Kaushal Kurapati\’s blog

Thoughts on Search, Technology and Management

Mobile Web & Search Usage in US

Posted by kaushalkurapati on May 25, 2007

30% of mobile users in the US access the web from their mobile phones, according to a study by iCrossing. US cell phone subscriber base is about ~230 million (www.ctia.org). So folks accessing the web from the mobile are ~69Million, which is significant. According to the study, half of these access the web several times a week. A majority of those who access the web from theri mobile devices, 75%, conduct searches. They mainly use GYMA search engines, reflecting their PC-search choices.

Local info, weather, maps/directions form majority of the queries. Interestingly, about 85% of these mobile-web users expect mobile versions of the web sites they visit. This implies that there is a high incentive to create mobile versions of web sites.

Posted in Mobile, Search, Stats | Leave a Comment »

Search Engine Market Share: April 2007

Posted by kaushalkurapati on May 25, 2007

comScore just released Search Engine market share data for April 2007. Google inches to ~50% of the market. Yahoo, MSFT, Ask, and AOL together account for 47%. Total searches conducted in April 2007 were 7.3B, unchanged from March, but 11% up year/year.

  • Google – 49.7% share (3.6B searches)
  • Yahoo – 26.8% share (2B searches)
  • MSFT – 10.3% share (757 million searches)
  • Ask – 5.1% share (376 million searches)
  • AOL – 5% share (364 million searches)

Posted in Search, Stats | Leave a Comment »

Internet Advertising growing at 35% y/y

Posted by kaushalkurapati on May 23, 2007

IAB reports today that Internet Advertising in 2006 grew by 35% over 2005 to $17B. Display advertising grew fastest y/y at 47%; classifieds grew 44%; keyword search grew 32%. Market share breakdown:

  • Display Advertising 22% – $3.75B (y/y growth: 47%)
  • Rich media display advertising 7% – $1.19B
  • Keyword search 40% – $6.8B (y/y growth: 32%)
  • Classifieds 18% – $3.06B (y/y growth: 44%)
  • Lead generation 8%

Posted in Advertising, Internet, Stats | 1 Comment »

Mobile Search User Behavior

Posted by kaushalkurapati on May 7, 2007

Google mobile search user behavior analysis was published in this interesting paper. Some key behavioral findings were as follows:

  • Average number of words per query were roughly same b/w mobile & regular searches: roughly 2.3 words/query.
  • Surprisingly 17% of queries the authors looked at were URLs. The number is much much lower (1-2%) for regular search logs.
  • Top categories: in cellphone based searches, “adult” was the top category (>20% of searches); Entertainment (>10%), Internet & Telecom (>5%), Local Services (5%) and Games (>2%) round out the top-5. The interpretation was that since a cellphone is considered a very private device, people look for more adult content than a computer, which may be used by multiple people.
  • Top categories in PDA-based searches differed: local services (>15%) was the most preferred category due to probably the profile of the user–business users. Entertainment, Computers & Technology, Internet& Telecom, Travel, Adult, and Sports each account for ~5% of queries.
  • Query distribution: the top-1000 queries on cell phone based search accounted for 22% of all queries, whereas it was 6% for regular search. This shows that cell based search has less variety in queries currently.
  • Number of queries/session in mobile was 1.6 whereas it was b/w 2 and 3 in regular search. It takes about a minute for a user to enter a query in mobile search! This fatigue could be the reason for searching less, but being specific (same # words/query in mobile as regular search) at the same time.
  • About 32% of consecutive searches are the same; 29% of consecutive searches are refinements of original query; 14% are spellcheck triggers. So the remaining 25% are not on the same topic. This suggests that mobile search is very focussed and is not exploratory in nature.

Summary: mobile searches roughly have same # of words/query as regular search; number of queries per session is much less than regular search; people often focus their searches and explore less while mobile. It currently takes too long to input search queries, which may be limiting the # of queries/session.

Posted in Internet, Mobile, Search | Leave a Comment »

Speed of data transfer over Internet

Posted by kaushalkurapati on March 20, 2007

Jonathan Schwartz of SUN has an interesting back-of-the-envelope calculation for how long it will take to move a petabyte of data from Sanfrancisco to Hong Kong over the Internet. He estimates 507 years at 1/2 MB/sec connection speeds. So it is easier to send it by sailboat he reckons!

In my own experience, we sent a box of indexed blog data tapes from US west-coast to east-coast by overnight shipping rather than over the Internet’s pipes, which would have taken a week at a minimum.

