Sneaker Reseller
Scaling listing accuracy without burning out the team
Tens of thousands of listings, countless quality issues
Sneaker Reseller is a fast-moving business with tens of thousands of live listings monthly across platforms like Amazon. That kind of volume helps them stay competitive—but it also opened the door to small listing issues that created big headaches.
Because the company lists under brand-owned product pages (like Nike or Adidas), they don't have direct control over visuals or metadata. And sometimes… things don't match. A red sneaker shows up labeled "Midnight Navy." A size 9 listing accidentally includes a size 10 image.
"We started seeing more support tickets, more returns but the main problem was negative 1 star reviews because the product received did not match what they thought they were buying. The inconsistency wasn't coming from us, but it was impacting us. Which means we need to be proactive to work with Amazon to solve it."
But how do you manually review tens of thousands of listings every month, within budget?
Looking for more than just scraped data
The team reached out to Audio Bee with a simple ask: "Can we get all of the listing data through web scraping that we can run our analysis to review?"
Rather than treat it as a simple data scraping job, we asked them to help us understand what they were trying to achieve—to see if we could solve their problem end-to-end.
A lightweight, AI-driven system tailored to their catalog
Audio Bee worked closely with Sneaker Reseller to create a complete solution:
-
1
Web scraping to get list of all products because Amazon's UI has a 10k limitation of what you can see at any point
-
2
Web scraping of each listing page to get relevant data like size, color and product images
-
3
AI powered image and size mismatch review
-
4
Human review for items that AI is not able to say for sure is a mismatch
"The accuracy of this was amazing. We really loved how you solved the entire problem rather than just giving us scraped data which we would need to figure out how to process it."
Error rate dropped from 5% to less than 2%
After just two monthly reviews, the number of listings with errors dropped dramatically. Since new listings are constantly added, some natural error rate was expected—but the improvement was significant:
- 80% of the review process was automated
- Manual size audits went from 30 seconds to 15 seconds
- Manual image audits dropped from 1 minute to 10 seconds
- Return rates dropped, and so did complaints
"Audio Bee helped us protect the buyer experience—without burning out our team. It saved us time, but more than that, it saved our sanity."
Ready to scale your operations?
Let's discuss how Audio Bee can help your team work smarter, not harder.
Start the conversation