Philippines staffing blog ·
Write a working agreement for client-requested policy exceptions
A short agreement on who approves an exception, what evidence it needs, and when it expires keeps one-off decisions from becoming hidden policy.

Direct answer
Operational guidance
Define the exception in customer terms
A client-requested exception is a case where the normal policy would produce one answer and the client asks for another. Write the request in customer terms before discussing the decision, because a technical description can hide who is affected. State the rule, the requested change, and the outcome the customer expects. Everyone should be able to read the sentence and agree on what is being asked.
Separate a one-off exception from a policy change. A single customer with a documented reason may deserve a decision, while a repeated request usually means the policy is wrong. The agreement should say which one is being considered and who can decide. Mixing the two produces exceptions that quietly become the new rule without review.
Name the approver and the backup
Name one approver and one backup for each exception type, with the maximum value or risk they may accept. An approver without a limit will either refuse everything or approve too much. Put the limit in the agreement and in the ticket form, so the agent does not have to interpret it. The backup matters because exceptions rarely arrive when the primary approver is at a desk.
Record the approval channel and the response time the desk should expect. A verbal approval in a hallway is hard to verify and easy to dispute later. Ask the approver to reply in the ticket or a linked record with the decision and the reason. If the approver is unavailable, the backup's decision should carry the same record.
Require an evidence packet
Ask for the evidence that makes the exception auditable: the customer's request, the policy clause, the reason for the deviation, and the outcome. The packet does not need to be long, but it must let a later reader reconstruct the decision. Attach the evidence to the ticket rather than describing it in a summary. A summary written by the requester may omit the fact that mattered.
Decide in advance which evidence the desk may not collect, such as medical details or full payment credentials. Collecting more than necessary turns a support decision into a privacy risk. Ask only for the fact that changes the answer, and record who verified it. The agreement should name the minimum evidence, not the maximum, so an agent cannot demand everything in the customer's file.
Write the customer-facing wording
Write the customer message with the decision, the reason, and the next step in that order. Avoid explaining the internal approval process unless the customer needs it to act. If the exception has a time limit, state the date and what happens afterwards. A clear limit prevents the customer from treating the exception as a permanent change.
Give the agent a short list of phrases that must not appear, such as a promise that the exception will always be honoured. A well-meant reassurance can become a commitment the client never approved. Review the wording with the client contact before the first use. The approved version should live with the exception type, not in a personal note.
Track the exception separately
Tag the ticket with the exception type, the approver, and the date, so the case can be counted without reading every note. Keep the normal disposition separate from the exception outcome, because the underlying request may still be denied. A report that counts exceptions as standard resolutions hides the cost of the deviation. The tag is the smallest useful data point.
Review the exception log weekly with the team lead and monthly with the client. Look for concentration by customer, agent, or request type, since a cluster often points to an unclear rule. Do not treat every exception as a failure, because some are legitimate and well documented. The question is whether the pattern is intended.
Review the pattern, not only the case
When the same exception appears repeatedly, bring the policy back to the client with the evidence. Show the number of cases, the reasons, and the outcome the customers received. Ask whether the rule should change or whether the desk should apply it more strictly. A decision either way is better than an informal practice that nobody owns.
If the policy changes, retire the old exception path and update the saved replies in the same week. An exception that outlives its policy becomes a hidden rule that new agents cannot find. Record the retirement date and tell the agents who used it. The log should show a closed decision, not an abandoned one that appears in the next audit.
Close the exception when the policy changes
Send the customer a short note when a temporary exception ends, unless the client advises otherwise. The note should say what changes, when, and who to contact with a question. A customer who learns about the change from a failed request will call the desk anyway. Early notice turns a policy change into a managed transition.
Update the agreement with the date, the reason, and the replacement rule. Keep the old version available for audit but remove it from the agent's working view. The next exception request should start from the current policy, not from a document that still describes the old one. A closed loop is what makes the agreement trustworthy.
Authoritative sources
- NIST SP 800-53 Rev. 5 security and privacy controls
- ISO 18295-1:2017 customer contact centres
- 16 CFR Part 314 (FTC Safeguards Rule)
- NIST Privacy Framework