Guruji launches…and local search engines in India are now REALLY HOT
The Indian internet landscape is getting more competitive by the day. Not only is it widening in breath, it is also getting deeper. Search engines have traditionally been at the heart of how the internet works, and we now have multiple local search engines to choose from. The latest (and most promising) entrant to this field is Guruji, which launched last week.
Guruji is a crawler based local search engine that focuses on Indian content. It has been co-founded by these two ex-IIT Delhi guys who have a strong professional background. They have the backing of Sequoia Capital for venture funding and Suvir Sujan (who co-founded Bazzee) as an angel investor. It is not entirely insignificant to note that Sequoia was one of the earliest investors in Google.
Off course if its about search, comparisons to Google are inevitable. Google’s search algorithm is almost like an industry benchmark and other search engines are likely to be assessed in terms of how closely they can replicate what Google does. But if my understanding of Guruji is correct, it is trying to do what Google does best, but in a very different way. To understand this better, you first need to understand how Google works for Indian content. Let me explain-
Google/Yahoo crawl the entire web and then categorize the content as Indian or not. It still retains the non-Indian content in its database and that content is included in Indian search results but gets lower priority. What Guruji does is in sharp contrast to Google’s approach. It discovers the content while crawling, stores only the India related content and throws out the irrelevant (non-Indian) data. This technique allows Guruji to crawl more, go deeper and hence index more India related data.
To understand Guruji’s second differentiator, take a look at this FAQ from its website - “Guruji currently has the data for Delhi, Mumbai, Kolkata……….free to submit any business listing you think is important and is missing“. When I first read this, it didn’t make any sense to me. How on earth is web indexing co-related to city specific data? I needed help from Guruji’s founder, Anurag Dod, who made me understand this better. Guruji calls this local (yellow pages) search. Using this section of their website (which seems to be an atomized listing of Indian websites), you can submit your establishment’s url for indexing. And when a query comes up, the user will get not only the web search results but also the business listing (almost like a yellow page listing). This feature is Guruji’s innovation and no other search engine does anything like this.
I think this local search is a very innovative feature. I also feel that this feature will almost work as a counter measure to Google’s very successful adwords program. Not convinced? Ok, let me explain how. I queried for “chinese food south delhi†on both Guruji and Google. The results are shown in the attached screenshots. On Google, you get a bunch of results and you also get a few featured adwords links on the right. On Guruji, the first difference is that the basic results seem more detailed and specific. Secondly, the local search (shown above the basic search results) lists two specific establishments, complete with their address/telephones. Now, isn’t that what the adwords program is supposed to do. This is my hypothesis. I tried to verify this, using other search queries as well but the results are not conclusive, probably because Guruji’s database is still very young and sparsely populated . Because of which, current comparisons of search queries on Google and Guruji would probably not be fair. We need to let Guruji gain traction before making judgments.
Anurag informed me that their future revenue model will be based on search engine ads (PPC) What I didn’t ask him was whether this is Google adwords or are they planning to have an advertising engine of their own as well?
I think Guruji is a very promising prospect. It has the right people backing it (i.e. founders, investors). And it is not trying to benchmark itself to what the Goliath in this field (Google) does. This is important, for one would need to think out of the box, in order to break Google’s stranglehold on Indian search.
While on local search, let me revisit two applications that I have reviewed in the past.
Onyomo- this is also a very promising local search engine, though its modus operandi is different from that of a classic search engine. For details, check out my earlier review. Its capabilities for local listing (or yellow pages search) overlaps with Guruji. It also has a very interesting mashup of city level maps with the search queries. Unfortunately it is available only in Delhi, Bangalore and Mumbai (future). In these cities, its search results are vastly superior to what Guruji gives at present (because of its modus operandi of adding content to its database).
MapmyIndia- this is the online mapping application that I reviewed a few weeks back and while it is not a search engine, it has a fairly potent local search section. You can search for services, facilities etc in the cities and they are marked out visually in the maps. It works fine, though the listings database needed to be better populated (when I reviewed it). And it has a pan-India coverage.
