Skip to content
Documentation
All docs

Core Concepts

Automating Your Response

Auto-hold high-risk returns, auto-approve low-risk ones, and route everything else for review.

Workflows are if-this-then-that rules that fire automatically whenever RefundSentry scores a return. They're how you turn risk scores into action without sitting in front of the dashboard all day.

Anatomy of a workflow

Every workflow has three parts:

  1. Trigger — When to evaluate the rule. Today there's one trigger: "Return scored." Future triggers (order created, customer tagged, chargeback filed) will appear in the same builder.
  2. Conditions — A set of filters the return must match. Common filters: risk zone, score range, return value, customer's previous return count, product tags, return reason cluster.
  3. Actions — What to do when the conditions match. You can chain multiple actions in one workflow.

Available actions

  • Hold the fulfillment — Pauses the return so a human can approve or deny it. The merchant sees a hold reason and the signal breakdown that triggered it.
  • Tag the customer — Adds a custom Shopify tag to the customer record. Use this to feed your existing Flow rules or segment the customer in Klaviyo / your CDP.
  • Tag the order — Same as above, but on the order rather than the customer.
  • Send a Slack alert — Posts to a channel via webhook. Useful for ops teams who already live in Slack.
  • Send an email — Notify a specific address (the merchant, the warehouse manager, the customer service lead). One email per return; never a digest from inside a workflow.
  • Add an internal note — Drops a note onto the return for the next person who reviews it. The note can reference any signal that fired.

Default workflows

Every new shop is seeded with three opinionated defaults that you can edit or delete at any time:

  • Auto-hold high-risk returns — When score ≥ 66, hold the fulfillment and add an internal note with the top 3 signals.
  • Tag medium-risk customers — When score is 31-65, add the refundsentry:medium-risk tag so your team can flag them in Shopify admin.
  • Slack alert on chargeback predictions — When the chargeback prediction score crosses the threshold, post to #fraud-alerts. (Disabled by default — connect Slack first.)

Building your own

Open Automation → Workflows and click New workflow. The visual builder walks you through trigger, conditions, and actions in that order. You can preview the workflow against the last 30 days of returns before saving — the preview shows you exactly which historical returns would have triggered the rule and what action would have been taken.

Start narrow. The most common mistake is writing one giant workflow with five conditions and three actions. Two simple workflows are easier to debug than one complex one.

Common patterns

  • Wardrobing defense — Risk zone HIGH + return reason cluster ≠ "defective" + days since delivery ≤ 5 → hold fulfillment + Slack alert.
  • Serial returner triage — Customer's lifetime return count ≥ 5 + score ≥ 50 → tag customer with refundsentry:serial-returner.
  • High-value VIP protection — Customer lifetime spend ≥ $500 + score ≥ 40 → email the customer service lead instead of holding the fulfillment. (You don't want to friction your best customers.)
  • Bracketing detection — Order has 3+ size variants of the same product + return includes 2+ of those variants → tag the order bracketed.

Fault isolation

Workflow failures never block scoring. If a Slack webhook is dead or an email send fails, the return is still scored, the customer is still tagged, and the workflow failure is logged on the return's audit trail. Open Automation → Activity to see what fired, what succeeded, and what didn't.

Next steps