Pixazza "Product in the Picture" Service: AdSense for Images

Pixazza is looking to turn Web photos into money. It might just work.

The startup, staffed by Netscape veterans with a $5.75 million Series A venture capital round, on Wednesday is unveiling a service that lets Web publishers apply tags to photos on their sites—they literally look like tiny yellow-and-blue price tags. Hover your cursor over one of them, and a box appears that lists the product, such as a coat or a pair of shoes, that looks like (or actually is the same) coat or shoes in the photo. You can see the price and the retailers offering the products and click over to a retailer’s product listing. You can check it out at one of the few Web sites such as imnotobsessed.com that have testing it since last fall.

Essentially, Pixazza is an ad network much like Google’s AdSense, which runs text and display ads on other Web sites that are related to the content of those web pages. Pixazza CEO Bob Lisbonne says he hopes the platform, as he prefers to call it, will help Web sites make some money from a commission on product sales. It’s an e-commerce version of the way thousands of sites survive by running AdSense ads. No surprise, then, that Google is one of the investors, along with venture firms August Capital and CMEA Capital and individual investor such as uber-angel Ron Conway, LiveOps CEO Maynard Webb, and Facebook Chief Financial Officer Gideon Yu.

But Google may be as interested in the system behind Pixazza as much as in the similarity to AdSense. Although Like.com, for instance, uses image recognition technology to help people find products similar to those it pictures on its site, Pixazza figures no algorithm can identify products in photos nearly as well as humans.

So the company has enlisted people--initially friends of company employees and their friends but ultimately anyone with some time and desire to share a bit of the product sale commission--to sign up to identify products in photos on Web sites and match them with some 2 million products in the databases of Pixazza's retail partners. Those include Amazon.com, Zappos, Bluefly, Macys, and many others. "Google recognizes that we have a lot of expertise in crowdsourcing," says Pixazza Chief Technology Officer James Everingham, former CTO of the crowdsourced call center service LiveOps, some of whose technology team is now at Pixazza as well. "It's incredibly difficult to design the community and tools to make this work."

For now, the only product category is apparel, though the company plans to branch out later into other products and even beyond images. It also plans to add by-the-click and by-the-impression pricing down the road, in addition to the current by-the-sale commission plan.

Pixazza will have to overcome some natural obstacles. For one, it better make sure these tags don't delay the loading of sites, like all too many apps and ad networks. Everingham says it only takes 200 milliseconds (in other words, not really noticeable), but it will have to prove that.

For another, some people may see tags on photos as intrusive or annoying. Lisbonne makes a good point that they're far less so than those highlighted words in text--often words that have little connection to anything someone would want to buy--that pop up ads unbidden. But they do mix content and ads closely enough that they may annoy purists.

Finally, I'm not entirely sure how many people who stumble across photos on random Web sites immediately wonder where they can guy the coat or blouse they see in the picture. Of course, I'm a guy, and one for whom shopping is a chore, so take that with a grain of salt. And maybe you only need a small percentage of people to click on the tag (and then fewer still to buy something) to make it worthwhile to merchants and Web sites.

    Before it's here, it's on the Bloomberg Terminal.