I am aware of another India focused search application that is currently under development (in the stealth mode). I have been shown a private demo but unfortunately, I can’t talk about it at this stage (I promised those guys). I think that’s very interesting as well.
Overall I feel that in about 6-12 months, these search applications are likely to become more mature. That’s when we shall really get to understand how capable they actually are. And its entirely possible that in the not too distant future, if you want to search local in India, you’d head to one of these, rather than Google. Webyantra says Amen with a capital A!
Shalin on Oct 20, 2006
I have been using it for last few days and I found it to be excellent. I like the city feature a lot, now numbers of my fav. restaurants are just search away. excellent job done. I would do anything to support a local search engine.
Madhur Khandelwal on Oct 21, 2006
I respectfully disagree with you on the note that throwing away non-Indian sites helps them in crawling more or in crawling deeper and getting more relevant information. It is just a matter of optimizing the resources on what to keep and what to throw away. Google has the money to buy tons of hard disk space and throw in more computers for crawling and so they could just easily crawl as many pages with the same speed. The real difference comes from how well the classication of Indian data is done and how the search results are ranked. For example, if a page is hosted on Wikipedia (which is not predominantly an Indian content site) and they have an article on “Taj Mahal” does the Indian search engine index it or not? If you don’t, you are missing out on a lot of good information that you can index. It will be interesting to see what Guruji shows as search result for the query “Taj Mahal”. Right now it doesnt show the wikipedia entry, but I for one, would like to see the Wikipedia page in the first page. Its too early to comment if they are doing the right thing or not, because I am not sure what is in the works and would wait for sometime before saying they are doing a great job or writing them off totally. Another differentiating factor would be if they plan to index other language sites or not. I have talked about this on my post at http://ileher.com/2006/10/12/gurujicom-baidu-of-india/
Also another minor point, you said mentioned that Google’s adwords program will work as a counter measure to listings. Not sure if I understand that correctly, but Google has local listings for search results in the US for quite some time and they serve up adwords as well and its quite clear what is what. for e.g. if you search for “pizza mountian view” you will see how it works. Even for Indian markets there are a lot of yellow page search sites where you can do such type of searches.
foobar on Oct 21, 2006
I dont necessarily agree with your post in entirety. To begin with, I think you are mixing up local search with web search. Also adwords is not same as local search.
Besides, I agree with the point made here http://www.startupdunia.com/2006/10/19/india-based-search-engine-guruji/
Why would I use Guruji for any regular search — for instance while searching for bollywood, why would I search on Guruji rather than Google. I think you’ve got it wrong in your review.
Ravi Venkatraman on Oct 21, 2006
I was doing some analysis of the search results from Guruji.com over past few days.
The search is very fast and relevant and it’s a great start.
One of the challenges they have is handling Parked Indian sites.
(Search for site:IndiaWorld.co.in in google.co.in and IndiaWorld.co.in in guruji.com engines)
At least 70% of the Indian sites are parked sites.
My definition of Indian sites.
- .in , .co.in and other 3rd level.in extensions (http://en.wikipedia.org/wiki/.in)
- .com with India suffix or Prefix
- Indian words written in English Shaadi, Naukri etc…
- .com names which used to have Indian content, but parked now.
- Genuine Indian Sites.
- In future, IDN Names..
One advantage Google has over others is:
Google is an accredited registrar and also provide feeds to most parking site and
can easily eliminate the listing of these sites by just looking into their DNS names or feed source.
Vaibhav Domkundwar - india 2.0 on Oct 21, 2006
Amit, I very much agree with Madhur here and posted on this topic on india 2.0 earlier: Search for the Baidu of India - http://blog.indiagoes.com/?p=36
It is really not so easy to identify Indian and non-Indian content - goes into the basic problem of context extraction which is not easy, not solved and very well documented. Google and Yahoo are pretty good in generic Indian content search already, and I would be surprised to see if Guruji or any other new search engine can match that breadth and accuracy.
Also even it Guruji is able to throw away non-Indian content, it is not trivial to crawl billions of pages - not something startups can attempt today, else we would have seen many more in the space.
