Is this really a better way to Search?
IdeaLab’s Bill Gross, who started Overture and sold it to Yahoo for $1.63 billion, seems set to continue a long line of profitable start-ups with http://www.snap.com/, one of the top five alternative search engines. Though Google, Yahoo, and MSN combined hold over 90 percent of the search engine market, Snap has already been the subject of some rave reviews. Indeed, despite being just two years young, it was included in Time magazine’s August 14 edition as one of the fifty coolest web sites for 2006. Are the reviews justified or are they as a result of the founder’s pedigree and the big bucks that venture capitalists have poured into the company?
I decided to find out. I wrote down my list of search terms for the day, which included "mechatronics," "cytometry," "Daniel Craig" and "lyrics for buttons by PCD," opened up several search engines and compared their results to Snap. I also dragged along several willing accomplices (including one who only uses www.ask.com and knows nothing about other search engines apart from Google).
After two days and some perplexing results (where are the paid listings?) I started writing this review. I will look at how good Snap is at returning relevant search results for some terms, as well as its "cool" factor, where it stands in relation to the Big Three plus Ask, and whether it is a good place to advertise products. I will also provide some tips on submitting your site to Snap and getting listed in its SERPs.
When reading up on how Snap generated their SERPs (Search Engine Ranking Pages), I saw CEO and co-founder Tim McGovern wax eloquent on the importance of their licensed ISP data fields. These fields factor into users’ behavioral data; Snap uses them to sort results in the SERPs, so it actually lists the most "useful" pages first according to the user’s behavior. Thus you are more likely to see the information you are seeking in the first few results, and surfing the page presented will actually lead you to crawling deeper into a web site, instead of flipping between the search engine listings and the web pages.
When surfing on "offshore web design" I discovered that Snap returned more pages from the same website than most search engines did. While Google would probably give me two results from the same web site, Snap could in some cases return as many as five. The listings were not more links to the same pages, even if they were from the same web site (this is not guaranteed; the SERPs are different). It’s worth noting that Snap also returned less listings on most search queries, and for my "lyrics for buttons by PCD," Snap couldn’t find any listings, nor did it offer alternatives, as Google is wont to do if I mistype one of my arcane queries.
The algorithms Snap uses are designed by their team of developers to rank "most used" pages, actually adjusting the SERPs to reflect users’ behavior. This causes the algorithms to reflect some form of human intelligence (what the herd does) instead of just allowing the "snapbot" crawler to collate the data and rank pages based on key word counts or links. Indeed, on some of the top pages which were displayed, the keyword occurred only once, and the only links which linked to it were inbound links! Your most highly optimized page may not appear in Snap’s SERPs first, but what the users regard as your most useful page will.
In Snap’s listings we actually have unique results which will be more relevant to the searcher (and to advertisers). It turns all the talk about "behavioral marketing" into reality, but believe me, the view takes some getting used to.
For my search queries, Snap delivered on all terms, including the scientific terms "mechatronics" and "cytometry." It fortunately did not take me first to the ASME (American Society of Mechanical Engineers) pages which Google does consistently(after paid listings asking me to buy some book). Instead, it took me to "robot assembly tips." This shows me what I already know: most people searching on the term "mechatronics" (which means mechanical and electronic) used the free robot assembly tips pages more than the "members only" ASME pages.
For cytometry, my accomplice confirmed the trend: Snap skips the "buy me" sites and goes to abstracts and journals. This behavior is very similar to Ask’s. Another point worth noting is that, apart from some similarities in SERPs (some, not too much), Snap shares a couple of other "cool" features with http://www.ask.com/, who by the way are partners of Snap.
For Daniel Craig, Snap delivered interviews with Daniel Craig, and not a bunch of news for "Daniel Craig is the new 007!" Still, it seemingly couldn’t translate my "lyrics for buttons by PCD" into "lyrics for buttons by Pussy Cat Dolls," so I received no results. In Snap’s defense most of the other search engines floundered until I changed PCD to Pussy Cat Dolls (for which I found one listing with quotes, and seven without). I found no advanced search buttons (they either do not exist or I missed them), no image search, no blog search, and no features which would allow me to check for videos, music or other things. I guess we will have to wait for upgrades. [Snap appears to have added an image search very recently. --Ed.]
One of my accomplices complained about irrelevant listings when searching for "Purdue." She also complained that there were no maps or images of maps in the first few listings. She said that the text on the preview pane is too small for her to read, but she calmed down when I pointed out the "settings" button used to change size. Consider this a warning to Google lovers: Snap takes getting used to. It’s actually different.
The major cool factor is the preview feature. I discovered that it lets you view the pages and decide whether you want to go there or not. The left pane shows the page summary and the right pane shows a screen shot of the landing page. This is super cool; you read and view the landing page, decide whether it is relevant or not before clicking on it (which it often is; the users have spoken). I can almost guarantee you will deep crawl every page you land on (view at least three pages on the site including the landing page).
