This year we will spend $25 billion buying clothes and accessories online. Which is not to say we will enjoy doing it. Most big shopping sites are ugly, poorly designed and completely clueless when it comes to showing you clothes you might actually like.

Google, the bastion of search, is hoping to change all that. Boutiques.com is designed to not only make it easier to find clothes you like, but to actually predict what those clothes will be.

The site is organized into a series of "boutiques," which aggregate clothing from more than 250 designers and online stores. The online pages are curated by bloggers, fashion people, actresses (or their stylists), designers and, to a degree, you.

The visual-search technology behind the online fashion aggregators teach the computer how to "look" for clothes in your style, setting Boutiques.com apart from other online shopping sites like ShopStyle, and Net-A-Porter.

While the boutiques are interesting, it's the potential of the visual-search technology to make online shopping an intuitive experience that stands out. The idea is that the machine will know what you want before you do. This is particularly exciting to those of us masochistic enough to insist on having specific items in mind when we look for clothes.

Keyword searches work well when shopping for a new camera. But using them to look for a new pair of jeans will often yield results that are incomplete or just plain wrong. Ladies, how many times have you searched for kitten heels and gotten kittens instead?

Theoretically, with visual search, the computer can recognize not just that a dress is short and blue, but also that it has a sailor collar and cap sleeves.

The people who make these classifications at Google are a team of fashion bloggers, journalists, buyers and design-school graduates. These are the folks creating the site and fine-tuning the algorithm.

The people who curate the boutiques are celebrities, celebrity bloggers and designers. They pick the clothes and develop the style parameters for each of the six genres featured on the site: Romantic, Classic, Street, Edgy, Bohemian and Casual Chic.

To decide which style category you fall into, Google invites you to take a style quiz before shopping. In the first segment, you choose between two images, deciding which you like better. These are not only pictures of the latest runway looks, but also queries designed to dig out quirks in your personality.

Are you ironic T-shirt quirky or meat dress quirky? Do you prefer a Cosmopolitan or a tequila shot? The person who chooses tequila presumably falls into the Street category, with Romantic rising – fast.

The quiz can be long or short, depending on the consistency of your answers. Once it took me 35 clicks to get to Casual Chic, and another time only 25 to get to Romantic. I decided to go with Romantic, because, well, I liked the sound of it. One of the best side effects of a trip to Boutiques.com is that everyone leaves having been bestowed with a style. And thankfully it's never "whatever's clean."

Next, I indicated my preferences in specific clothing categories: I like dresses that are strapless. I like black and red. I don't like V-necks or reptile prints. Which brings us to the main problem with the quiz: As with Pandora and other recommendation engines, it's better not to get too specific. You invariably end up weeding out things that you might like. More often than not, we don't like or dislike something because of one detail like a V-neck. Except for reptile print. That’s never okay.

The quiz complete, I was taken to My Boutique, a page populated by a collection of items inspired by my answers. The filters seemed to be working pretty well, as I specifically indicated I liked round-toed red shoes, not pointy-toed red shoes, and I was shown lots of round-toed red shoes and no pointy-toed shoes. That didn't mean I liked them all, or even most of them. The filter is like going into your favorite store. It makes liking more likely, not a sure thing.

My goal was to find a dress for New Year's Eve. The dress I saw myself wearing was something reminiscent of the shimmery sleeveless number The Little Mermaid wears when she finally gets her legs. Not something I would go into a mall looking for, but this was a mall powered by Google, the bastion of search. I was optimistic.

I typed "glittery dress" into the keyword-search bar, and I was surprised by how many contenders popped up. None were exactly right, but several were close enough to warrant a click on the Visually Similar Items button, an inspired feature of the site. Of course, you hope they will be similar in the ways you want and not in the ways you don't, but there's no way to tell it that. In this case, a lot of sparkly dresses showed up, but they were all short, something I had specifically indicated in my style preferences that I hated (their term, not mine). Apparently, the filters I set up didn't work with the Visually Similar Items button.

The main problem of trying to do this kind of very focused shopping on Boutiques.com: It's hard. In offline shopping, you manage your expectations. I would never go to the mall with the goal of buying a dress even remotely similar to the one worn by a cartoon character in a 20-year-old movie. Online shopping is more focused, which is fine if you need a TV or black socks.

When you're looking for clothes, it's much more difficult to find something you like even if you have every dress ever made to choose from. Sometimes, one dress in the flesh can be worth 10,000 images. And sometimes even 10,000 dresses isn't enough.

I may not have found my Little Mermaid dress, but after some refining, Boutiques.com has the potential to combine the fun of offline shopping with the efficiency of not having to put on pants to do it.

With its large images and simple, logical design, Boutiques is great for browsing and stumbling into unexplored lands like Bohemian or Carey Mulligan. From cheap Forever 21 dresses to $3,000 McQueen gowns, all clothes get the same visual treatment. And in this context it's fun to see how arbitrary pricing can seem.

However, if they want people to actually buy clothes instead of just looking at them, they need to make some improvements. It seems to be curated for the young, thin and trendy. The fashiony boutiques are great (who doesn't need an easy place to find Harem Inspired clothing?), but the people at Google would do well to include more practical and accessible collections for specific body types, ages or professions.

The filters themselves work pretty well. If you tell it you hate latex stilettos, you won't end up with a ton of stripper heels in your boutique. However, you still get a lot of shoes you don't like. Hopefully with time and more interaction with the site, it will get better at recommending flats you like and not just any old ugly flats. Whether people have the patience to teach a program a thing they already know remains to be seen. At this point, Boutiques.com is more entertaining than practical, but it's a good start.

Someday, maybe we'll have online shopping technology that makes it possible to buy a pair of pants without trying them on first. Now, that would be revolutionary.