I totally agree with the local search opportunity, but I disagree that the results you see on Guruji are impressive. It won’t be difficult to crawl some yellow page sites and accumulate such information.
Local search is definitely a big pain and a need but there is inherent lack of data to search across. Local search plays in the US are all built over robust yellow pages databases which are very accurate. There is barely anything of this sort in India right now - atleast not something with APIs and so on. In this situation, any local search play will be fairly shallow. I just searched for “pune coffee shops” on guruji and I got names of 2 coffee shops at the top - I grew up in Pune and have never heard of these.
That said, this is not a criticism of guruji at all, but a reality of the inherent problem/need/opportunity.
StartupDunia on Oct 22, 2006
I dont necessarily agree with your post in entirety. To begin with, I think you are mixing up local search with web search. Also adwords is not same as local search.
Besides, I agree with the point made here http://www.startupdunia.com/2006/10/19/india-based-search-engine-guruji/
Why would I use Guruji for any regular search — for instance while searching for bollywood, why would I search on Guruji rather than Google. I think you’ve got it wrong in your review.
Shalin Shekhar Mangar on Oct 22, 2006
Amit, thanks for revisiting Onyomo. You’d be happy to know that we’ve revamped our website and our core search engine as well. Moreover, that search is now available on mobile phones as well, through a SMS Search service. More information can found at http://www.onyomo.com/sms
As far as local city-specific search is concerned, I think that if you’re really serious about it, you’ll need to look beyond the net for your data. Otherwise you’ll miss a lot of things.
Madhur Khandelwal on Oct 22, 2006
I have posted some more thoughts on Guruji.com on iLeher clarifying the difference between web search and local search and how guruji could fill in the blank in the local search space.
http://ileher.com/2006/10/22/more-on-gurujicom-web-search-or-local-search/
Animesh Bansriyar on Oct 23, 2006
Whatever guruji.com is it must have some serious technology or a business model to be able to get Sequoia to invest in the company.
What I make out of guruji.com
1. It is trying to create a new space which might be “india specific content” search along with kind of classifieds and maybe a adsense type system soon enough.
2. guruji.com definitely has some kind of a local search model in place, but we need to wait and watch. “Local Search” hasn’t been all that successful with advertisers even in the US. It is really difficult to convince local advertisers to put in their ads on the Web.
3. Guruji is here to stay since I think the guruji getting acquired by anybody is a long shot right now. It will need time to get a foothold in the market.
4. I am not sure the branding has been perfect. I believe the branding of the entire search engine could have been better.
But, I am convinced that guruji.com does have something cooking.
-Animesh
amit on Oct 23, 2006
Madhur,
“I respectfully disagree with you on ………………………. deeper and getting more relevant information.†– You probably know more about search engines than I do. But instead of debating about search algorithms, we should judge the technology by what it actually delivers. Even at this stage, some searches on Guruji are far better than what Google gives. That’s not true for all searches though (and I mentioned that in my blog post). Which means that we should give this application some time to mature and then lets compare apples to apples. Right now, comparing Guriji and Google is like comparing apples to oranges. My overall sense is that the application looks very promising, even when you compare it to Google.
About the local search and adwords, I understand the difference between adwords and yellow pages. What I am trying to suggest is to look at it from a user’s perspective when he is trying to query a search. What he wants is good, relevant and reliable information along with his basic search result, which may be possible from local listings the way Guruji is doing it. Although the manner in which Guruji is asking websites to submit their urls for indexing may have scalablity problems.
Foobar,
Check my reply to Madhur to know what I meant on the adwords issue.
And, if it does not impose on you in any other way, can I request you to leave behind recognizable names when commenting. We are all grown ups here.
amit on Oct 23, 2006
Ravi,
Thanks for adding your perspective on the parked domain names issue. I have always found your knowledge about these issues very comprehensive and indepth.
Animesh,
Thanks for adding your views about Guruji. You probably know more about search engine algorithms than everyone else who is debating the point here.
StartupDunia on Oct 23, 2006
Amit,
Firefox had saved & prefilled my values. I realized that goof up and so posted again, with the correct values.
