Pixsta's ad network for fashion is all about image

Pixsta is a London-based startup with an image search technology which allows you to browse and search visually. Since launching trials with a handful of partners including Elle’s browse&buy as well as others like the News of The World, it’s now going on the warpath with an image-based advertising network for the online fashion retail market, similar to Google’s adwords model, but based on images not text.

This is how it works: You click on an image, say a dress, and new associated images (matching handbags, shoes etc) appeal in a cloud around it. Clicking on any of the displayed items re-submits that item as the search term. Pixsta argues that a click on its AdImage network is worth more than an AdWord click because for the advertiser, customer acquisition costs are lowered, with lead-to-customer success rates higher. Media brand owners get a new, visual way of monetising their site which is perceived as content rather than advertising. Consumers get to see all the products from various retailers presented in a homogeneous way, like in a shopping mall, with no need to visit numerous individual sites. Clicks go direct to the product landing page.

Pixsta’s patent-pending image search technology is based on research by Pixsta co-founder Dr Daniel Heesch, who came up with algorithms that enable any image to be compared with any other, to determine their similarity in terms of shape, colour, texture and more. Pixsta CEO is Alexander Straub, CEO and co-founder of Pixsta and of early dot-com Mondus back in 1999.

I don’t know where they plan to head with this technology but it occurs to me that if they allowed users to improve the image search by adding their own tags, the search may be improved. However, I dare say they thought of that already and are confident their algorithms can do the heavy lifting.