Some friends and I have been working on and off on a service called GetSize (https://www.getsize.shoes) since 2019 and feel we finally have enough pieces in place to take it somewhere. We’re however uncertain about quality and would love your feedback on any and all aspects of it.
The core insight is that pretty much everyone has returned shoes they bought online that didn’t fit well. There are 3d scanners in stores which are cool but that won’t help you at home, and AI photo apps are generally in a pretty sad state still. Consumers do however have multiple shoes with a preferred fit at home, and when applied at scale it’s a simple data problem:
If I have 5 shoes that fit me well, and you have 5 shoes that fit you well, and we have a shoe in common; then I can buy all the same models and sizes that you have.
We are now building a dataset for this purpose, and to do it we’ve simplified the size chart tango that all retailer sites offer today. We’ve ingested a bunch of brand specific size charts so that you no longer need to look at excel sheets essentially. Just a simple recommendation based on what you already own. And whatever shoes you add will contribute to creating a dataset that makes size charts obsolete.
So why does this matter? - Around 20-30% of shoes bought online are returned, half of that is due to poor sizing. - Retailers usually solve this in isolation or not at all. There are some great solutions, but isolated to specific sites because of their business model. - Size charts alone aren’t enough. They differ across brands and in many cases across models. And due to this, larger retailers can’t even map the correct chart to all of their models (See Europe’s largest online retailer Zalando as an example). They often use generic size charts.
The website design is a bit… developer-flavored. Our priority so far has been getting the backend, data collection, and recommendation engine working, and the texts are tailored for SEO to build organic traffic. Kind of hate that part personally but I guess it’s a necessity. Or is it?
I have written a first draft of our remote MCP server where the true value lies, but it’s not yet live. I hope it will put less emphasis on the need for a web frontend.
Regardless, would love to hear your thoughts and appreciate any feedback I can get.
Best regards, Thomas