Probably should have included a note/update in there.
Anyways, just wanted to clarify so that I dont labelled as anonymous coward
thnx,
Pranav.
Madhur Khandelwal on Oct 23, 2006
Amit,
Onboard with you and others on the potential of Guruji.com and the fact that they are just taking off right now. May be I didnt make it clear in my earlier response, but I am NOT trying to compare with Google. In fact I am trying to advocated exactly the opposite thing on my blog that Guruji is different from Google at least as of now. Guruji is Local search, Google is Web search. There is no overlap. I have written about this elaborately at http://ileher.com/2006/10/22/more-on-gurujicom-web-search-or-local-search/
The above statement is also in line with what the technology delivers right now. Google does not do Local search in India right now, so obviously “pizza bangalore” does better in Guruji.com. So if I have to use Guruji, I would use it for “local” queries like that. For everything else (”Taj Mahal”, “Amitabh Bacchan”) I would use Google right now. If Guruji wants me to use them for these queries, they have work cut out to do better classification of Indian content regardless of what the site (which I am sure they will do) and going forward index local language content (which Anurag has already mentioned they will be doing soon)
Like I mentioned in my post, the real comparisons will only start when BigG or Y! start doing local search in India. For now I think the closest match is onyomo or other online yellow page listing sites.
For adwords with local listings, its a matter of UI design and I think both G’s have a smart team to make it clear for the users.
Deep on Oct 23, 2006
I really like the speed with which it comes up with the result, and I think it has capability to become Baidu of India.
Keep up the Good Work !!!
Dev on Oct 24, 2006
I am not too sure if Guruji.com will catch on in the indian space. Its very different world altogether. People search for specific stuffs around their city are miniscual, I think. And for a search engine to grow, it would need mass user base in the shortest period of time.
But never the less, the crawler algorithm is pretty efficient. I did find in few case, close to what ever I was looking for in guruji.
just a regular user on Oct 24, 2006
i use Just Dial services (find it faster and currently more relevant than a search) - although the data they have is not all that rich.
if i have to get onto the net for searching local listings, it has to offer much more meaning than just finding phone numbers/ addresses.
subsham on Oct 24, 2006
Search engines such as G, Y have a local play as well (e.g. local.google.com or local.yahoo.com). Basically, it is their listings business with a richer format (e.g reviews, maps, coupons, directions,).
A majority of local searches come from the generic search page (e..g google.com). So, these engines examine the query and then make decisions that this might be a local query and fetch local listings right @ the top (along with generic search results)
So, it is probably just a matter of time when these players build/acquire data in the Indian market. (yellow pages, maps etc.) Once, the data is present, these guys are up and running.
All the very best to Guruji to combat the competition in the coming months.
I might add that It might be a good acquisition target for some of the new Internet entrants like MIH or NDTV if Guruji gets some good consumer numbers.
Sunder Iyer on Oct 25, 2006
I think Guruji is a step in the right direction. And I do agree with DomKundwar comment that a lot of information in India is very unstructured. But this represents a major opportunity if someone can take all this unstructured information and piece it all together. Anyone who is able to tap the community and structure this information is in for a huge play. I think Indian information available on Wikipedia is some of the best on the net. If somebody can replicate this for local search etc , its going to take off really big.
Vaibhav Domkundwar - india 2.0 on Oct 25, 2006
Sunder - totally agree with your point.
The above JustDial comment is also really on the mark here. They just raised money as per contentSutra today and I wonder if their content is one of the most structured local content so far, for the metros they cover. There is no way to tell as their DB is closed.
Amit, I still somewhat disagree with your “local search is better on Guruji” comment, because it is fairly quick to build a crawled index of a small set of sites, identify meta data for the city and run a keyword search on it. The difficulty is when you crawl the entire web and try to extract data - and I do not believe Guruji is doing that. Its impossibly expensive to do that today.
On another note, isn’t it strange that none of the Guruji guys have commented on any of the posts on them so far?
amit on Oct 25, 2006
Vaibhav,
I am a big believer in the “proof of the pudding is in the eating” theory. Accordingly, my contention is based on actual results that I got on Guruji and not from opinions/judgements about the efficacy of search algorithms.
