Josh's Thoughts

So I just took a couple minutes to check out Cuil, because Techcrunch can’t stop posting about it. Not too shabby for an early version. They need to work on their speed. I searched for “shoes” and it took quite a while for the site to render. The 3 column layout makes my head hurt. It’s tough to scan quickly. A friend once told me that there’s a lot of evidence to support single column vertical list layouts for results because they’re much easier to scan. I believe her because she worked at Y! and because scanning staggered blocks of text w/ images in multi-column layouts makes my head hurt. One last thing on Cuil – their categorization system is pretty cool, but I wonder if clicking and discovery what’s in those categories might be too much work for some folks. See: “Satisficing”

I’m a search user, and in this case, I’m probably closer to being a typical user than a power user. When I use Google, I never use Advanced Search or fancy-schmancy queries. I just type in what I think I want, hit the button, and see what I get. If I don’t get what I think I want, I try again. It usually works. The main limitations being:

1. I need to be looking for something, and I need to know something about what I’m looking for.
2. If the perfect search result isn’t indexed and I’ve never seen it before, it functionally does not exist for me. For people who use search engines as their primary gateway into the Web, this ends up making their engine of choice the canonical source for all things Web related. Something anyone working on the next big search innovation should remember.

Now on to the not-so-secret search secrets that mention in the title of this post.

I guess there are two ways to approach search. One way is to position your search as a service. In real world terms this would be similar to seeking advice from an expert like a travel agent, interior designer, matchmaker, personal shopper, etc. Some cool things about search as service are:

1. People usually don’t mind waiting for it or paying for it, because they wait and pay for it in the real world.
2. Search as service is a filter not a sort. Presumably you wouldn’t need to even attempt to index the entire Web, you just need to know the best results for the given search, and it’s likely that only a handful are necessary. The best search as service would likely return 1 perfect result.
3. You get to have an opinion.

Some uncool things about search as service are:

1. The amount of content and rate of change on the Web makes it extremely difficult to become an expert. Of course the work it takes to become and maintain expert status is what people pay for. This is probably why you see the search as service model used primarily for limited scope search. Makes me think of sites like Expedia, Kayak, eHarmony, Match, Hotels.com, etc (wow a lot of IAC companies in that list, huh?).
2. You have to convince people that you are an expert. Jeeves tried it but just couldn’t pull it off. He has since been fired.
3. Service industry folks know that you can’t make everyone happy all the time. For search as service this means that you’re likely to find people who just don’t like your expert opinion. This will reduce your user base.

The other would be position yourself as a utility. By utility I mean a utility like the phone company, power company, etc. This is the direction most traditional search engines go in. The cool things about search as utility are:

1. The focus is on being useful, not “right”. As long as you can return useful results you’re in good shape, and of course, the more useful the results the better you’ll do. It is different than being an expert because you don’t need to have your own opinion.
2. If people find you useful, they’ll come back often. You can monetize people who come back often.
3. If done right, you can become like Google, MSN or Y! and become your user’s canonical source for all things Web.

Hard parts:

1. You have to know one hell of a lot about the Web. You’ve got to do everything in your power to know about as much Web content as you possibly can. You never know what people will be looking for, and to be useful you’ve got to do your best to give them something.
2. As a utility, people take you for granted. You exist to be used. People want to use you to get to somebody else. That sucks, and there’s ultimately an ego issue some companies can’t get over.
3. You have a responsibility to be fair. People trust you like they trust the water company. You can’t poison them with ads or artificially manipulate results to serve your financial interest. As soon as you do you risk losing the trust of your users.

Google just happens to be the best at it for a few reasons:

1. Google is fast. It’s damn fast. It’s so fast that many people use it multiple times a day and don’t even realize it.
2. Google has indexed a lot of pages, to the point where if it exists on the Web and Google doesn’t know about it, it’s probably not worth looking at. – even my blog is indexed. Google has become the canonical source for many folks the way AOL was way back in the day.
3. Google’s business model and their features mesh better than any other Web business. Google does everything in it’s power to send you away from Google as fast as they can with the hope that you will come back and do it again. Their ads are performance based, so the combination of people leaving quickly and coming back often makes for more clicks on ads and more revenue for them and their advertisers. This is different than MS or Y! who both, I believe, sell impression-based ads. This means they make more money by keeping you around. Can you see the problem in that if you’re supposed to be ego-less and unbiased? Note: There is some debate about potential Google bias in results since the launch of Knol last week. Use your search engine of choice and search for “is Google a content company Knol” to see what I mean.

OK. So that’s my big long post of the month. I’m not an expert at search. I just use it a lot. So take it as you like it.