Posted in Data, Internet | Leave a Comment »

Library Model for TV

Posted by kaushalkurapati on February 24, 2007

So what is the “Library Model”? You have a membership in a local library and you can check out certain number of books in a time period. How would that apply to TV shows? Imagine a connected network of DVRs or PVRs (digital/personal video recorders) and other machines (PCs) with video content on them. I should be able to pay for and pull anything from this network and watch whenever I want to. Simple! I am a subscriber who pays a flat fee of, say, $49.99 per month, which gives me permission to ‘check out’ 300 hours (10 hrs/day: 2.5x of what an average American wathces today) of TV shows/movies from the network. The network contains video content that users have chosen to upload and the network could in fact include libraries of shows from TV-stations themselves: NBC, HBO, PBS, BBC could charge per check-out of a show from their libraries and it will count towards your monthly subscription rate (flat or variable).

The power of this network is independence from a set broadcast schedule of course. It is on-demand video. But its much more than that. Lets imagine that the broadcast model won’t go away still. There is a certain power of ‘live TV’ still (sports, Oscars, you want to watch live I presume). The TV-stations may not put everything online or not every TV-station will put stuff online either. In such a scenario, the power of the network of DVRs kicks in. Users in the network may have stored interesting shows / movies that the TV-stations, movie-studios have not put online yet. I should be able to legally pay and pull the content from a networked-DVR and watch it whenever I want to. The copyright owner should get a cut for this model to work.

Ever since I have had a TiVo (since 2000), I have dreamt of this scenario. Even with a TiVo I am still at the mercy of what is being broadcast on the pipe. The recommender system on the TiVo is not smart enough to know all the shows I would like and so it misses some. But if I could tap into other DVRs/machines (PCs) that caught the shows I’d like, that would be awesome. I am willing to pay for this service and be legit about it but the technology is not there yet. Imagine a ‘search’ + ‘personalized recommender’ service over such a network. That is couch-potato-nirvana!

Isn’t YouTube doing this already? Well…no. YouTube has interesting material (sometimes) that users have uploaded but its missing all the great content produced by TV-stations and movie studios from across the world. Yes, they are trying to do content deals…not there yet.

Is Joost  the closest to this model? Yes and No. Joost just extends on the current TV viewing model, so it does not require disruption of user behavior. It is trying to minimize disruption to the current business model as well, which is smart in a way. Joost is trying to create a broad distribution platform for copyrighted video content. Content owners decide what goes on Joost. So it is a walled-garden approach. In that sense Joost does not come close to the Library Model described above. Viacom seems to be interested in finding an alternative to YouTube and has just licensed its content to be put on Joost. This reminds me of Yahoo giving a small search startup called Google to handle search, which was the right boost at the right time. Could this Viacom deal be that inflection point for Joost? It may just be.

Joost is not close to the full-viral p2p, library model for TV that I described because it is not letting users put anything and everything on its network. By tapping into millions of users’ libraries, video gems get on to the network that do not have any other outlet. 

Joost was chosen as one of the next net 25 startups to watch by Business 2.0.

Posted in p2p, Technology, Tivo, TV | Leave a Comment »

Search Engine Market Share: January 2007

Posted by kaushalkurapati on February 24, 2007

comScore released US search engine market share statistics for January 2007:

  • Google: – 47.5% (3.3B searches); 1xG
  • Yahoo: – 28.1% (1.9B searches); 0.59xG
  • MSN: – 10.6% (733M searches); 0.22xG
  • Ask.com: – 5.2% (361M searches); 0.11xG
  • AOL:  – 5% (342M searches); 0.105xG

A total of 6.9 billion searches were conducted in January 2007 (up 2% over December 2006). This represents a healthy 26% growth in query volume year-over-year.

A different perpective on search market share is offered here.

US Internet audience size is 175M. In terms of unique visitors to respective network sites, here is the breakdown: Yahoo sites (#1 129M), AOL (#2, 117M), MSN (#3, 115M), Google (#4, 113M), Ask.com (#8, 49M).

Posted in Search, Stats | Leave a Comment »

Search Engine Marketing in India

Posted by kaushalkurapati on January 18, 2007

Pinstorm in association with IAMAI (Internet and Mobile Association of India) has prepared a very interesting report on the state of search engine marketing in India. Some interesting stuff from it, along with my own analysis:

  • They consider Indian companies spending in India and Overseas companies spending in India — so companies (Indian or otherwise) targeting Indians.
  • They essentially ‘sampled’ extensively and inferred search volume, ad-coverage, depth of advertiser market, average ad-spend, etc.
  • 1Billion searches in a month in India vs. 6.7 Billion searches in the US in December 2006. That number seems fairly high at first glance given that PC penetration is much lower in India compared to the US. On the other hand, see this — 37M Internet users in India (Sep 2006) vs. 250M Internet users in the US: ratio of 6.75:1. That ratio mirrors the search volume ratio exactly! (6.7 billion to 1 billion).
  • According to the report the number of advertisers targeting Indians is around 40,000. Thats a good size in itself, and will grow rapidly depending on (a) more users getting on to the Internet (larger target audience), and (b) more vernacular content and India-specific content coming online (more ad-sense shelf space so to speak).
  • US online ad market is roughly $10bn in 2006. Indian online ad-market, according to this report, is $52M — 0.5% of the US market. 
  • So who are the top spenders currently? Naukri (jobs), 99 acres (real estate), jeevansathi (online matrimonials – match making). Top categories? Banking / Financial services, Automotive, Matrimonial — all wanting to sell services to Internet savvy Indians (hence you may be making a certain amount).