I ran about ten queries on Guruji and Google and if I remember correctly, four of those results were far better on Guruji. Some of the results on Guruji were pathetic and didn’t show up anything at all. But that is a sign of an unpopulated database rather than anything else.
Your suggestion that it is easy to do a quick local crawl but difficult to do a sweeping overall web crawl is true, but thats trying to benchmark search to how Google does it. Which is exactly the point Guruji seems to be making- if a local crawl can give good results, why do a global crawl? Even on Google search, nobody ever ventures beyond the first two pages of the searc results; so why crawl globally and throw up those hundreds of thousands of results that nobody anyway reads. Also the point about how Guruji will counter Google’s huge servers farms doesn’t appeal to me. I am sure when Google started, they started with meagre resources as well.
I read Madhur’s post on the distinction between a web and local search. I see his point of view and I understand that difference. But I feel that distinction is more for the tech/geeky kinds and doesn’t have much relevance from an end user’s perspective. The average Indian internet user, incidentally is not a geek, and in his/her mental model, the distinction between a web search engine and a local search engine is insignificant. And companies & products exist because of users and not the geeks who make those products.
Guruji or no Guruji, I have on many occassions been frustrated with the search results I get on Google about Indian queries. And I think there’s a window of opportunity for somebody smart enough to exploit. And given the background of its founders, investors etc, Guruji seems a very good candidate for that role.
amit
Vaibhav Domkundwar - india 2.0 on Oct 25, 2006
Amit, I agree with you on all the points above. A smaller index of focused content delivers better search. Some of this is also the foundation for iNods (www.inods.com) which is our reviews search engine as Lava.
Lets leave Guruji aside for a minute - because it is unfair to judge them without really knowing more about their methodology.
My main point was that as we discuss about vertical/local search engines there have to be some fundamental search technology related points to be considered when assessing a search play. A small index of 1000 sites with local Indian content, structured by some basic meta data like “city” can definitely deliver very good and relevant results for a keyword like “pune coffee shops” (assuming there are 1000 or more sites with rich content). And I agree that any user will love that - all good.
However, what is the barrier to entry for anyone else to whip up something like this? Some of my friends who have spent a few years in the GYM search groups can whip it up in a week and fine tune it as well.
So as we discuss search startups, I felt it is important to discuss the sustainability and IP aspects in all its geeky depth… hope I am getting my point across.
Cheers!
StartupDunia on Oct 25, 2006
Amit,
When you say:
Guruji or no Guruji, I have on many occassions been frustrated with the search results I get on Google about Indian queries.
What exactly do you mean by ‘Indian queries’ ? Do you mean searching for local stuff (coffee shops in pune) or do you mean searching for Indian stuff like ‘Bollywood’ ?
If you imply local search, then I agree with you on it, there is room to get a stronghold here. However, if you imply web search for ‘Indian queries’, then I fail to see how Guruji stands to outsmart Google.
Pranav
Animesh Bansriyar on Oct 26, 2006
StartupDunia,
My take on your view:
For example, I did a small test with both google and guruji for the search query “linux” on guruji and “linux” and “india+linux” as two separate queries on google.
Guruji gave me “linux-india.org” as the first link and “foss.in” as the 4th link and so on. Google gave me worldwide listings for the query “linux” and for the query “india+linux” it gave me “linux-india.org” as the first link, “indlinux.org” as the second link, “linux.org/groups/india” as the third and so on.
So, I feel this is not a pure local search play, but it tries to limit the pages which have content specific to India. It does fail at times but it limits the superset of indexed pages to “only pages which have india specific content”.
The rankings for “linux” on guriji and “india+linux” on google gave results which were completely different, this is because if
1. guruji follows a pagerank/HITS kind of algorithm then the links to and from are from this superset only, while in the case og google it is completely different.
2. guruji might have a different ranking system.
