Kosmix, the universal search engine that dynamically generates guides to search queries using dozens of different content sources, is quickly gaining momentum. According to today’s latest comScore numbers, the site has jumped up to 3.2 million monthly uniques in March – a 419% growth since February.
Kosmix is meant to serve as a search engine for when you want to get a quick overview of any topic (In some ways, its approach is similar to Mahalo’s). But unlike Mahalo, all ‘guides’ are built using search algorithms rather than human editors. Data is pulled from a variety of sources, including YouTube, Wikipedia, and news sites to generate each guide. This makes the engine very flexible, as it can attempt to build a page for any search query, but it has also can lead to some quirks – occasionally you’ll find items in a Kosmix page that seem out of place, which human editing would presumably avoid.
But for the most part it works well, and it’s growing quickly. The site began testing its universal search in a private alpha last summer but only began promoting it to the public in December, when it also announced it had closed a $20 million funding round led by Time Warner (the company has raised $55 million since 2005). And while its technology is primarily geared towards search, it is also flexible: last month it launched a personalized newspaper called Meehive, which uses the Kosmix engine.
Update: I’ve updated the image below to show Kosmix’s traffic over a longer period of time. Note the steep drop last year, when Kosmix switched from its old model as a vertical search engine for health, automobiles, and travel to the current site. It may be picking up quickly, but it still has a long ways to go before it reaches its past peak. Kosmix has clarified that the drop is due to a change in the way comScore counts its data. Before October, comScore treated RightHealth (which currently has 6.6M monthly uniques) as part of the Kosmix network, and now treats them as separate entities.










looks like a hockey stick graph, nice going!
This is definitely the wave of the future.
I actually prefer Yauba, http://www.yauba.com due to its privacy features.
Yauba http://www.yauba.com and Kosmix http://www.kosmix.com are the future of search.
The consolidated search is the new new thing.
These are all copying the Korean search engines which have been doing this for years and whipping Google’s ass.
Just goes to prove Korea is 5 years ahead of everyone else.
wats the URL to these korean search engines?
Daum and Naver are the two leading search engines in Korea.
It seems obvious to me that Kosmix and Yauba are looking at the Korean successes in beating Google and bringing this to US.
It is no surprise that both Kosmix and Yauba are India based companies, and that Korean products are very popular in India.
Kosmix seems to be more of a portal site, providing additional content on the homepage, while Yauba is strictly search, with a bare bones homepage like Google.
You are right. Kosmix is more like Yahoo.com, msn aol less like google. i like their home page better than the others.
I prefer Yauba http://www.yauba.com which just launched.
I second the recommendation on Yauba. Mainly because they also offer real-time search, like getting results from less than 1 second ago.
55 million dollars is a lot of money to aggregate search results.
Navgle.com also provides Universal Search using Google Search Services and Twitter search.
Check it out at http://www.Navgle.com
Oh oh. I smell a million splogs brewing on this thing…
It’s a solid concept and I like Meehive all the more. Well done team! Keep it up.
These guys are famous for buying keywords to artificially inflate traffic (see RightHealth). Doesn’t demonstrate any type of adoption for Kosmix. Just means that they can spend a lot of money.
See what happens to their traffic in a year when their search budget decreases.
yep. this is ALL about buying keywords and arbing traffic so that foolish blogs will right about them as a real company.
Sean, why isn’t your site included in this article?
Our traffic doesn’t even compare, but version 2.0 is almost ready and it should give us a similar effect.
The use of the term “universal search” does not make any kind of intuitive sense to me. Unless I am mistaken, it is not as though they are indexing feature transforms that would allow them to data mine non-textual sources.
Universal search means searching all different filetypes. For example try a query like this on a universal search engine:
http://www.yaub...amp;x=0&y=0
and compare it to what you would get on Google.
That is what universal search represents … basically “search once, find everything”
Yup, that’s part of what it takes to be a universal search engine. An additional requirement is being able to rank the different types of searches. E.g., on Kosmix, if you search for “Somali Pirates” today, you’d see news &blogs near the top of the page, but if you searched for Kauai, you see a widget from Tripadvisor, travel articles, images etc. near the top.
