AI-powered inspection

Return Fraud Detection
for Shopify Brands

Return fraud is one of the fastest-growing threats to ecommerce profitability. In the UK alone, fraudulent returns cost retailers hundreds of millions of pounds each year. Most brands absorb it as a cost of doing business because they have no way to catch it. ReturnRev changes that.

📅 Last updated: May 2025 ⏱ 6 min read ✍ ReturnRev
🚩 An estimated 6 to 10% of all ecommerce returns in the UK involve some form of fraud or abuse.

For a brand doing 500 returns a month, that is up to 50 fraudulent returns slipping through unchecked. Without an inspection process, you are refunding those customers in full and restocking items you will never be able to resell.

What is return fraud?

Return fraud covers a range of behaviours where a customer exploits a returns policy to receive a refund or replacement they are not legitimately entitled to. It is distinct from genuine returns, where a customer changes their mind or receives a faulty product, and from friendly fraud, which is a chargeback filed on a legitimate transaction.

Return fraud exists on a spectrum. At one end is opportunistic abuse, such as wearing a dress once and returning it claiming it was never used. At the other end is deliberate criminal behaviour, such as returning a box filled with rocks or a product stolen from another retailer. Both cost brands real money. Both are preventable with the right inspection process in place.

The shift to online shopping has made return fraud significantly easier. Without a physical store interaction, brands cannot inspect items at the point of return. Returns are often processed centrally in a warehouse where speed is prioritised over scrutiny. And because UK consumer law gives customers a broad statutory right to return within 14 days, brands are reluctant to push back even when something seems wrong.

The most common types of return fraud

Understanding what you are dealing with is the first step to catching it. These are the fraud types most commonly seen in UK ecommerce returns:

👗

Wardrobing

The customer buys a product, uses it fully (wearing clothing to an event, using electronics for a project), then returns it claiming it is unused. The item is in a condition that cannot be resold at full price.

📦

Empty box fraud

A return is submitted and the package arrives containing packaging only, or weighted with unrelated items. The customer claims the item was inside. Without photographic evidence at receipt, brands have no defence.

🔀

Item switching

The customer returns an older, damaged, or lower-value version of the product instead of the one they purchased. A newer model is kept; the defective older one is returned for a full refund.

🏷

Tag and receipt fraud

Items are purchased from one retailer, tags are removed and replaced with tags from a different (usually higher-value) retailer, and the return is claimed against the wrong receipt or order.

🛒

Wrong item returned

A customer returns a completely different product, sometimes one purchased elsewhere or one they never bought from the brand. This is either opportunistic or an attempt to dispose of unwanted goods at someone else's expense.

💳

Friendly fraud (chargeback abuse)

The customer files a chargeback with their bank claiming the item never arrived or was not as described, while keeping the product. This bypasses the returns process entirely and can result in a double loss if you also process a return.

What return fraud actually costs your brand

The direct cost of a fraudulent return is the refund amount. But the real cost is considerably higher once you account for the full picture:

For fashion, electronics, and homeware brands in particular, where average order values are high and return rates can reach 20 to 40%, even a small percentage of fraudulent returns represents a meaningful hit to gross margin.

Fraud type Detection difficulty (manual) Detection with AI inspection
Wardrobing Hard: relies on staff spotting signs of wear AI compares photos against original product images, flags wear indicators
Empty box Caught at receipt, but no evidence trail without photos Photos taken on receipt create a timestamped evidence record
Item switching Very hard: requires staff to know every product variant AI matches returned item to Shopify product catalogue; flags mismatches automatically
Wrong item returned Moderate: obvious to staff but no proof without documentation Automatic Grade C hold; item cannot be matched to any inventory record
Condition misrepresentation Subjective: staff opinions vary, no consistent standard Consistent AI grading applied to every return using the same criteria

How ReturnRev's AI inspection detects fraud

ReturnRev's warehouse inspection process is built around the principle that every return should be documented before any refund decision is made. This alone closes off the majority of fraud vectors, because fraudulent returns rely on the brand having no evidence of what was actually inside the box.

Step 1: Photography on receipt

Every return that arrives at our UK warehouse is photographed at multiple angles before it is opened or processed. The outer packaging, any damage to the box, and then the contents are all captured. These photos are timestamped and stored against the original order record. If a customer later disputes the return decision, you have a complete photographic record of exactly what was received.