One really cool factor is that you can open the web site you want on the preview pane, instead of moving off of the search engine site and hitting the back button when you wan to check other listings. When it comes to "cool," Eric Schmidt and his Stanford grads may have to pass the title to these Cal Tech-powered geeks. Google, move over.
The Price of Cool
I started using Snap in the good days. According to some reviews, Snap was actually slower than it is today. This is amazing because the first time I used it on a 115 Kbps dial up modem, it took literally forever to load listings after I typed in my query. I am pretty sure it loaded under twenty seconds, but since I wasn’t waiting around for it to load, I had already put the same queries to Google, Ask, and Bloglines and gotten results before checking. Google and Ask returned results at warp speed.
I was warned by several sites that Snap is for "broadband users," but I ignored the warning as is my habit and plowed on undaunted. I finally laid my hands on a high speed connection the next day (100 mbps), and it was all right. Still, I wondered: is cutting off close to half the market really worth being "cool"?
I am predicting that by 2010 most users (70 percent) will be on some form of broadband connection, 3G CDMA EVDO for mobiles, Wi-fi and Wi-max or some other new technology. Until such a utopia exists, however, how long does Snap intend to hold out? We are talking about three years, an eternity in Internet time. Will Snap’s current customer base support it through the slug fest that the other major search engines put newbies through?
How Snap Will Compete
Snap is different from other search engines; it is actually "alternative." With its graphical-preview-based SERPs and its behavioral-laced algorithms, it has not one but two differentiating factors from the rest of the market. It is also certain to win its target audience, namely "visually oriented broadband users who want an alternative," no matter what the competition does.
In a sense, then, Snap does not really have any direct competitor on those two fronts. However it may soon see its behavioral algorithms being duplicated by Yahoo with "Panama." And it may start to lose advertisers as they jump ship and spend their hard earned advertising dollars where the eyeballs are.
Bill Gross has proved that he has incredible foresight before with Overture and the first pay per click model. Now with algorithms which also take user data into consideration, he may have started a bandwagon which the other search engines may seek to improve on. If they do not, it is probable they will just buy him out for another few billion dollars.
Advertisers have run the gamut of the click fraud-ridden Adsense to Yahoo’s exorbitant PFI/PPC program Sitematch. Snap has a confirmed one million users, and is designed around the advertisers’ listings. The cream for advertisers is that their listings are part of the SERPs. The Snap SERPs actually blend organic and paid listings seamlessly, while adjusting for user behavior to achieve continued relevance.
This answered my question of where the paid listings were! They were right there along with the organic listings. Not only does the advertiser get to have his pages previewed, but advertisers also have the chance to get the best and most honest deals.
Snap is promoting a cost per action (CPA) model rather than a pay per click (PPC) model. The cost of each advertiser-determined action is a minimum of $1 per fixed action performed, and then a minimum percentage (5%) of the sale for a variable action performed. For their PPC package, they have a minimum bid value of five cents per click, and there is no prepaid package; they charge you after a hundred dollars. In other words, there is no money down; the money is paid "after the fact."
Snap’s advertising program is called "Adsys." For now it does not offer broad matching of keywords, and only returns paid results for exact matches. You pay a non-refundable $50 fee up front, register, and "snap" you get reduced risks and a higher return on your investment. The more you are willing to bid, the higher you will rise on the search listings.
It is okay for the user also because sites with higher CTRs (click through rates) move up on the SERPs, so relevance is preserved. The advertiser only pays for performance, so an ad campaign can be run on Snap with a minimal budget. As the MySpace Internet Marketing forum says, "Go for it!" I say "at least give it a try, what’s the worst that could happen?"
Submitting and getting listed
You can go to Snap’s home page and submit your site, but unless I am wrong, they may take forever to review it and may even throw it out. I recommend submitting to one of their partners such as http://www.dmoz.org/. If you already get listed on Ask, you will definitely get listed on Snap. Putting the Snap browser "Snap Anywhere" on your website will trigger the "snapbot," but for high listings you may have to pay that $50 non-refundable fee. My research has not discovered how to differentiate paid listings from organic; check out the blog on this topic as we seek to answer "how to get high listings on Snap." In the meantime, even if you are not a "visually oriented broadband user" (sounds like a disorder) feel free to perform your next search on Snap.
Some Snap Mashups
Watch out for Snap job search in partnership with http://www.simplyhired.com/, as well as a beta shopping search package with http://www.smarter.com/. Other Snap partners include desktop search company x1 Technologies and Gigablast. These companies power portions of Snap’s SERPs, so getting in their databases should help you get listed on Snap itself.