An e-commerce analytics company had a small internal team tagging product data by hand, and it needed something faster that didn't give up quality.
This company works with large sets of product data, and every item needed tags: what kind of product it was and which brand made it. A small internal team was doing that by hand.
They needed it done faster, but not at the cost of quality, because those tags fed their own analytics.
We started by categorizing products and picking out the brand from each item. That first batch of tagged data then became training data for the projects that followed.
The next one was product matching. For any two listings, was it the same product or not? We worked with them on step-by-step instructions to cut down on mistakes, and built training so our review team understood the task before they started.
After the proof of concept, they increased the amount of data they sent us.
This was one of our earlier projects, from when a lot of our work was building training data for machine learning teams.
If your team tags or matches product data by hand, that's the work we did here. Talk to us about what you're working with.