Okay, I can see a “universal search” referring to file types. For me the term intuitively maps to *data* types, which is a much stickier (and cooler) technical problem that no one has solved (yet apparently).
i love Kosmix. i have it bookmarked, check the site daily for updates http://www.kosmix.com
I think Meehive (from Kosmix) is equally cool, having a personalized newspaper created for me in real time, even including the latest tweets on my favorite topics.
As the technology matures this will become vital to certain professions (traders, defense, …)
This is the first search experience outside Google that I really like. Well done.
and, with mahalo there is no hockey stick, which might be the biggest difference of all. jason is going to have a wicked hard time trying to raise money again.
Kosmix is utterly cool. it’s the only place i can go to see “what’s happening on the web” Where else can you find twitter yahoo buzz, google trends, twitter and the onion all in ONE PLACE!!!!!!!!
Do anyone know about their founders? Good guys? Seems like Junglee was a decent hit.
well, i go way back with Venks, Anand, and much of the old Junglee and Amazon folks who work there. i know these folks for what they are — honest, brilliant, funny, and deserving.
total load of bullshit hockey stick
these guys are as sketchy as you can get online
total keywords buying exercise, have seen it before, why the big jump, the pages are fabricated crap
They’ve not even bothered to format wikipedia properly for one of the examples on their homepage. 50 million bucks for this? Not very impressed (:=
http://www.kosm...oline_Wozniacki
I second that, they are not even bothered about the security of their site. the search is vulnerable to XSS (just reported it to them. 50 million Bucks Hah!
Yauba is about a thousand times worse than kosmix!!!
Hey Yauba is kinda cool. Does anyone know how much funding they have? Where are they based? Who is backing them?
Do they also have $50 million in the bank like Kosmo?
I’ve read they have raised over $25 million from the son of an Indian billionaire.
the way you can move boxes around on the homepage is cool. wheres the social and personalization? you have to give users a reason to call it Causemix home. content mashups are everywhere. its gonna get very scary for algore based engine makers that dont have a social platform. where are there professional social channels that are consumer and business driven. how can they help small business?
best of skill to my fellow engine makers.
HomemadeLocator.com – start from scratch
Jason, it would be nice if you did a little bit more research into where this traffic is coming from. Maybe I’m wrong, but I sort of doubt that a bunch of people all decided in a one-month span that traditional search engines weren’t suiting their needs and then all flocked to Kosmix. Assuming that’s not the case, then we can probably conclude that Kosmix closed a large distribution deal somewhere, or is buying loads more traffic than usual. If it’s the latter, then they’re either doing it because it’s monetizing or because they’re going for some sort of funding event or exit. But this story as-is doesn’t even address these issues. Why not?
The quality of results from Kosmix is excellent. Take a look at the “Related in the Kosmos” results for any search and it should become obvious that there’s some truly semantic intelligence in the application for Knowledge Classification. I wouldn’t be surprised if some of the bigger search guys are taking serious notice.
If you would like to take the Kosmix experience to the rest of the web, give MashLogic a swing (see the link at the top of this page on TC) and be sure to activate Kosmix.
Kosmix indeed has a great categorization engine. “Related in the Kosmos”, topic to module relevance are just some of the applications for that engine.
I wish they could open the APIs like SemanticHacker, OpenCalais for developers to extract semantic info.
http://www.quan....com/kosmix.com
These are directly measured. Notice that the average pageview per visit is about 2… not very encouraging.
The question is how they’re generating the search traffic. Is it natural/viral or have they cranked up their SEM efforts to buy traffic from the general search engines? If the former, it’s very impressive. If they’re buying the traffic, then it’s less impressive but no doubt still economically viable if they’re effectively monetizing it on their site via advertising. This is basically the same arbitrage model that the comparison shopping search engines are built on.
Kosmix isn’t intended to be a replacement for Google. It is more of a discovery engine for broader topics, and for this purpose it is a great tool.
It seems that server is sometimes not responding.
You all need lives. What a pathetic existence.
best for you,
55 millions to create kosmix. That’s a lot of money. Try a new start up competitor:
Tagsup.com
It will catch up Kosmix soon, without 55 millions!