Decide whether to wait, retry, or escalate a stalled onboarding step
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
A WhatsApp business onboarding step appears stuck. Someone refreshes the page, another person resubmits it, and a third suggests starting again. Now the team has more attempts but less certainty about which submission is active.
Choose the next action from the evidence, not from a universal waiting period. Wait when a submission is acknowledged and current guidance supports leaving it alone. Retry when a specific correction or documented retry instruction makes another attempt meaningful. Seek clarification or escalate when the state is contradictory, the outcome is unknown, or a published escalation condition has been met.
The operating method below applies across onboarding steps without assuming that every pending label means review, every error means rejection, or every delay has the same cause.
1. Identify the exact step before diagnosing the delay
“Onboarding is stuck” describes the business problem, but it is too broad to guide an action. Name the smallest step whose outcome you cannot establish: submitting business information, completing an account connection, confirming a prerequisite, or receiving a review outcome.
Then separate that step from what comes after it. A completed account setup does not establish that a template has been approved. Template approval does not establish that a message will be delivered. The current WhatsApp Business Messaging Policy is the source to consult for precise messaging requirements; an onboarding status is not a substitute for checking them.
Create a short incident header before anyone changes anything:
- Affected object: the relevant workspace, account, or submission identifier, shared only through an appropriate support channel.
- Blocked step: the specific action and expected confirmation.
- Observed state: the visible label, error details, or absence of confirmation.
- Business impact: what cannot proceed, without treating urgency as evidence of technical failure.
- Owner: one person who coordinates further attempts and records results.
Other people can collect evidence, but coordinated execution prevents overlapping changes from obscuring what happened.
2. Capture a state snapshot before another attempt
A useful snapshot distinguishes what the system shows from what the team believes it means. “The status has not changed” is an observation. “The review has failed” is an interpretation unless an authoritative outcome says so.
Record the observation time and timezone, the displayed status, any confirmation or reference identifier, and the last known change. Preserve relevant error details privately, removing credentials, tokens, and unnecessary personal information from material shared more widely.
Also capture the attempt history. For each submission or retry, note what changed and what response followed. If nobody knows whether the first submission was accepted, that uncertainty should remain explicit rather than being rewritten as a failure.
Use three working categories:
- Pending review or processing: there is evidence that the step was accepted and remains unfinished.
- Explicit failure: a visible response identifies an unsuccessful action, possibly with a correction or next instruction.
- Unclear status: acceptance or completion cannot be established, or different surfaces show conflicting information.
These categories describe the available evidence. They do not assign undocumented meanings to a provider’s labels.
3. Match the next action to current guidance
Check guidance for the exact step and the service responsible for it. Start with the current official instructions, relevant service notices, and the support route responsible for that part of the process. Provider-specific behavior should be checked with that provider rather than inferred from another service’s documentation.
Record the page consulted, when it was checked, and the instruction relevant to this incident. Look for three things: whether waiting is required, whether a retry is supported, and what condition warrants contacting support.
Elapsed time matters only in context. Keep both the original submission time and the most recent attempt time. Do not assume that a refresh, edit, or resubmission restarted a review unless current guidance establishes that behavior. If a published timeframe exists, verify its scope and starting event before applying it. If none is available, do not invent one.
A technical error can provide a more specific instruction. For example, HTTP 429 indicates rate limiting, and a response may include a Retry-After header. MDN’s explanation of HTTP 429 also notes that providers apply limits differently. A retry instruction associated with one request should not be treated as the processing deadline for the whole onboarding journey.
If guidance is missing or contradictory, the immediate task is clarification—not discovering an undocumented fix through repeated submissions.
4. Use this filled decision table
Fictional worked example: The following decision rows describe hypothetical incidents. Their states and actions illustrate a triage method, not actual provider messages, processing times, or retry allowances. Each row functions as a completed decision-table entry.
Row A: Pending review with an acknowledged submission
- Observed state: A business-information submission has a reference identifier and remains pending review. No failure is displayed.
- History: One submission is recorded; a later read-only status check shows no change.
- Guidance check: In this hypothetical case, the applicable instructions say to await the outcome. No applicable escalation condition has been met.
- Decision: Wait without resubmitting.
- Reason: There is evidence of acceptance, but no evidence that another submission is required.
- Next checkpoint: The owner will inspect the existing record at the team’s scheduled review and check for revised guidance.
- Change trigger: A new outcome, a published escalation condition, or conflicting evidence changes the decision.
Row B: Explicit failure with a supported correction
- Observed state: The submission reports a validation failure identifying a particular field.
- History: One failed submission; no accepted submission is recorded.
- Guidance check: In this hypothetical case, current instructions explain the accepted format and support resubmitting after correction.
- Decision: Correct that field and make a supported retry.
- Reason: The next attempt contains a relevant, documented change rather than repeating the same input.
- Next checkpoint: Record the response immediately after submission.
- Change trigger: If the same failure returns despite the documented correction, preserve the evidence and seek clarification rather than improvising further edits.
