Coming To A Screen Near You: Veveo Beefs Up Mobile Video Search With Personalization, Recommendation & The Wisdom Of Crowds; Gears Up For Handset & Operator Deals
Mobile Search Expert / AltSearchEngines
In-Brief: An in-depth analysis of Veveo’s vtap mobile video search offer and product roadmap.
Regular readers will recall that MSG has tracked the rise of exciting mobile vertical search engines from the start. Why? Because the mobile screen real estate is limited and so is users’ patience. Combine these factors and we have the perfect Petri dish conditions for vertical mobile search. In my view, there’s no need to “out google” Google (and Google is so good at what it does that any efforts in this direction waste resources.) Instead, carve out a niche, develop some clever algorithms and maintain a sharp focus on the customer, and you’ll have much of what it takes to play in this space.
Many companies are rising up to embrace this model, which is one reason I so look forward to cross-posting my pick of super-cool search engines and services at MSG and AltSearchEngines come June. BTW: If you are a vertical search engine active in mobile – or have ambitions in the mobile space soon, then I encourage you to contact my PA, Andrea Henninge (andrea@msearchgroove.com) for a briefing.
One vertical search engine that I have placed high up on my must-watch list from day one is Veveo. This recent article in BusinessWeek is a good primer but it only scratches the surface of why we’re sure to see much more of this mobile video search engine over the next months. I know from my briefing with Daren Gill, Veveo VP/GM responsible for the company’s flagship mobile search offer vtap, that the company is set to offer paid search results soon. It’s also going to announce a tie-up with a major European mobile operator “as early as this summer.” Search advertising is also on the roadmap.
Thanks again to Daren for walking me through a demo, being forthcoming with answers to ALL of my questions and helping me create a personalized mobile channel/feed that alerts me to all things related to Into the Wild (my current fav movie). I should add that creating the channel was a piece of cake – so high marks for usability in my book.By way of background Veveo – with HQs in Andover, MA and Bangalore, India — was founded in 2004 by Murali Aravamudan and Ajit Rejasekharan. The company is VC-funded and, in May 2007, reported that Promod Haque, managing partner, Norwest Venture Partners, invested another $7.8 million in the company. (This post from Unwired India picks up on the investment and puts it into perspective into perspective.) That brings total investment to $28 million.
To read the entire article, please click here.
Caronte Web
Federated Search
Highlight Health
MSearchGroove
Ramenos Blog
Reel SEO
Software Abstractions
The Genius Files
The Next Web
May 28th, 2008 at 10:13 pm
‘v-tap’ : Google opportunity in mobiles
http://unwireindia.com/2008/05/v-tap-google-opportunity-in-mobiles/#comment-497
May 29th, 2008 at 12:05 pm
Looks like these so called “analysts” will get carried away by any company propaganda. Its total wishful thinking to say mobile search giants may tie-up with this company. That too for what is basically a metadata search. If you strip away all the marketing hype and jargon in the article, metadata search is what this so called “video search” is. Its a simple metadata text search just like any other search engine. That technology has been around for so many years. No search major is that brain-dead enough to have to go out and buy another metadata search company. They crawl 100,000 sites. Major search engines crawl billions of sites. I dont see why this search engine is more exciting compared to any one of the other thousands of search engines out there.
May 30th, 2008 at 5:40 pm
This is Daren from vtap. Most of the excitement about vtap relates to its use on mobile. For devices that do not have a full qwerty keyboard, a metadata index that is built at the character level as opposed to keywords is extremely convenient and useful for helping users to find what they are looking for with minimal effort. Many of us have experienced how easy this is when we search contacts in our address book and more results appear with each additional letter. When this approach is applied to 160 million web videos, as opposed to say just 500 contacts in your address book, and works just as efficiently — I can confidently say that is definitely one of the features our vtap users are excited about. In the end, with any search engine, it is not about how many sites you crawl but whether you can efficiently deliver what the users are searching for.
May 30th, 2008 at 9:33 pm
Another excellent mobile web video application that I have used is Avot MV. You can check out the comparison of Avot MV with vtap @ http://avot.tv/corp/index.php?option=com_wrapper&Itemid=48
May 31st, 2008 at 10:35 am
@Daren G-
Your focus on character by character search as in a local phone book is interesting. Character by character search may be cute from a demo perspective, but doesn’t scale in a cost effective manner. Instead of one search for every keyword, you will be incurring server and bandwidth costs for every character. For large dataspaces and for large user numbers, the cost of server capital and operating costs (especially with today’s power costs) will be so high, today’s mobile eCPMs that one sees will not be able to sustain the costs incurred. Character search has been considered before and not adopted for this cost reason. Plus, in your system you are incurring media delivery costs as well since you are streaming out media from your servers. But as you mentioned, you won’t be able to monetize the video streams since you don’t own the content. Do you really have a business case that shows this system to be viable as a commercial system beyond just a demo?
