1. Turn the proposal into a working brief
Record deliverables, exclusions, milestones, acceptance checks, client inputs, the approval owner, and the ongoing support arrangement. Give each missing input an owner and a due date. If it blocks delivery, make the dependency visible and revise the schedule together.
For every step, assign someone to do the work and someone who can approve it. Those may be the same person in a small team, but approval should never depend on guessing who has authority.
2. Collect access and assets once
Use an intake checklist for the agreed work: business details, brand assets, service boundaries, working hours, reply owners, approved copy, and relevant account access. Ask for what the project needs.
Have clients invite named users through platform access controls where available, with permissions appropriate to the work. Keep ownership and recovery arrangements clear. Do not collect passwords or private customer lists in an ordinary onboarding form.
| System | Client owner | Agency access | Review point |
|---|---|---|---|
| Business profile | Client retains ownership | Named manager for agreed updates | Project close |
| Client CRM | Assigned client owner | Account user for build and test | Handover |
| Brand files | Client team | Relevant project folder | Project close |
Add the work requiring access, whether the invitation was accepted, and when access should be removed. A request sent is not access verified.
3. Record the starting point
List what is live, who uses it, and how a request moves today. For a follow-up project, possible observations include real inquiries received, recorded replies, and booked appointments during a clearly stated period. Confirm the source before calling a number a baseline. Keep test records separate; an unavailable measure is unknown.
Agree who checks each number. A sent message is not proof of inbox delivery. A booking is not completed work. A pipeline estimate is not cash collected. These differences belong in the client report.
4. Test a complete path and its exceptions
Use approved test data and recipients. For an inquiry-to-appointment project, start with a permitted request, an appropriate reply, and a suitable booking. Then test what happens when the normal path breaks.
- New request: it arrives once and has a responsible person.
- Reply and escalation: the message reaches its intended destination and a person can take over.
- Booking: availability, time zone, and approvals match the client’s operation.
- Changes: cancellations update the record and stop obsolete reminders.
- Opt-out: applicable follow-up stops and suppression stays in place.
- Failure: an owner sees the issue and knows the fallback.
- Access: the client can perform the agreed tasks without unnecessary access.
These are proposed acceptance checks, not a claim that a client account has passed. Record actual results, evidence dates, and unresolved issues. Our controlled email test illustrates how to separate an observed result from a broader claim.
5. Hand over an operating guide
Provide a short walkthrough, the tested process, known limits, monitoring responsibilities, and the support route. Record which open issues prevent release and which the client accepts. Include ownership, access removal, and asset handover when the engagement ends.
The handover note
- What changed, and how the client uses it.
- Checks completed and the date of the evidence.
- Open issues and the person responsible.
- Monitoring, response expectations, and support boundaries.
- Next review, access review, and agreed fallback.
Keep reporting useful after the handover
Set a review cadence that matches the scope and the client’s decision cycle. A report should answer what happened, what remains uncertain, and what happens next. Explain the reporting window and source, any tracking gaps, and whether test activity was excluded.
- Delivery: which agreed work shipped, and what is blocked?
- Process: where did requests, approvals, or replies get stuck?
- Outcomes: which results are supported by the data, with what limits?
- Next decision: continue, fix, reduce, or expand a specific part of the service.
Agree on an owner and date for each action. Track repeated support issues and time spent against the service allowance. Recommend ongoing work when there is a clear job to do; a recurring invoice needs a recurring responsibility.
When access breaks: a release and recovery checklist
A connection is a dependency: another account or service your work relies on. An invitation accepted last month does not prove today’s access works. Before handover, have the responsible person open the correct client account and perform the agreed task with their own authorized login. Do not ask for their password or verification code.
| Check | Evidence to record | If it fails |
|---|---|---|
| Account and ownership | Client account identifier, client owner, named agency user and agreed role | Ask the owner to confirm the intended account and grant only the required access |
| Task access | Last successful authorized read or action, its timestamp and time zone | Record unavailable; do not replace a missing result with zero |
| Connected service | Correct destination, current connection status and an approved end-to-end check | Owner reconnects through the service; then repeat only the affected checks |
| Recovery and coverage | Who receives an alert, agreed response window, backup owner and permitted fallback | Use the agreed escalation route; a monitor cannot restore permissions by itself |
| Offboarding | Access review date, assets to return and person authorized to remove access | Resolve ownership and handover before removing access needed for the transition |
Keep this register in a private client workspace. Store evidence references and safe account identifiers, not passwords, recovery codes, tokens, customer message bodies or public links that expose private records. A connection-status badge is one signal; it does not prove a particular email arrived or a booking was recorded.
Worked example: the dashboard opens, but the client mailbox does not
Hypothetical, not a customer result: an agency is handing over a five-email sequence. Its authorized workflow history shows a send, but the client’s mailbox login is unavailable. Mark native send evidence as observed and inbox receipt as unknown. Do not mark delivery passed, resend the series or enroll a real customer just to fill the gap.
- Record which check is blocked, the last successful observation and the next check time.
- Ask the mailbox owner to restore their own access through the provider’s normal process. Do not broaden unrelated account permissions.
- After access returns, inspect the already authorized test’s actual receipt. Record the receipt time separately from the send time.
- If evidence is still missing, investigate the affected path and agree on the release decision. Any new controlled test requires appropriate permission and must remain separate from customer totals.
Release decision: hold the affected acceptance criterion when required evidence is missing. If work can proceed safely within a narrower scope, write down that scope, the limitation, the approving client owner and the review date. Client approval does not turn an unknown result into a pass.
Copy this exception note into the handover
Download the access-readiness and exception worksheet. It works with a shared document or your existing project tool; no signup is required.
- Blocked check: what cannot currently be verified, and what work depends on it?
- Known evidence: source, actual observation time, time zone and test/customer distinction.
- Decision: hold, proceed within an approved limited scope, or restore and recheck.
- Owner and due time: who acts, who approves, when to escalate and what the fallback is.
Agree on this coverage before promising ongoing monitoring. A scheduled local check requires the computer, application and relevant authentication to be available. Missing access means missing coverage; silence is not a healthy result. This is a proposed operating method, not a claim that it prevents every failure.
Where HighLevel can help
HighLevel is one platform agencies can use for customer records and automated workflows. It distinguishes agency-wide access from selected client accounts, called sub-accounts. For a person who only needs selected client accounts, its Account user type can be assigned those sub-accounts; module and supported action permissions further control the work they can perform. Choose access around responsibilities.
A snapshot is a reusable copy of selected configuration. HighLevel’s snapshot documentation distinguishes these assets from live records and third-party connections, which do not transfer. Reconnect required services and verify client-specific settings after loading. A template still needs a client-by-client acceptance test.
The onboarding process also applies with other tools. Check account requirements and budget communication and software usage for the work you agreed.
Sources and working template
Access and snapshot facts rechecked October 22, 2026 against HighLevel’s agency permissions, client-account permissions, snapshot overview, and Google’s profile access guidance. The stages and checks above are our proposed working templates.
Download the agency worksheet or check the economics of your delivery plan.