Row C: Unclear status after an interrupted request
- Observed state: The browser stopped responding during submission. A later page view does not establish whether the request was accepted.
- History: One attempted submission with an unknown outcome; no further writes.
- Guidance check: No applicable instruction establishes whether resubmission could create a duplicate.
- Decision: Seek clarification through the responsible support channel.
- Reason: A failed browser interaction is not proof that the underlying submission failed.
- Next checkpoint: Review the support response or an authoritative submission record.
- Change trigger: Confirmation that no submission exists, together with applicable instructions, can justify a new attempt.
The distinction is not “short delay versus long delay.” It is accepted and pending, explicitly unsuccessful, or unresolved.
5. Follow a completed incident timeline
Fictional worked example: A team cannot confirm whether a business-information submission completed. All times below are hypothetical event timestamps, not recommended waiting periods.
- Tuesday, 09:10 UTC — Attempt: Maya submits the information. The browser stops responding before a confirmation appears. She records the attempt as “outcome unknown,” not “rejected.”
- 09:16 UTC — Containment: Leo proposes submitting again. Maya becomes incident owner and pauses further writes until the original outcome can be established.
- 09:25 UTC — Evidence: Maya records the affected account identifier, the current screen, the interrupted request time, and the absence of a confirmation. A read-only check still does not establish acceptance.
- 09:40 UTC — Guidance: Maya checks the relevant current setup instructions. In this hypothetical case, they do not explain how to resolve this ambiguous outcome. She records that gap rather than assuming a retry is safe.
- 10:05 UTC — Clarification request: Maya sends the responsible support team the timeline and asks whether a submission exists and what action is supported.
- 14:20 UTC — New evidence: In this fictional scenario, support confirms that the submission exists and is awaiting review. Maya records the reference and changes the working category from unclear status to pending review.
- 14:35 UTC — Decision: The team leaves the accepted submission unchanged. The next internal review is scheduled for Wednesday at 09:30 UTC; that appointment is an ownership checkpoint, not a promised completion time.
The incident has not finished onboarding, but it has produced a useful outcome: uncertainty about acceptance is resolved, duplicate attempts are avoided, and the next action has an owner.
6. Make each retry a test with an observable result
A retry should answer a question. Before acting, write down what changed, why that change is supported, and what response would count as progress.
For the hypothetical validation failure above, the question is: “Will correcting the identified field according to current instructions allow this submission to be accepted?” The observable result is a new acceptance response or a specific failure—not simply that someone clicked again.
- Before: Preserve the prior error and confirm the correction applies to this step.
- During: Have one owner make the supported attempt without unrelated changes.
- After: Record the time, changed input category, resulting state, and any reference identifier.
- If unresolved: Stop the local experiment when it no longer has a documented basis. Seek clarification with the new evidence.
This is not a universal retry limit. Any actual limits or retry conditions must come from current applicable guidance. The operating principle is that each attempt should have a reason and produce interpretable evidence.
Avoid speculative recovery actions such as deleting the setup, changing unrelated account details, or creating replacement accounts. Without supported instructions, these actions can discard useful evidence and make the original problem harder to isolate.
7. Escalate with a question someone can answer
“Please fix onboarding” leaves the recipient to reconstruct the incident. A useful escalation states the unknown and the decision it blocks.
For the fictional interrupted submission, the request could read:
We cannot establish whether the business-information submission was accepted. The attempt occurred Tuesday at 09:10 UTC; the browser stopped responding before confirmation. We have made no further submissions. The attached record contains the affected account identifier and current status evidence. Please confirm whether a submission exists and whether we should leave it unchanged, correct something, or submit again under the applicable instructions.
Add the guidance already checked, previous attempts, relevant errors, and operational impact. Share sensitive identifiers only through the appropriate channel. Do not include passwords or access tokens.
Business urgency can justify asking for attention, but it does not prove that processing has failed. Keep those statements separate. Maintain one coherent case history where practical, and append new evidence rather than sending conflicting accounts through multiple routes.
8. End the triage with a recorded next action
Waiting has a cost: it may delay discovering a missing prerequisite. Retrying has a cost: it may introduce another uncertain attempt. Escalation also has a cost: a poorly documented case can generate another round of questions. The best choice reduces uncertainty without creating unnecessary new state.
Before leaving the incident, complete this decision record:
- State: pending, explicitly failed, or unclear, with supporting evidence.
- Action: wait, make a supported retry, or seek clarification/escalate.
- Basis: the relevant current instruction or the specific evidence gap.
- Owner and checkpoint: who will review it next and when.
- Change trigger: the outcome or new information that would justify another action.
Measure the next checkpoint by what you can establish: whether acceptance is confirmed, a correction produces a new response, an escalation receives an actionable answer, or an unknown remains explicitly documented. A quiet screen is not automatically progress or failure. A justified next action is the immediate goal.
Put the next conversation to work.
Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.
Explore Communications