Triage NDRs by the action needed before the next delivery attempt

The NDR queue contains “customer unavailable,” “incorrect address” and “payment issue.” The team sends the same reattempt instruction to every shipment. Some customers are still unavailable, one address is incomplete and a payment dispute remains unresolved.

Triage non-delivery reports into owned next actions. The carrier's reason is the starting evidence, not necessarily the confirmed cause. The workflow should establish what must change before a further attempt is useful and permitted.

Keep the report and the investigation separate

Record the AWB or shipment identifier, attempt reference, reported reason, event time, receipt time, service, current destination and applicable action deadline. Preserve the original reason even if later evidence corrects it.

Shiprocket's NDR guidance describes reattempt, buyer contact and return-to-origin actions. Availability and timing depend on the actual account, service and carrier procedure; the guidance does not make every reattempt successful. Shiprocket — What is NDR?

Do not hard-code one universal number of attempts or response deadline from a generic article. Obtain the current service-specific terms and portal behaviour, and retain their source in the operating configuration.

A case should distinguish reported cause, verified finding, authorised instruction, carrier acknowledgement and final outcome. That separation makes both disputed reasons and failed instructions visible.

Twelve reports need different work

This illustrative queue is reviewed at 09:00 Asia/Kolkata on 15 October 2026. All twelve cases share a confirmed 12:00 action cutoff for this example only.

Six cases are ready for an instruction. Six remain unresolved: three customer confirmations, one address issue, one payment review and one disputed attempt.

The team has not “saved six RTOs.” It has prepared six supported actions. Delivery outcomes must be measured later against those same shipment identities.

Make each action appropriate to the evidence

For customer unavailability, record the confirmed receiving arrangement and whether the carrier can service it. A customer asking for 20:00 delivery does not establish that the selected service offers that window.

For an address correction, preserve the original and proposed address, the authorised request and serviceability result. A new postcode or materially changed destination can require rerating, a different service or another approved process. Updating the CRM does not prove the carrier's active shipment record changed.

For a payment dispute, verify the transaction through the authorised payment or finance record. A message saying “already paid” is not enough to instruct a carrier to stop collecting COD. Avoid asking the customer to expose sensitive payment credentials.

Delhivery's B2B exception guidance includes payment-mode discrepancies as a distinct operational issue. The appropriate action must follow the actual shipment's service and verified payment state. Delhivery — B2B exceptions management

For a disputed delivery attempt, preserve the reported event and request the relevant carrier evidence. Do not mark a driver or customer at fault solely from a reason code or an unverified message.

Prioritise cases that can still change the next attempt

Sort by the applicable deadline, remaining work, customer commitment and consequence. A high-value case with no verified address may need earlier human attention than a low-effort reattempt ready to submit.

Keep attempted contacts, responses and instructions separate. Restrict access to personal contact and delivery details to those handling the case. Use approved communication channels and retain only the evidence needed for the decision.

Reconcile the instruction with the carrier

A portal submission can fail, time out or be accepted too late. Store a stable case and action reference, payload version, submission time and response. Query uncertain outcomes before sending another contradictory instruction.

If a reattempt request is accepted and the customer subsequently cancels, create a new authorised action and resolve which instruction is current. Do not overwrite the first request without history or assume the second message reached the delivery team.

An RTO request is also a state transition requiring acknowledgement and later physical tracking. The parcel has not returned to stock merely because the request was sent.

Compare the control options

A structured queue can be sufficient when the team has manageable volume and clear ownership. Native carrier or aggregator tools reduce repeated entry when they expose action status and evidence. An integration is useful when multiple carriers use inconsistent reason codes, but it must retain the original code and each service's action rules.

A model may suggest a category from free text. It should abstain when the evidence is ambiguous and leave payment, address and return decisions to the authorised process. Measure category errors by their consequence; a mistaken payment instruction is not equivalent to a harmless routing suggestion.

Test an unavailable buyer, changed postcode, unverified payment claim, disputed attempt and lost submission acknowledgement. Include late responses that arrive after the case has moved to RTO.

Evaluate the actions and the eventual outcomes

Report cases with supported action before cutoff, carrier acknowledgements, unresolved reasons and later delivery or RTO outcomes. Keep the original intake population, including cases that never received a response.

Compare similar services and reason groups before attributing an outcome change to the workflow. A queue containing easier cases can show better delivery without better execution.

The immediate objective is an accountable next action for every NDR. The longer-term measure is whether those actions improve completed service without creating incorrect instructions.

Next actions

1. Map carrier reasons to the evidence and authority each decision needs.
2. Capture service-specific action deadlines and acknowledgement states.
3. Trial the twelve-case classification with unresolved cases retained.
4. Follow instructions through to delivery or return outcomes before claiming improvement.

For an NDR-workflow review, identify the reason group that consumes the most follow-up and the carrier response your team cannot currently verify.