What I learned from yet another failed experiment: The Gift Feed app

Yet another idea that seemed to have potential, but it was not: Gift Feed: a mobile reel-like feed of gift ideas, with the idea to make money via affiliate links to those products.

Seems to have potential at first glance, but here' what I figured out as I poured hours into it

1. Traffic isn't free and isn't cheap. The more qualified, the more expensive it is. And paid traffic need to convert enough to make it a sustainable business model (see next item)

2. In this number's game, you need to draw traffic in so that a percent of the visitors clicks on the affiliate link and then a smaller percent buys from that website. With ~0.25€ per click, out of 100 visitors, I didn't manage to get any affiliate links... which brings me to the next item

3. The algorithm counts. Product recommendations need a lot of specialized data and are hard to pull off. I tried using some AI to cluster the products a little bit to offer better results but with my 35k cralwed products I wasn't able to build a diverse enough dataset of products (see the enxt one)

4. The data counts. I crawled some products which made the data very messy. The images more so. At Vuuh I work with product and image data everyday and I expected it. But lacking the resources to clean everything up, the data quality took a serious hit.

5. So without investment in data, algorithm and traffic sources, this idea was dead in the water. It sounded interesting enough at the first glance, but as soon as I started to dig into it, these issues became more and more important and I lacked the resources and energy to handle them.