So, guruji has shrunk the web to only india specific pages when its searching for your query. And this is when Local Search comes into play - and hence tieups with classifieds, yellow pages. It is game on. A kind of adsense for advertisers targeting Indian Users follow. SMART ain’t it. And searches are not only for pizzerias or restaurants, it could encompass searching for Nokia N93 reviews, prices. Maybe searching for Hill stations in Himachal.
And for the geeks, if I were to build a similar search engine as a 8 weeks college project, I would think like the college student I was and
1. Lookup domain registrars and whois records to create a list of pages which are 99% Indian sites. This is trivial to do.
2. Lookup wikipedia, DMOZ, Yahoo/Google directories, del.icio.us, yahoo bookmarks to get as much “certain India specific structured data ” as possible. This takes up a lot of perl.
And gives me my starting point and a database of “a set of domains and urls”
Now I start crawling with a customized C++ crawler, and all my crawler does is starts from the domain/url db, dumps the pages and extracts all the links and puts the links into a url server.
Now I get smarter and vector index the dump of collected webpages. I now create a small subset of keywords which are definitely india specific and another subset which is definitely non-indian (manually or algorithmically derived). Then I go ahead and start crawling from my urls in the url server. Run a clustering or even a trivial naive bayes classifier to classify all incoming webpages as indian/non-indian. I add the indian pages to the index and all non-indian pages are hence discarded. This takes a few days to run, consumes all the bandwidth my college has and then I am all set.
I run a complex recursive algorithm (pagerank/HITS) and rank the pages. I throw up some HTML/AJAX and am all set for my india specific search engine.
I go ahead take some screenshots, the most dirty looking code, get some printouts and submit it to my project in-charge who keeps this as a project to show to the junior batch. And all I get is a hardtime with my regular papers.
And if I were to do it now, I will need to take about 3-4 months for the basic algo. And about 6 months to come to a public Beta, around 4-6 system programmers, 20-30 servers, a couple of angel investors, about $200,000 and lots of luck.
Apologies, if I have been off-topic and my attempt at humor sucks, I am really stressed out today and have been listening to some Pink Floyd/GnR.
-Animesh
amit on Nov 2, 2006
Animesh,
Thanks a zillion for taking out the time to explain this so well. My earlier statement about your knowing more about search algorithms, more than me or others in this discussions, surely was true.
Pranav,
By Indian queries, I meant searching for local stuff (i.e. coffee shops in pune). However I don’t see that distinction between local and web search as you or madhur are advocating. Its possible that guruji is better for local search right now but I don’t see why that logic cannot extend to web search as well, once their databse is populated well enough.
amit
Madhur Khandelwal on Nov 9, 2006
This has been quite a discussion! I had a chat with Anurag, their CEO where I have discussed all these points and few others. In simple terms they are trying solve the following two problems:
1. Fix the confusing user experience when searching for information related to India on google. Search on Google.co.in in “web results” or “pages from India” or search from Google.com with “india” keyword added?
2. Provide a value add by providing city specific local data.
Not completely new or ground breaking, but taking advantage of the nascent market in India by being comprehensive. Of course some clever marketing will also be in the plan to win over new Internet users coming in next few years in India. More at:
http://ileher.com/2006/11/09/exclusive-interview-with-ceo-of-gurujicom/
Rahul on Dec 17, 2006
Hey Guys,
I have seen guy in Bombay. He is champion.He is linux Guru.
He traced footprint of our hacked servers.
He does web site development,Web hosting,Netowork Setup(Routers,Switches)He is all rounder.
Anybody wanna contact him?
His name is Yogesh Saswadkar 9819340999
Ashish on Jan 10, 2007
Here is my review of existing search engine players in the Indian market: link
My Technopark on May 23, 2007
This is the forum operated by people in Trivandrum technopark. Please add this to guruji.com
hiutopor on Sep 18, 2007
Hello
Very interesting information! Thanks!
G’night
Mitali on Dec 4, 2008
Hi,
Guruji is based on the concept “Think Global,Act local”.Well this is a gud start for the search engine market in India.But for the timebeing they can’t compete head-to-head with Google.
Regards
http://www.chandnichowktochina.net