Posted in India, Search | Leave a Comment »

Search Engine Market Share: December 2006

Posted by kaushalkurapati on January 18, 2007

comScore networks released December 2006 search engine market share data and it shakes out as follows:

  • Google – 47.3% (3.2 billion queries) – 1x G
  • Yahoo – 28.5% (1.9 billion queries)  – 0.6x G
  • MSN – 10.5% (713 million queries)  – 0.22x G
  • Ask.com – 5.4% (363 million queries) – 0.11x G
  • AOL – 4.9% (335 million queries) – 0.10x G

Search volume is growing 30% year-over-year. It was up 1% for December 2006 over November same year. I am a bit surprised that it was not up more in December, over November. I would’ve assumed that travel and online shopping should’ve driven more searches. On the other hand, may be niche travel and shopping engines are probably taking most of that growth rather than general purpose engines. Also, what may have balanced the uptick in shopping and travel searches, is a lull in school / work related search activity owing to vacations and such. The yearly search volume growth (30%) is healthy though. It would be nice to see a break up of the key search categories that are gaining in that incremental 30%. Some of the top ones may be as follows: Local, Health, Travel, Shopping, Reference, Sports and Entertainment. The usual suspects I suppose.

For more on the search market’s winners and losers see Danny Sullivan’s post at  Search Engine Land.

Posted in Search, Stats | Leave a Comment »

Stock Exchange for Start-ups: good or bad?

Posted by kaushalkurapati on December 4, 2006

Rajesh Jain has insightful posts on his blog usually and I am a regular reader. One of his latest posts is about creating a stock exchange of sorts for start-ups to raise capital in order to address the inefficiencies in the Indian VC market to be able to fund fledgling entrepreneurs. From my experiences in advising friends’ startups and having seen their trials & tribulations in getting the venture funded and flying, I am not sure I agree that a financial marketplace is the right way to fund start-ups. Here are my counter points:

  1. No VC Rolodex Power: Start-ups get funding from VCs for sure but they need hand-holding early on to allow their strategy, product vision to mature. A good VC often provides such ombudsman like guidance and more importantly opens doors/introductions to potential customers, partners through their rolodex network. A stock exchange type setting may help the young company raise capital but it may not be able to provide invaluable contacts to generate those important first customers and cash flow necessary for the working capital of the company.
  2. Syndicating further rounds of funding: VCs typically can get other VCs to get to invest in the company as syndicates in future rounds of funding. This is another example of VC-rolodex-power. Stock-exchange based investors cannot provide that kind of future investment potential.
  3. Capital Stability: VCs provide capital, but typically wait 5+ years before they start to look for exit routes for their portfolio companies. They provide that somewhat longterm umbrella of capital stability. Exchange based investors could be fickle. They have been trained on regular stock markets of the world, which are focussed on earnings growth. Such investors lack the patience that start-ups need from their investors. Seeing no cash flow or foreseeable earnings, they would want a quick exit from the company thereby draining the company financially and forcing the company to focus on short-term financial vagaries of investors instead of getting a solid product to the market.
  4. Market liquidity: robust stockmarkets are highly liquid. Start-up marketplaces could be devoid of liquidity and may not provide individual investors with exit-routes when they want to get out of a stock. The investors could be harmed themselves with such exchanges.

It is hard for entrepreneurs to raise capital and this is true anywhere, US or India. The US VC market may be more deep enabling higher risk taking from VC funds, which means more ventures get funded. Lack of depth/variety in venture business plans themselves could be a reason for VC firms/funds in India not funding as many ventures as their US cousins. Lack of entrepreneur track-record is a major reason I’d suspect VC firms in India may not invest easily. This takes time to mature. Especially as technology penetrates many facets of Indian life, more and more people will create ventures to address the needs of a large market and further, the VCs will now fund those start-ups as the market is now big enough to support the 5x, 10x, 100x returns VCs crave for. As some of these ventures achieve success, executives cashing out from such ventures spawn out and create new ventures — now, with success on their resume, they should find it easier to raise capital than the first time around.

Posted in Innovation, Management, Technology | 1 Comment »