The problem with e-tailing

NEW YORK (Fortune) -- Recession notwithstanding, U.S. consumers seem pretty addicted to online commerce: The U.S. Census and Forrester Research both expect total sales of goods and services procured online will reach $235 billion this year, up about 15% from 2008 spending.

But that doesn't mean the online shopping experience couldn't stand a lot of improving. In the terrestrial world, browsing is an important part of how we shop. We scan hundreds of items at once as we walk into Macy's, say, and pick up whatever object happens to catch our eye.

Online, by contrast, we arduously click through product after product or maybe compare a few at a time. Who knows how many spontaneous sales are lost by our inability to experience the pull of something beautiful that we didn't know existed?

One new startup founded by a couple of computer science PhD candidates is hoping to change this dynamic. Modista scrapes the inventory of several shoe, watch, handbag and sunglass retailers and serves them up at a blisteringly fast pace. The effect is a browsing experience that rivals, if not surpasses, anything offline -- and certainly anything else on the Web.

The best way to understand it is to check out http://www.modista.com, but here's quick explanation of how it works. Let's say you're looking for eyewear. The site presents a 9x9 grid full of products as a starting point (the size of the grid depends on the width of your browser window), everything from Wayfarers to ski goggles. You click on an appealing pair, and the site loads 80 new products, almost instantly, that are closely related to your choice. In the new grid, you may find a more appealing option still. You click on it, and the page is again redrawn relative to your new choice, and so on, and so on. There's very little text on the page -- just an optional price tag. That's by design. The founders built the site to work as a visual browsing engine, to expedite purchases that are based largely on aesthetic appeal.

"The idea started when I was watching my girlfriend shop online for shoes," says co-founder Arlo Faria. "She found a pair of shoes she really liked, but she wanted another pair just like it, but cheaper. So she was clicking the NEXT button over and over -- for an hour."

Both students at UC Berkeley, Faria and co-founder Ajeet Shankar figured they could automate that process. They began reading up on computer vision and signal processing (in a nutshell, image retrieval and interpretation software that allows computers to see), and developed an algorithm for comparing products to each other based on color and shape, while also including attributes like brand, material, et cetera. Then they developed a way to serve up the images at a blistering pace. "When you make a query to the server, in 5 milliseconds or less, it'll give you a response," says Shankar. Adds Faria, "Most of the time you're waiting, it's just for the images to load."

Shortly after the duo had developed the technology, they won a business plan competition at Berkeley, which scored them an office, some startup funding and a slew of contacts. In December, they unveiled a proof-of-concept portal and began attracting interest from big online retailers like Zappos.com. In addition to feeding visitors to retailers like Zappos, 6pm.com and eBags, in exchange for a referral fee, Modista also began powering the explore.zappos.com portion of Zappos last month.

Now, with a couple of patents in hand, Faria and Shankar are reaching out to every big retailer you can imagine about doing licensing deals -- everything from apparel to jewelry, furniture and faucets. It's easy to imagine a Modista engine (or a Modista-like engine) enabling all manners of online shopping in the future. "Anything with a visual component," says Shankar.

It's far too early to call their project a success, of course. Both co-founders have been overwhelmed at just how rapidly inventory turns online. And they're not getting much help from the search engines, which, they say, don't crawl Modista very well. But the company has already succeeded in shining a light on how ecommerce can be made into a more user-friendly experience. The good news is that the company has almost no burn rate. It runs on Amazon's Web services platform and makes enough to survive from commissions generated at Modista.com.

If nothing else, their foray has opened their eyes to a world that was completely foreign to them -- and isn't that what college is supposed to be about? "We're just two guys," says Faria, "with 250,000 kinds of shoes."