June 1st, 2008 at 11:19 am
@Jon-
Not sure what is meant by “large” dataspaces, but we are indexing over 160 million videos and it is commercially deployed today. We definitely believe in the long term business potential and the activity of over 1 million unique visitors on vtap last month helps support our models. We actually do not fear the scaling issues with respect to the incremental search features; it’s the transcoding costs that are much more resource intensive. However, we feel the transcoding will diminish as opposed to increase as more content is offered in formats which the devices can play.
June 1st, 2008 at 2:05 pm
@Daren-
Your answer on your traffic number now makes sense. I am closely familiar with the operations of a search major and can only say that 160 million records and 1 million unique visitors (with probably one or two visits each) a month are just peanuts on the web. Any scalability problem can easily be covered by putting in a good number of servers but the question is whether that is cost effective and practical at web scale. Also, the day when devices support most web video formats are at least a few years away and you will have to incur media streaming and transcoding costs without being able to monetize the same.
June 2nd, 2008 at 4:35 pm
Very informative article. Myself and a colleague from one of the “AltSearchEngines top 100″ were intrigued by Veveo spokesman’s claim of having built up 3 million global content topics. By using the vtap web site we also found where those topics are coming from. If you go to http://vtap.com and search with any keyword, it gives a list of “Related Topics”. If you click on any of those, it gives videos on that topic and in addition gives information about that topic from Wikipedia, IMDB or last.fm. Now, Wikipedia has close to 2.4 million topics, the other two databases probably have another 600 thousand more unique topics, all put together 3 million topics! So, it appears Veveo’s 3 million content topics are coming from Wikipedia, IMDB or last.fm. These open source projects are an awesome source of information and nothing wrong in leveraging them, but it would have been better if Veveo had acknowledged the sources instead of appearing to claim to have built the content topics themselves.
June 3rd, 2008 at 12:17 pm
I will try to be more clear and overt in acknowledging sources such as Wikipedia, IMDB and last.fm which are great repositories of user contributed data and, as you mention these sources are displayed in our product. Our topic index is also influenced by many of the UGC sites we index such as YouTube, DailyMotion, Metacafe, etc. The real problem we solve is not “building the topic list” but correlating multiple data sources to understand each topic better (music artist, movie, tv show, actor, etc.) AND, more importantly, correlating relevant video clips (again from multiple sources) to each topic.
June 4th, 2008 at 5:22 pm
Daren: Wikipedia has much more than just topics. Wikipedia articles have human edited text descriptions from which keywords that correspond to that topic can be extracted. The articles also list related topics and external links all of which are useful in “correlation”. IMDB articles have keywords, all kind of “related” information and rich descriptions. Similarly for last.fm. Each web video also has its own metadata and keywords. Once you have a web video with its metadata/keywords and a corpus of open source articles with their own metadata/keywords, there are well known statistical techniques to associate the two. You can see such “correlation” in action in systems that provide the “Related” function. These are well known techniques. Thanks for the article and the discussions. They helped us understand what you are doing.
June 5th, 2008 at 4:33 pm
As recently as early May Veveo’s CEO has been quoted in Indiatimes as saying vtap has one hundred thousand (one ‘lakh’ in Hindi) users worldwide (http://economictimes.indiatimes.com/Features/Special_Pages/LiveITUP/Vtap_Video_search_engine_on_mobile/articleshow/msid-3023455,curpg-2.cms : see second para). Now they are claiming one million unique visitors last month which is difficult to believe. They are also advertising very aggressively as I could see from the many WAP sites I visit. With the low CPCs on mobile ad networks, it is not difficult to build unique users. Once a user clicks on an ad and goes to vtap she is a “unique user”. A better measure would be to see what proportion of these unique users are repeat users who come to vtap on their own and not by clicking Veveo’s ads. That way, any ad effects could be removed. I didn’t find the service interesting at all. The video quality was very poor and it takes forever for videos to load.
June 21st, 2008 at 10:36 am
Character by character search engine Boopsie is in my opinion the best such search engine in that category for mobiles. Check it out at http://boopsie.com
Their “smart prefix search” is really cool. To search for “Jacqueline Kennedy” all you need to type is “ja ken”, ie, only the few characters from the search words. Boopsie has indexed a number of information sources such as Amazon, CitySearch, Wikipedia and many others.