Queued Is Not Delivered: Decide What Each Campaign Status Allows You to Do
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
A campaign marked queued is not evidence that its messages reached recipients. A submission acknowledgement is not a delivery confirmation. And a missing delivery update is not, by itself, proof of failure. Treating these states as interchangeable creates two avoidable problems: reporting success before you can substantiate it, and resending messages that may already have arrived.
The practical answer is to attach an evidence requirement and an action rule to each status. Keep internal processing separate from external delivery evidence, classify uncertainty explicitly, and authorise retries only after checking what happened to the original attempt.
The model below is a recommended operating design, not a description of any particular provider's status vocabulary. Map it to the documented meanings of the evidence available in your workspace. Where that mapping is unclear, preserve the uncertainty rather than inventing certainty.
1. Make the status describe evidence, not optimism
A useful operational status answers two questions: what do we know about this message attempt, and what does that knowledge allow us to do next?
Use these five evidence categories:
- Queued: an internal record shows that an attempt is waiting for processing.
- Submission acknowledged: the external service has acknowledged the submission in a way its documentation defines, but delivery has not been confirmed.
- Delivered: a documented delivery confirmation can be matched to the relevant message attempt.
- Failed: documented evidence establishes that the attempt failed at a particular stage.
- Unknown: available evidence cannot reliably establish the attempt's current outcome.
These categories are not necessarily a simple sequence. An attempt can move from queued to a confirmed pre-submission failure. A submission timeout can leave an unknown outcome. A delayed delivery confirmation may resolve an unknown attempt later.
Do not overwrite the underlying facts when changing the summary status. Preserve the event, its source, its event time if available, when your operation received it, and the attempt it belongs to. That makes later reconciliation possible without pretending the team knew the outcome earlier.
2. Use this filled status decision table
Fictional worked example: a team is sending a service update. The following plain-text decision table uses hypothetical records and operating thresholds. These are examples to adapt, not provider definitions or recommended universal waiting periods.
Status | Evidence required | Permitted action | Claim operators must not make QUEUED Evidence: Internal job Q-101 exists; no dispatch or external acceptance evidence; dispatch check shows it remains pending. Action: Inspect queue age and blockers. Cancel only through a verified pre-dispatch cancellation process. Do not claim: The provider accepted it, or the recipient received it. SUBMISSION ACKNOWLEDGED Evidence: External acknowledgement for attempt A-102, with a matching message reference; documentation identifies this as submission acceptance, not delivery. Action: Track delivery evidence and investigate overdue updates. Do not resend solely because delivery is unconfirmed. Do not claim: Delivered, received, or completed successfully. DELIVERED Evidence: Documented delivery confirmation matched to attempt A-103 and its intended recipient record. Action: Count this attempt as confirmed delivered; exclude it from delivery-recovery resends. Do not claim: The recipient read, understood, or acted on the message. FAILED Evidence: Attempt A-104 has a documented terminal rejection for an invalid destination; no evidence of acceptance for this attempt. Action: Hold retry, correct or exclude the destination, then review permission and authorise any new attempt. Do not claim: Every failure is retryable, or every attempt to this recipient failed. UNKNOWN Evidence: Attempt A-105 timed out after dispatch began; no acknowledgement or delivery confirmation is available. Action: Reconcile using available records and documented lookup options. Require review before a replacement attempt. Do not claim: It failed, it was never sent, or resending cannot duplicate it.
Keep this table beside the campaign runbook. Add a named owner and a review trigger to each category. For example, the hypothetical team assigns unknown attempts to its operations lead for review after 30 minutes. That threshold starts an investigation; it does not transform uncertainty into failure.
The failed row deliberately names the stage and reason. A failure before submission is operationally different from a failure reported after acceptance. If a system displays only failed, inspect the underlying evidence before deciding whether a new attempt is appropriate.
3. Track attempts separately from recipients and campaigns
A campaign-level label is too coarse to support many resend decisions. One campaign can contain queued attempts, confirmed deliveries, and unresolved submissions at the same time. One recipient can also have several attempts with different outcomes.
Maintain a record for each attempt using whatever authorised records your tools make available. Useful fields include:
- Campaign reference and intended recipient reference.
- Internal attempt reference and external message reference, when available.
- Current evidence category and the evidence supporting it.
- Dispatch, acknowledgement, and delivery event times where available.
- Last reconciliation time, responsible operator, and next action.
- Any replacement attempt reference and its authorisation reason.
Separate the latest observed event from the inferred current state. A late acknowledgement should not automatically replace an already verified delivery confirmation merely because it arrived later. Conversely, contradictory evidence should trigger review rather than an improvised status precedence rule.
For conflicting events, first check whether they concern the same attempt. A delivered first attempt and a failed replacement are not contradictory. Collapsing them into one recipient-level status can conceal a duplicate send or falsely suggest that no message arrived.
If your tools cannot retain all of these fields, keep a controlled reconciliation record with the minimum needed references. Do not assume a particular interface or export exists; check your workspace and current plan for available records and controls.
4. Separate processing completion from delivery outcomes
Processing complete can be a useful operational statement, but only if its meaning is explicit. It might mean that every planned job has been examined and no internal queue work remains. That does not establish delivery for every recipient.
