AI refund decisions and returnsYour refund policy.
Applied the same way every time.
Your AI assistant checks each return or refund request against your written rules: the return window, the item's condition, the reason and the order itself. It tells your rep what the policy says and drafts the reply. Issuing the refund stays with a person.
Find the order date. Count the days. Reread the policy. Check past refunds. Ask the lead. Wait for an answer. Write the reply. Or just ask.
Does this return qualify, and what can we offer?
Your assistant looks up the order and the customer's history, then walks through your refund and return decision tree: inside the window or not, which reason, which condition, any exclusions. It answers with the outcome your policy gives (refund, exchange, store credit or no) and names the rule that decided it. Then it drafts the reply in your support voice for the rep to check and send.
The decision and the reason, not just a yes.
Every answer names the rule behind it, so your rep can check it and your lead can see why. Two reps asking about the same order get the same answer. When a case falls outside your rules, your assistant says so and points to the person who decides.
How to write a refund decision tree. Six questions, in order.
Most refund arguments start because the rule lives in someone's head. Write it down as questions with yes or no answers, in the order you'd ask them. The same tree works for a new rep, a busy lead or your AI assistant.
- 01 Is it inside your return window?
Decide whether you count from purchase or from delivery, and write it down. If it’s outside, go to your late-return rule, for example store credit or a polite no.
- 02 What is the reason?
Damaged, wrong item, doesn’t fit, changed mind. Each can lead somewhere different. Many stores let damage and their own mistakes skip the window.
- 03 What condition is it in?
Unused with tags, used or missing parts. Say what “unused” means in plain words.
- 04 Is the item excluded?
Final sale, custom-made, personal care or gift cards. List them by name or product tag.
- 05 Is the amount over your limit?
Above a set amount, a named person approves. Below it, your rep can go ahead.
- 06 Has this customer had refunds before?
Decide what counts as a pattern and who looks at it. That should be a person, not a rule.
Each path should end in one outcome: refund, exchange, store credit, a polite no or a hand-off to a person. If a path ends in “it depends”, that branch needs another question.
Money back is a person's call.
Issuing the refund
Your assistant says what your policy allows. A person issues the refund or exchange in your store or payment app.
Goodwill and exceptions
Saying yes outside the rules for a loyal customer is a judgment call. The tree simply tells you when you’re making one.
Patterns and disputes
Repeat refund requests, chargebacks and threats go to a named person, set out in your escalation rules.
Your rules, written down. Then followed every time.
Set up for you as part of the Customer service package, typically in one to two weeks. No technical staff needed: one person who knows how the work flows, plus logins to your apps.
What your assistant learns
Rules we write down with you
Shortcut · for this job
Reply drafts the answer when the rules settle it. When the tree says a person must decide, use Hand off and it writes the note for whoever decides.
Works with
App names are trademarks of their owners. Figure Hawk is not affiliated with or endorsed by them.
Can AI decide refunds?
It can apply your refund rules and tell you what they say, which is most of the work. The decision to give money back stays with a person. In our setup, your assistant recommends and drafts; your rep or lead approves and issues the refund.
What is a refund policy decision tree?
It’s your refund policy written as a short list of yes-or-no questions in a fixed order, each path ending in one outcome. It stops two reps giving two different answers to the same request. The six questions on this page are a good place to start.
What happens when we change our return policy?
You update the document and your assistant uses the new version. Your SOPs refresh every night and old versions are kept, so you can always see which rule applied to a past decision.