Step 2: AI product matching

Our AI vision model compares the returned item against your product catalogue, pulled directly from your Shopify store. It checks whether the item returned matches the product ordered, including variant-level details such as colour, size, and model. Mismatches are flagged immediately. If no match can be found in your inventory at all, the return is automatically placed on a fraud hold and you are notified.

Step 3: Condition grading

Every item receives a condition grade based on the AI's assessment of the photos:

A

Grade A

As new, sellable at full price. No signs of use or damage.

B

Grade B

Minor defects or signs of use. Sellable at a discount or as refurbished.

C

Grade C

Significantly damaged, wrong item, or fraud hold. Merchant review required before any refund.

Step 4: Fraud flags and merchant notification

When the AI detects specific fraud indicators, it raises a flag in your merchant portal with the supporting evidence. Fraud flags include: wrong item returned, signs of heavy use inconsistent with a change-of-mind return, item not matching any active product in your Shopify inventory, packaging inconsistencies, or missing components. You review the flagged return and decide whether to approve the refund or contest it. No refund is processed automatically on a flagged return.

Step 5: Evidence-backed dispute resolution

If a customer disputes your refund decision or files a chargeback, your portal provides a full export of the inspection record: timestamped photos, AI grade, fraud flags, and the original order details. This gives you a concrete evidence package to submit to your payment processor or to share with the customer directly.

What ReturnRev's system catches automatically

🔍

Wrong item detection

If the returned item cannot be matched to any product in your Shopify inventory, the return is automatically graded C and held. No refund without your approval.

📸

Condition evidence trail

Every return is photographed before processing. You have timestamped visual proof of exactly what arrived, which is the single most effective defence against fraudulent disputes.

⚖️

Consistent grading

AI applies the same grading criteria to every return. No variation between staff, no pressure to approve a refund quickly. Grade B items are not accidentally restocked as Grade A.

🚩

Fraud flag alerts

Specific fraud patterns trigger instant notifications in your portal. You can review flagged returns before any refund decision is made.

🔗

Shopify order matching

Every return is matched to the original Shopify order. Returns without a valid matching order are flagged immediately.

📋

Chargeback evidence package

Export a full inspection record with photos and AI assessment for any return. Structured to meet payment processor evidence requirements.

Why manual inspection is not enough

Most brands that process returns manually face the same structural problems:

Common questions

Can I legally refuse a refund if I suspect fraud?
Under the Consumer Contracts Regulations 2013, a customer's statutory right to cancel applies to a change-of-mind return, but it does not require you to refund an item that was not the item ordered or that has been substantially damaged beyond normal inspection handling. If the returned item is not what the customer purchased, or has been used in a way that significantly reduces its value, you are entitled to make a deduction from the refund or decline it. Having photographic evidence is essential to justify this position if the customer disputes it.
What happens when ReturnRev flags a return as fraudulent?
The return is placed on hold in your merchant portal and you receive a notification. The flag includes the AI's reasoning and the inspection photos. You review the return and decide whether to approve the refund in full, make a partial deduction, or decline the refund. No refund is processed automatically on a flagged return. The item is held in the warehouse pending your decision.
Can I override the AI grade?
Yes. You can override any grade from your merchant portal and leave a note explaining the reason. This is useful for edge cases, new product lines the AI has not seen before, or returns where you have additional context the AI does not. Override decisions are logged in the return record.
Does ReturnRev share inspection data with payment processors?
ReturnRev does not share data directly with payment processors. You export the inspection record from your portal and submit it as part of your chargeback dispute evidence package. The record includes timestamped photos, the AI condition grade, any fraud flags, and the matched Shopify order details.
How accurate is the AI grading?
The AI grading is calibrated against your specific product catalogue. Item-matching accuracy is high where your Shopify product images are clear and up to date. Condition grading is designed to be consistent rather than perfect: the goal is to apply the same standard to every return so that Grade A always means the same thing. The merchant portal shows the AI's reasoning for every grade, so you can assess confidence and override where needed.

Stop absorbing fraud as a cost of doing business

AI inspection on every return, fraud flags before refunds are issued, and a full evidence trail for disputes. Free to set up, no contract.

Get started, it's free to set up

From £2.50 per return · No monthly fees · Cancel anytime