Use separate reporting lines for processing and outcome. A campaign can be finished internally while delivery outcomes remain unresolved.
Fictional worked example: a campaign contains 1,000 initial message attempts. At a stated reporting cutoff, its mutually exclusive evidence categories are:
- 80 queued.
- 120 submission acknowledged, without confirmed delivery.
- 700 confirmed delivered.
- 60 failed.
- 40 unknown.
A defensible report says: At the reporting cutoff, 700 of 1,000 initial attempts have confirmed delivery. A further 120 have submission acknowledgement only; 80 remain queued, 60 have failed, and 40 require reconciliation.
It should not say that 920 messages were delivered by adding queued, acknowledged, and delivered counts together. Nor should it call all 300 unconfirmed attempts failures.
State the denominator and cutoff whenever presenting a delivery percentage. In this hypothetical example, confirmed delivery is 70% of initial attempts at that cutoff. If replacements are later added, distinguish total attempts from intended recipients so retries do not inflate the apparent reach.
Administrative closure is another separate decision. A team may close an investigation with unresolved outcomes after recording its rationale and residual uncertainty. Label that as closed with unresolved outcomes, not fully delivered.
5. Treat unknown as a reconciliation task, not a resend list
Unknown is useful because it prevents a missing signal from becoming a false conclusion. Common reasons include a timeout after dispatch began, an unmatched external reference, a gap in event collection, or conflicting records.
Use a bounded investigation:
- Locate the attempt. Establish its internal reference, recipient, dispatch time, and any external reference.
- Check dispatch evidence. Determine whether the message demonstrably remained internal or whether external submission may have occurred.
- Reconcile available sources. Compare authorised logs, status records, exports, or documented lookup options, if available.
- Check for delayed or mismatched events. Look for evidence attached to another campaign or attempt reference before creating a replacement.
- Record the decision. Resolve the status when evidence supports it, or retain unknown with a named owner and next review.
Do not create a rule that unknown becomes failed after an arbitrary duration. Time without confirmation may justify escalation, but it does not establish non-delivery.
There is a real trade-off: waiting can make a time-sensitive update less useful; resending can send the same message twice. Make that decision explicitly according to the consequence of a duplicate and the consequence of delay. For a high-consequence message, escalate the unresolved case to a human rather than letting a generic retry rule decide.
6. Put an authorisation gate before every retry
A retry is a new operational action, not a harmless status correction. Before authorising it, answer three questions: is there adequate evidence that another attempt is needed, has the cause been addressed, and is contacting the recipient still appropriate?
For a queued attempt, first establish whether it can still leave the queue. Creating a replacement while the original remains eligible for dispatch can produce two sends. If cancellation is available, verify its documented effect and the attempt's state rather than relying on the fact that someone requested cancellation.
For a failed attempt, classify the reason. Correct a destination problem before considering another submission. For a temporary rejection, use the documented retry guidance rather than repeatedly submitting unchanged work. MDN's guidance on HTTP 429 explains that this response indicates rate limiting and that a Retry-After header may specify when to retry. Providers differ, so that signal must be interpreted in the context of the affected request; it is not a campaign-wide delivery verdict.
For an acknowledged or unknown attempt, check the original outcome before replacement. Where the tooling supports documented duplicate-prevention controls, use them as designed, but do not assume those controls exist or cover every resend path.
Permission and message eligibility also need another check. Honour opt-outs received since the first attempt. Under the WhatsApp Business Messaging Policy, approved templates are used to initiate Platform conversations and outside the 24-hour customer-service window, which opens or resets with a user message. Template approval is not a delivery guarantee. A previously prepared message should not bypass those checks merely because an operator calls the next send a retry.
7. Recover from premature reports and duplicate attempts
If a report has already treated acknowledged submissions as deliveries, correct the classification rather than silently changing the numbers. Publish the reporting cutoff, the corrected confirmed-delivery count, and the unresolved categories. Explain that the earlier figure represented a different stage of processing.
If unnecessary replacements have already been created, pause further recovery sends for the affected group while reconciling original and replacement attempts. Preserve both records. Where a verified cancellation process is available, use it for work that has not dispatched; do not promise recall of messages already submitted or delivered.
Then identify the decision rule that failed. Was queued labelled sent? Did a timeout trigger an automatic replacement without review? Did a campaign summary hide attempt-level uncertainty? Change that rule and test it against the affected records before restarting the recovery process.
The observable outcome should be a clearer account of what happened: which recipients have confirmed delivery, which attempts remain uncertain, and which replacements were authorised. A cleaner dashboard alone is not evidence of a better delivery outcome.
8. Run one status audit before the next campaign
Start with a small, explicitly scoped review. As a hypothetical exercise, inspect 25 recent attempt records covering all five categories. For each record, ask whether another operator could reproduce the status from the evidence and reach the same permitted action.
Record three measures:
- Evidence coverage: reviewed attempts with enough supporting evidence to reproduce their classification.
- Unresolved age: how long each unknown attempt has remained unresolved and when it was last reviewed.
- Retry traceability: replacement attempts with a recorded reason, original-attempt reference, and authorisation.
Assign an owner to every gap before the next send. The immediate objective is concrete: no delivery claim without delivery evidence, no unknown attempt silently counted as failed, and no replacement attempt without a documented decision about the original.
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