How to Use Approval Workflows for Refunds, Returns, and Exceptions
A customer asks for a refund. Your agent thinks it's reasonable. They don't have the authority to approve it. So they email their manager. The manager is in meetings all afternoon. The customer waits. By the time the refund comes through the next day, they've already left a one-star review.
Welcome to the approval-process tax — the hidden cost of doing things safely but slowly.
Approval workflows exist to solve this exact problem. They embed the approval process directly into your helpdesk, ensuring the right person signs off on refunds, returns, and policy exceptions without creating a customer-experience bottleneck. Here's how to set them up, when to use them, and how to avoid the common traps that turn "quick approval" into "sorry for the delay."
When approval workflows actually pay for themselves
Not every support action needs a sign-off. Approval workflows add value only when they're protecting something that needs protecting:
Financial impact. Refunds, credits, discounts, or compensation above a threshold need oversight. The Consumer Rights Act 2015 in the UK and FTC consumer protection rules in the US make clear that refunds aren't optional — but how much discretion to give frontline agents? A workflow keeps that decision consistent.
Policy exceptions. When a customer's situation doesn't fit your standard terms, someone with authority should review it before the agent improvises.
Legal or compliance matters. Data deletion requests (the ICO has detailed guidance on individual rights), regulatory enquiries, or legal threats — these should always be reviewed before you hit send.
High-value accounts. Your largest customers often warrant an extra layer of eyes, even if the action itself falls within agent authority.
Irreversible actions. Account deletion, data purging, permanent bans — these should require confirmation from someone beyond the frontline agent.
The key principle: approval workflows exist to prevent bad decisions, not to slow down good ones. If you're requiring approval for every £10 credit, you've made the bar too high.
How the workflow moves the ticket forward
When an agent identifies a ticket that needs approval, they trigger a workflow. Here's what happens:
The request. The agent selects an approval option (or the system flags it automatically based on rules — e.g., "refund over £100" automatically routes for review). They include the amount, their recommendation, and the key context. Done right, this takes 30 seconds.
The routing. The workflow sends the request to the right person based on rules you define: refunds under £100 go to the team lead, refunds over £100 to the support manager, policy exceptions to the department head, anything legal to legal. If the primary approver is unavailable, it routes to a backup. This is critical — a single point of failure (one person on holiday) can stall dozens of tickets.
The review. The approver sees the request in context. They can read the full conversation, the agent's reasoning, and supporting info. Then they approve, deny, or ask for more detail.
The resolution. Once a decision lands, the workflow notifies the agent. The agent then either processes the approved action (issuing the refund, authorizing the return) or explains the denial to the customer and offers alternatives.
The record. Every step gets logged — who requested, who approved, the timestamp, the reason. This audit trail is non-negotiable for financial compliance and internal governance. It also becomes a training tool: when you review denied requests, you learn where agents' judgment needs sharpening.
Designing your workflow for speed
Set clear thresholds, then stick to them
Define which actions require approval and at what level:
| Action | Threshold | Who approves |
|---|---|---|
| Refund | Under £50 | Agent only |
| Refund | £50–£500 | Team lead |
| Refund | Over £500 | Support manager |
| Policy exception | Any | Team lead |
| Return authorization | Unusual circumstances | Team lead |
| Account deletion | Any | Manager |
Agents given clear authority to handle routine refunds independently move faster and feel more empowered. The HBR research on customer service actually shows that frontline discretion — within clear bounds — is one of the strongest predictors of loyalty.
Set SLA targets for approvals themselves
An approval workflow is only as fast as the slowest approver. Define decision targets: routine approvals (refunds under £500) within 2 hours, complex ones (policy exceptions) within 4 hours, urgent matters (escalated customer, legal) within 1 hour. If the approver hasn't responded within the window, escalate automatically to a backup. Don't let approval requests become black holes.
Build escalation into the workflow from day one
Primary approver unavailable? Route to secondary. Still no response in 2 hours? Route to the department head with an alert. This prevents one person's absence from tanking your response times.
Update the customer while they wait
The customer should never experience radio silence during an approval. As soon as the agent triggers the workflow, they should tell the customer: "I've submitted your request for review by our team. You'll hear back within [timeframe]." When the decision comes through, follow up immediately. This transforms "why am I waiting?" into "they're taking this seriously."
Real-world scenarios
Refund requests. Customer requests a refund. Agent checks policy. If it's within agent authority, process it immediately. If it's above threshold, trigger the approval workflow with the amount and justification. Approver reviews. Agent then processes the refund or explains the alternative.
