BranchVoice
Demo
New

Return vs Listing Match: Catch Return Fraud with AI

Upload the listing image and a photo of the item you received: the AI says whether they are the same product, how closely they match and where the difference is. It ignores angle, lighting and background, and looks for another model, colour, wear and signs of a fake.

Listing image

Photo of the item received

Example photos

Example photos: Museum Rotterdam, CC BY-SA 3.0 (Wikimedia Commons). Listing on the left, item received on the right.

Result

The result will appear here.

Results are AI estimates and may contain errors. Your text is not stored; the result for the same text is cached briefly.

Labels

The model scores the probability of each damage type below for every photo. You can extend the damage types for your processes or use it with labels made for you.

Different brand
The item is another brand from the listing.
Different model
Another model or design, even if the brand is the same.
Different colour
The colour or pattern does not match the listing.
Used or worn
The item looks used, worn, washed or dirty.
Missing parts
Parts, accessories, tags or the box shown in the listing are missing.
Signs of a fake
A wrong logo, poor stitching or cheap materials.

Where is it used?

Examples of how teams use this model in their daily work.

  • 01

    Stopping return fraud

    Photograph returned items at the warehouse and catch a different or fake product before you approve the refund.

  • 02

    Wrong item complaints

    Compare the customer's “I got the wrong item” photo with the listing and verify the complaint in minutes.

  • 03

    Marketplace seller audits

    Check samples of what sellers ship against their listing images.

  • 04

    Spotting used items

    Separate items returned as new that are actually worn, washed or missing their tags.

Use this model in your own system

Change the labels, add new ones for your industry, or connect the model to your CRM, contact centre or e-commerce stack through our API. For API access and custom models, write to us at [email protected].

Write to us

Frequently asked questions

Is this tool free?
Yes. You can analyse a number of texts a day without signing in; heavy use gets a short cool-down.
Which languages does it support?
It understands multilingual text, Turkish and English first; results are shown in your interface language.
Is my text stored?
No. The text is only processed for the analysis; the result is cached briefly so repeated requests return quickly.
Can I analyse in bulk?
The BranchVoice platform collects your reviews from their sources automatically and classifies them with these models in bulk; request a demo to try it.

Other tools

All tools

Want to analyse thousands of reviews automatically?

BranchVoice collects reviews from Google Maps, complaint sites, social media, app stores and marketplaces every day, classifies them with these models and turns them into alerts, reports and dashboards.

Request a demo