E-commerce & Dropshipping · n8n

Detect High-Refund Dropshipping Products Before They Drain Your Profit

An n8n workflow that every week calculates each product's refund, dispute and return rate, and flags the ones crossing your danger threshold with an AI-summarized reason from the refund notes — so you spot the quietly toxic products and pause or fix them before they eat the profit your winners make.

difficulty Intermediatesetup 55 minresult A weekly report of your worst products by refund rate, with AI-summarized reasons and a recommendation to pause or fix
  1. 1

    Run weekly

    A Schedule Trigger set to Weeks (e.g. Monday morning) pulls a trailing window of orders and refunds. Weekly cadence smooths out noise and gives enough volume for meaningful rates.

  2. 2

    Fetch orders and refunds

    HTTP Request nodes pull the period's orders and refunds/disputes from the Shopify API, including each refund's reason or customer note — the raw material for both the rate and the cause analysis.

  3. 3

    Compute per-product refund rate

    A Code node groups by product, dividing refunds by orders to get each product's refund rate over the window, and keeps only products above your threshold with enough orders to be significant.

  4. 4

    Summarize the cause with Claude

    For each flagged product, an HTTP Request sends its refund reasons to Claude, which returns the dominant cause and a suggested action — fix the listing, add a sizing note, change supplier, or discontinue.

  5. 5

    Send the danger report

    A Slack node posts the flagged products with their refund rate, AI-summarized cause and recommendation, ranked worst-first. You get a five-minute weekly review that protects the profit your winning products work so hard to make.

Frequently asked questions

Why track refund rate by product, not overall?

An overall refund rate hides the problem. Your store might sit at a healthy 4% while one product is refunding at 30% and silently erasing the margin from three good products. Because dropshipping margins are thin, a single high-refund item — wrong sizing, misleading photos, a defect-prone supplier — can turn your whole store unprofitable while the top-line looks fine. Per-product visibility is the only way to find and cut it.

How does it figure out WHY a product is refunding?

Refunds and disputes usually carry a note or reason (customer complaint, 'not as described', 'defective', 'never arrived'). The workflow feeds those reasons for each flagged product to Claude, which clusters them into the dominant cause. 'Runs two sizes small' is fixable with a sizing note; 'arrives broken' means a supplier problem; 'not as pictured' is a listing issue. Knowing the cause tells you whether to fix or drop.

What threshold should flag a product?

It's yours to set, but a common rule is flagging any product above roughly 2-3x your store's average refund rate, with a minimum order count so a single refund on a new product doesn't trigger a false alarm. The workflow computes each product's rate over a trailing window and compares against that threshold, which you keep in a config value.

Does it pause products automatically?

No — it recommends, a human decides. Some high-refund products are worth keeping with a fix (a sizing chart, better photos); others should be cut. The weekly report gives you the rate, the AI-summarized cause and a suggested action per product, so you make an informed call in minutes rather than never noticing the problem at all.

About this recipe. Recipes on FlowRecipesHub are written for business owners, not developers, and are tested before publishing — how recipes get made. Some ingredient links are affiliate links that cost you nothing — full disclosure.