Research question and scope
This protocol asks what an assignment-end access review actually catches when a Philippines-based worker leaves a client engagement. The unit is one assignment closeout, and the outcome of interest is the difference between the access that was believed to be removed and the access that was still active at verification. The study does not measure individual trustworthiness or intent. It measures whether the offboarding process produced verifiable removal.
Set the scope to one client, one provider, or one category of systems and record the access model, the closeout checklist. The identity provider in use. If the team changes its offboarding workflow during the window, preserve the earlier definition for the full sample or analyze the change separately. An assignment that ends without a formal closeout is still in scope and should be recorded as a process failure.
Population and unit of analysis
Define the population as assignments that ended during the window, including planned ends, early terminations, and transfers to another client. The unit is the assignment closeout, identified by the worker, client, end date. Access set. A worker who held two unrelated assignments has two units if the access and closeout records are separate.
Record the system count, the number of shared accounts, the presence of administrator access, and whether the worker had access to customer or payment data. These strata matter because a single shared mailbox is not equivalent to production database access. Exclude assignments that never received system access. Record their count so the population is transparent.
Definitions that make the measure reproducible
Define access removal as the condition where a named reviewer confirms that the account or permission is no longer active in the source system. Define a residual access finding as any active credential, group membership, token, or shared password that the worker could still use after the agreed end date. Define a complete closeout as one with an owner confirmation for every access item, not merely a request submitted. Each definition should include a positive example and a boundary example so two reviewers can classify the same case consistently.
Distinguish account disablement from permission removal. A disabled account may still exist and be re-enabled, while a removed permission may leave a shared credential usable. Record the evidence type for each item: screenshot, system export, ticket closure, or owner attestation. Owner attestation alone is weaker than a system check, and the report should show the difference.
Evidence standard and data dictionary
For each closeout, capture the access inventory, the removal request, the verification method, the verification time, and the reviewer. Capture any shared credentials, API keys, or device tokens that were in scope. Code each item as verified removed, reported removed but unverified, still active, not applicable, or unknown. Keep the raw evidence link in the case file.
Add a field for the time between the end date and verification. A removal that is confirmed a month later leaves a real exposure window even if the final state is clean. Also record whether the worker returned equipment or local copies and whether a knowledge-transfer acceptance was completed. These are separate controls, but they belong in the same closeout review.
Sampling and collection
Draw the sample from the HR or assignment system that records end dates, not from a list of completed offboarding tickets. Record the export date, the filters, and the selection rule. Include assignments that ended early or were transferred, because those are often the cases where access removal is least orderly. If the population is small, review all of it and report counts rather than rates.
Verify access in the source systems rather than relying on the ticket status. If a system cannot be queried, record the limitation and use the strongest available evidence. Do not reconstruct a verification that never happened; mark it unverified. Note the collection date so the reader can see how much time passed between the end date and the review.
Analysis and reporting
Report the number of closeouts, the share with a complete access inventory, the share with verified removal, and the share with at least one residual finding. Separate administrative accounts, shared credentials. Customer-data access because the consequence differs. Show the time-to-verification distribution and the number of items that required a second request.
Inspect the residual findings individually and describe the system, the permission, and the reason it remained. Avoid naming individuals in a way that turns a process finding into a blame exercise. If the same shared account appears repeatedly, the problem is the access model, not a single closeout. Report that pattern with a specific recommendation.
Limitations and sources of error
The review can catch only access that the inventory knew about. Shadow accounts, personal cloud storage, and credentials shared outside the approved system may never appear. A clean verification therefore does not prove that no access remains. The protocol also cannot prove that the worker used or retained information, and it should not imply misuse without separate evidence.
System exports may be incomplete, and some identity providers retain disabled accounts for audit purposes. A disabled account is not the same as an active session or token, so the verification method matters. If the review happens long after the end date, the team may have removed access for other reasons, which weakens the attribution to the closeout process. Record these limits plainly.
Review cadence and next action
Give the result to the access owner and the client sponsor and ask for one change: add a system check to the closeout, remove shared credentials from the role, or shorten the time to verification. Name the owner and the date. If the finding involves a high-consequence system, treat it as an incident rather than a quarterly improvement item. The owner should also confirm whether the same finding appears in other assignments using the same access model.
Repeat the protocol after the change and compare the same access categories. If residual findings persist, inspect whether the inventory is incomplete or the removal process lacks authority. Close the review when the owner accepts the finding and the next verification is scheduled. The goal is a closeout that can be trusted without relying on a person's memory.
Sources
- [U.S. Department of Labor, Wage and Hour Division](https://www.dol.gov/agencies/whd) — primary guidance on hours, overtime, and recordkeeping.
- [Occupational Safety and Health Administration, recordkeeping](https://www.osha.gov/recordkeeping) — incident and exposure record requirements.
- [U.S. Small Business Administration, manage your business](https://www.sba.gov/business-guide/manage-your-business) — operating and workforce planning guidance.
- [NIST Cybersecurity Framework 2.0](https://www.nist.gov/cyberframework) — governance, access, and recovery language used in the access checks.
What the result can change
The result can change the access inventory, the closeout checklist, or the time allowed for verification. It cannot prove that no information was retained or that no misuse occurred. Those questions require separate evidence and, where appropriate, a security or legal review. The protocol should make that boundary explicit in every report.
If residual findings cluster around shared credentials, the fix is an access-model change rather than another reminder. If they cluster around late verification, the fix is a closeout deadline with an owner. Match the recommendation to the pattern instead of treating every finding as an individual mistake.
A worked reporting example
Suppose fifteen assignments ended in the window and twelve have a complete access inventory. Of those twelve, nine have verified removal and three have at least one residual finding. Report the counts and describe the systems involved, then separate administrative access from customer-data access so the reader can judge consequence.
The example does not establish a rate for the provider or the client. It shows how to present a small sample without overclaiming. The next step should be a specific verification change, such as a system check for the shared account, with a named owner and a date.
Reporting the limitation beside the result
State that the review covers only the access items in the inventory and the systems that could be queried. It cannot prove that no information was retained or that no credential was shared outside the approved system. If a system could not be checked, name it and describe the evidence that was available instead.
The report should separate a documentation gap from a security incident. A missing verification is a process finding. An active credential after the end date may require immediate response. The owner should know which category applies before deciding on the next action.
What the result cannot settle
The review cannot show whether a worker used access improperly or retained information after the end date. It also cannot show whether the client followed its own removal obligations. Those questions require a separate security or legal process, and the report should not imply a conclusion the evidence does not support.
A residual access finding is a signal to act, not a verdict about a person. The owner should decide whether the case needs immediate containment, a broader review, or a process change. The protocol gives that decision a factual starting point without replacing the judgment of the responsible team.
