whether the change was applied before the check finished. It does not judge whether the supplier's new account was legitimate.

Set the scope to one small business, one vendor master, and one accounts-payable process. Record the systems that can receive a change request and the people who may approve one. If the process changes during the window, keep the earlier definition for the full sample or analyze the change as a separate period. A new callback rule can look like a faster check when it is only a new step.

Population and unit of analysis

Define the population as every bank-detail change request that reached the business during the window, including requests that were withdrawn, rejected, or never completed. The unit is the request, not the supplier, because one supplier can request several changes. A request that is forwarded to another team remains one unit with a handoff interval. A second request for the same account is a separate unit linked to the first.

Record the supplier category, the spend band, and whether the supplier is new or established. These strata matter because a new supplier and a long-standing supplier carry different fraud risks. Include requests that arrived through a sales contact rather than the finance inbox, because the base rate is part of the finding. Test the unit definition on two ordinary requests before collection begins.

A supplier can send a change from a personal address while the contract names a company domain. Record both the sending address and the domain on file so a later reviewer can see the mismatch. A request that cannot be matched to a supplier record is still a unit, but it should be labeled unmatched. Do not silently drop it, because unmatched requests are often the ones that matter.

Definitions that make the measure reproducible

Define the start time as the moment the request was recorded in the change log, not the time it was first mentioned in a call. Define the first contact as the first outbound attempt to a phone number the business already had, not a number supplied in the request. Define a completed check as a conversation with a known contact who confirms the change. Define a late check as one completed after the business's stated verification window.

Do not count an email reply as independent verification when it goes to the address that requested the change. Do not count an internal note as a verification attempt. A request that was applied and later reversed should keep both dates. Each definition should include a positive example and a boundary example so a second reviewer can apply it the same way.

Evidence standard and data dictionary

For each request, capture the supplier name, the old account, the new account, the requester, and the channel used. Add the time the request was logged, the time of the first outbound call, the person who made it, and the number dialed. Record whether the call reached a person, a voicemail, or no answer. Keep the call log with the change record rather than in a separate system.

Also capture whether the request was applied, who applied it, and whether the verification came before or after the change. Record any document the supplier provided and whether it was checked against an independent source. If the request was rejected, record the reason. A second reviewer should be able to rebuild the timeline from the change record and the call log.

If the business uses a callback form, keep the completed form with the request. A form without the number dialed and the person reached is only a claim that someone called. Record the device or line used for the callback when the team has more than one. The evidence should show independence, not just activity.

Sampling and collection

Draw the sample from the change log rather than from a list of suspected fraud. Record the export date, the filters, the row count, and the selection rule. Include requests from every channel and every day of the week so the result is not dominated by one busy period. If the population is ordered by time, a fixed interval spreads the sample across the window.

Collect the request and the verification as they existed at the time, not as they look after a later cleanup. If a note was added days later, record the addition date and analyze the original delay separately. Do not infer a verification from a change that was later reversed. Note the collection date separately from the request date.

Analysis and reporting

Report the number of requests, the share verified within the stated window, the share verified later, and the share applied without a completed check. Show the counts by channel, supplier type, and spend band. Present the median time to first contact and the median time to completed verification. Use the same denominator for every rate.

Inspect the requests that were applied before verification and describe the pattern without naming the supplier or the employee. A cluster on one channel may point to a missing rule, while a cluster at month end may point to pressure to pay. Compare the result with the day of week, because a Friday request has fewer business hours before the window closes. Report after-hours requests separately.

A useful additional view is the share of requests that were verified but never logged as complete. A callback without a recorded outcome can leave the record looking late when the work was done. Show the closure rate beside the verification rate and describe the difference. That view helps the owner decide whether the problem is speed or documentation.

Report the result in a short table with one row per channel and one column per measure so the owner can compare like with like. Add a sentence that states the denominator and the sample window beneath the table. A reader should not have to reconstruct either from the prose. Keep the raw counts available so a later review can test a different cut.

Limitations and sources of error

The record cannot show whether the person who answered the callback was authorized to confirm the change. It can show only that a call was made and what was recorded. A fast verification is not proof of a good control, and a late one is not proof of fraud. The protocol also cannot capture a check performed through a channel the business does not log.

Phone systems can mislabel the time zone, drop a call attempt, or record a duration without a connected party. A request that appears unverified may have been checked through a conversation that was never noted. Record those cases and their treatment instead of dropping them silently. If the same person requests and verifies a change, note the independence limitation and analyze that subgroup separately.

Review cadence and next action

Give the result to the owner and the person who manages the vendor master and ask them to choose one observable change. The change might be a required callback to a number already on file, a named verifier, or a hold on every change until the check is complete. Record the owner, the effective date, and the measure that will show whether the change worked. Avoid changing the verification window and the approval role at the same time.

Repeat the protocol with the same definitions after the change and compare the same fields. If the verification time fell but the share applied before verification did not, the team may be calling faster without waiting for the result. If the share improved but the closure rate fell, the new step may be adding work without a record. Close the review when the owner accepts the finding and the next sample is scheduled.

Sources

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