Automating return decisions with rules

When a return request arrives, someone has to decide: offer the customer an exchange, store credit, or go straight to a refund? In a small store, whoever looks at the request that day usually makes the call. As return volume grows, this approach becomes inconsistent, slow and impossible to explain. The short answer: automating return decisions with predefined rules that always give the same result ensures every customer in the same situation gets the same offer, every decision can be explained, and policy can be changed in a controlled way.

What is a rule that always gives the same result?

Such a rule always makes the same decision in the same situation. A rule such as "if the reason is size and the next size is in stock, offer an exchange" gives the same decision for the same return today and tomorrow. There is no randomness, no guessing and no mood of the day involved.

This matters for decisions with financial consequences. If you offered one customer bonus store credit, you need to be able to explain why another customer in the same situation did not receive it.

The problems with ad-hoc decisions

Return decisions not based on rules tend to suffer from the same issues:

• Inconsistency: two customers returning the same product for the same reason get different offers depending on who handles the request.
• Slowness: every request waits for a person to approve it, so queues grow in busy periods.
• No explanation: when you later ask "why was this customer offered credit?", there is no clear answer.
• No way to measure change: when you adjust your policy, you cannot tell whether a shift in results came from the change or from something else.

Rule-based automation answers each of these directly.

Why rules rather than AI?

Predictive models are useful in many areas, but they are hard to use as the decision-maker for return offers. A model may not give the same answer in the same situation every time, its reasoning is hard to show clearly, and the merchant cannot get a definitive answer to "why did this offer appear?" When an offer has a monetary consequence, explainability and repeatability are requirements, not preferences. Readable rules you can try out in advance are a sounder foundation for the decision path.

Example rules in plain language

A good rule set should be simple enough for a non-technical team member to read and understand. For example:

1. If the reason is size and the next size up or down is in stock, offer an exchange for that size first.
2. If the reason is "did not like it" and the order value is above a set threshold, offer store credit with a small bonus on top of the refund amount.
3. If the item is in electronics and the reason is "incompatible", offer an exchange for a compatible option of the same model first.
4. For discounted campaign items, do not offer bonus credit; offer an exchange and the standard refund.
5. For accounts that start many returns in a short period, do not offer incentives; show only the standard refund.

Two fixed boundaries should sit above all of these rules:

• The standard refund is always visible and one click away. Rules only decide which alternatives are offered in addition and in what order.
• No incentives when the merchant is at fault. If the item arrived defective, with missing parts, wrong or damaged in shipping, only a free replacement of the same item or a standard refund is offered, whatever the rules say.

A worked example

Suppose a rule offers 15 percent bonus credit on "did not like it" returns. For a 500 TL return the offer works out as 500 × 0.15 = 75 TL bonus, or 575 TL in store credit. If the customer accepts, they receive 575 TL of credit; if not, they choose the 500 TL standard refund in one click. Because the rule is fixed, every return with the same amount and reason gets the same offer.

Preview before publishing

Seeing the effect of a rule before changing it prevents the most common mistakes. A preview re-evaluates past return requests with the new rules and answers the question: "what would have happened if this rule had been live last month?"

What to look at in a preview

• How many returns would get a different offer?
• How does the mix of exchange and credit offers shift?
• How much does the total bonus credit go up or down?
• Is an unexpected category or reason group affected?

If the results match your expectations, you publish the rule. If not, you fix it before any customer is affected.

Rule versions and a record of every decision

Keep your rules in versions: editing a rule creates a new version rather than overwriting the old one. Every return decision should also be stored together with the rule version that produced it.

This gives you:

• Explanations after the fact: when a customer or colleague questions a decision, you can show which rule, based on which details, produced which result.
• Safe rollback: if a new version behaves unexpectedly, returning to the previous one is straightforward.
• Measurable change: comparing outcomes between two versions isolates the effect of a policy change.
• Transparent billing and accounting: every offer with a financial consequence leaves an auditable record.

A checklist to get started

1. Write down who makes return decisions today and on what criteria.
2. Turn those criteria into a few simple "if … then …" rules.
3. Define merchant-fault reasons and the visibility of the standard refund as fixed boundaries.
4. Preview the effect of the rules on past data and review the results with your team.
5. Publish the first version and store each decision with its rule version.
6. Review results regularly and prepare a new version when needed.

How rules work in Revoop

In Revoop, offers are decided by rules that always give the same result for the same situation; no AI is used in that decision. Merchants define their rules, can preview the effect of a change on past returns before publishing it, and every decision is recorded with its rule version. The standard refund always stays one click away, and no incentives are offered for merchant-fault reasons. See the FAQ for more.

Conclusion

Tying return decisions to rules is the most reliable way to treat customers consistently and fairly, protect your team's time and change policy in a controlled way. With plainly written rules, a preview before publishing and every decision stored with its version, the question "why did this offer appear?" always has a clear answer.