Return authorisations. Customer wants a return. Agent checks eligibility (timeframe, condition, standard policy). If standard: issue return authorization instantly. If outside standard (late return, heavily used product), trigger approval. Approver decides. Agent either approves and provides instructions or communicates the standard policy position and any alternatives. For a deeper dive on handling these moments without frustrating the customer, see our guide on handling ticket escalation without frustrating customers.
Policy exceptions. Customer's situation doesn't fit your standard terms. Agent explains the standard position, then triggers approval with the specific reasoning. Approver reviews the circumstances. If approved, agent communicates the exception and documents it. If denied, agent explains the standard policy and explores other options (discount, store credit, next-time gesture).
Use approval workflows alongside other tools
Approval workflows work best as part of a larger support system. Automated ticket routing can send refund-request tickets to experienced agents in the first place, reducing the need for approval. Automation rules and triggers can flag certain tickets as "approval needed" automatically, so agents don't have to remember. Canned responses and macros let agents respond to common refund scenarios instantly, reducing back-and-forth. AI copilot can draft the approval request on behalf of the agent, reducing friction.
These tools together create a system where approvals happen fast because the groundwork has been done.
Measuring what matters
Approval request volume. Are more requests coming in over time? Growing volume might signal that your approval thresholds are too low — agents could handle more themselves.
Approval time. How long from request to decision? This directly impacts customer wait time. If your target is 2 hours and you're hitting 6, something's broken.
Approval rate. What percentage of requests are approved? If it's above 95%, your threshold is probably too low. If it's below 50%, agents might need better training on the policy.
Denial reasons. When requests are denied, why? Common reasons become training moments.
Customer impact. Compare CSAT scores for tickets that went through approval versus those that didn't. You might find that a quick approval (even if it takes 4 hours) beats a slow no-approval process.
Common pitfalls to avoid
Thresholds too low. If agents need approval for a £10 credit, you've built a bottleneck, not a safeguard. Trust your agents within reason.
No backup approver. One person on holiday shouldn't stall your entire refund queue.
Hiding the process from the customer. Silence during approval kills trust. Always communicate that a decision is being reviewed and when they'll hear back.
Skipping the audit trail. Approvals without documentation create compliance risk and training blind spots. Log the who, what, when, and why.
Making the request too complex. If requesting approval requires filling in a ten-field form, agents will avoid the workflow. Keep the request structure simple and fast.
Frequently Asked Questions
Q: How do I decide whether to use automatic routing or manual approval triggers? A: Use automatic routing for high-volume, clear-cut scenarios (e.g., all refunds over £100). Use manual triggers for judgment calls (policy exceptions, complaints). Combine both if you can — automatic routing for obvious cases, manual override for edge cases.
Q: What if an approver rejects a request without good reason? A: Review the denial reasons regularly. If a manager is rejecting legitimate requests, that's a training or authority issue. Approvers need clear guidelines on what "good reason" looks like.
Q: Should we notify the customer if an approval is denied? A: Always. Explain the decision, cite the relevant policy, and offer alternatives if available. A transparent "no" is better than a silent disappearance.
Q: How do we handle urgent refunds that need approval? A: Define an "urgent" lane in your workflow with a shorter SLA (e.g., 1 hour). Use it sparingly — if everything is urgent, nothing is.
Q: Can the same request go to multiple approvers at once? A: Avoid parallel approval if you can. Sequential approval (agent → lead → manager) is clearer. Parallel approval creates confusion about who's accountable.
Q: What if a refund is approved but the customer has already churned? A: Still process it. You're honouring your policy. The refund might even win them back. Plus, the customer can't say you refused — they'll remember that you eventually made it right.
Q: How do we audit approval workflows? A: Review the audit log monthly. Track who approved what, how long it took, and whether the decision was correct in hindsight. This feeds into agent training and policy refinement.
Q: Can agents see who approved or denied their requests? A: Yes, and they should. Transparency builds trust. If approvers are anonymous, agents can't learn from feedback.
The balance
Approval workflows exist to solve a real tension: you need to protect the business from unauthorised actions, but you also need to serve the customer with speed and decisiveness.
The best approval processes are almost invisible. The customer experiences a brief, communicated pause while the team confirms the right course of action. Then the resolution arrives — correct, authorised, and documented.
The worst approval processes create visible friction. The customer waits, asks for updates, receives vague responses, and eventually gets a decision that took far longer than it should.
The difference isn't whether you require approvals. It's how efficiently your workflow moves them through the system.
If you're managing this across multiple support channels or scaling your team, a modern helpdesk with built-in approval workflow automation removes the manual back-and-forth entirely. The ticket moves, the decision is logged, the customer stays informed — all without anyone checking email.