Research ·
Fallback intake channels and duplicate records during a portal outage
A portal outage pushes customers onto channels with less structure, so the desk needs a tested fallback plan and a reconciliation step before duplicates pile up.

Key stats
intake channels covered by the fallback plan
duplicate checks before a record is merged
reconciliation pass after recovery
Methodology and findings
Question and scope
This review asks how an outsourced help desk should handle intake when the client's portal is unavailable and customers switch to another channel. The population is requests that arrive by email, phone, chat, or walk-in during a portal outage. The unit is one request and the record it creates across channels. The question matters because a fallback that works for the customer can still leave the desk with duplicate, incomplete, or unowned records. It treats continuity as a records problem, not only a communication problem.
The scope covers the desk's fallback instructions, the intake fields used on each channel, and the reconciliation after recovery. It excludes the portal's engineering recovery and the client's internal incident command. The review draws on incident response concepts in NIST SP 800-61 and contact-centre guidance in ISO 18295-1. It does not assume that every channel can perform every action, because identity checks and attachments often differ. Where a channel cannot meet the normal standard, the desk should record the gap rather than pretend it does not exist.
How a portal outage changes intake
A portal normally carries structure: required fields, file upload, status lookup, and a record that the customer can see. When it disappears, customers move to channels that carry less structure and more variation. Email arrives without a template, phone arrives without a transcript, and chat arrives without a ticket. The desk must replace the missing structure at intake or accept a queue of records that cannot be routed.
Volume also changes during an outage. Customers who would have waited may call, and customers who already had a ticket may open a second request because they cannot check status. The desk should expect a short spike and a second wave after recovery. Plan the fallback for both, not only for the first hour. A fallback that handles new requests but not status questions will still overload the phone line.
The duplicate record problem
Duplicate records are the most common failure of a fallback intake. The same customer emails, calls, and then opens a portal ticket when service returns. Each record may contain a different piece of the story, and none of them is complete. Without a matching rule, the desk works the same request three times and gives the customer three different answers. The cost is not only wasted effort but also inconsistent commitments.
Define what counts as a duplicate before the outage: same customer, same underlying request, and a time window. Use the customer identifier and the request description together, not one alone. A customer with two genuine requests should not be merged just because the messages arrived close together. Record the merge decision and keep the later record linked so no evidence is lost.
Choosing the fallback channel
Choose the fallback by what the request requires, not by what is easiest to advertise. A password reset may need a verified phone call, while a document upload may need a secure link. Publish a short table that maps request types to channels and states what each channel cannot do. Customers make better choices when the limits are visible before they start.
Test each fallback before an outage, including the acknowledgement the customer receives. A channel that accepts a message but sends no confirmation leaves the customer unsure whether the request arrived. Check who monitors the inbox or queue and how quickly. A fallback with no owner is not a channel; it is a place where requests go to wait. Record the test date and the person who confirmed the message arrived.
Keeping the customer informed during the fallback
Tell customers which channel to use and what information to include before they write. A short checklist reduces incomplete requests and repeated questions. Publish the same checklist on the portal error page, the phone greeting, and the social account. The checklist should name the request, the account identifier, and the best way to send evidence. Keep the checklist to five items so it is read rather than skipped.
Set an expectation for the fallback's response time, and keep it separate from the normal promise. A fallback queue may be slower, and saying so is more honest than letting the customer assume the usual service. Update the expectation when the queue changes. A customer who knows the wait is less likely to open a second request. Publish the expectation in the same place as the fallback instructions.
Identity and attachment limits on a fallback
An outage does not suspend identity checks, even when the usual verification tool is down. Decide in advance which requests can wait and which need an alternate verification path. Do not let urgency turn a callback to a new number into an approved method. Record the method used on every fallback request, because the normal audit trail may be missing.
Attachments are the other common gap. Email may reject a large file, and phone cannot accept one at all. Offer a tested secure upload or a scheduled drop-off, and record where the file is stored. If the file cannot be delivered, note what is missing and set a follow-up. A fallback request without its evidence is not ready for the queue.
Staffing the fallback queue
Move named agents to the fallback queue rather than asking everyone to watch every channel. A named owner can learn the workarounds and answer consistently. Give that person a backup and a supervisor who can approve exceptions. The arrangement should be written before the outage so it does not depend on who is online at that hour. Write the assignment in the coverage plan so it survives a shift change.
Track the fallback queue's age and volume separately from the normal queue. A fallback can look small while a few requests wait for days. Review the oldest record at each shift change and assign it. When the portal returns, return the agents to their normal work only after the fallback queue is reconciled. A daily count by channel shows whether the fallback is keeping up.
Reconciliation after recovery
When the portal returns, reconcile the fallback records before closing the incident. Match each fallback request to an existing ticket or create a new one with the correct channel and time. Check for customers who submitted twice and for files that arrived in a different system. The reconciliation is finished when every fallback request has one owner and one status.
Send a short confirmation to customers whose request arrived through a fallback. Tell them the record is now in the portal and give the ticket number if one exists. This message closes the loop and reduces status calls. It also gives the customer a simple way to correct a detail that the fallback channel could not capture. Ask them to reply only if a detail is wrong, so the thread stays short.
Reporting the outage period
Report the outage period separately from normal weeks. Mixing the two hides the effect of the fallback on volume, first response, and duplicates. Show the counts by channel and the number of records reconciled. Label the period on every chart so a reader does not compare it with a normal week. The client should see both the service impact and the recovery work.
Record how many requests arrived through each fallback and how many needed a correction. A high correction rate may mean the fallback instructions were unclear, while a low rate with long delays may mean the channel was understaffed. Use the numbers to choose which fallback to keep for the next outage. Keep the report short and attach it to the incident record.
Working with the client
Agree the fallback plan with the client before an outage, including who can approve a change during one. The client should name the person who can confirm which channels are open and which actions are paused. Put the plan in the service agreement or an attached schedule. An agent who has to invent a fallback during an outage will invent a different one on every shift.
Review the plan after every outage and after any change to the portal or the identity system. A fallback that depended on a feature the client removed is no longer a fallback. Update the customer-facing page and the agent runbook at the same time. Give the plan a review date so it does not wait for the next incident. Keep the review notes with the plan so the next shift can see what changed.
Limits and a practical standard
No fallback can replace a portal that holds the customer's record and status. The desk can control the instructions, the intake fields, the duplicate check, and the reconciliation. It cannot prevent every duplicate or recover a file that was never delivered. State those limits in the incident review so the team is judged on the controls it owns and can test.
A practical standard is a tested channel table, a required evidence set, a duplicate definition, a reconciliation step, and a separate outage report. Rehearse the plan once with the client and keep the approval contacts current. Review the fallback after each use and record what changed. Continuity is visible in the record, not in the claim that the desk stayed open.
Sources
- NIST SP 800-61 Rev. 3, Incident Response Recommendations -
- ISO 18295-1:2017 customer contact centres -
- NIST Cybersecurity Framework 2.0 -
- BLS Occupational Outlook Handbook: customer service representatives -