• Print

The Twitter Gold Mine & Beating Google to the Semantic Web

twitterads.jpg

There’s always been jabs at Twitter for not having a viable business model and the chatter has increased in the current economic climate. In a recent interview Evan Williams, Twitter CEO, said “We had planned to focus on revenue in 2010 but that’s no longer the case, so we changed the plan quite a bit… We’ve moved revenue higher on our list of priorities…”.

I believe Twitter, potentially, has an incredible business model.

In The New York Times R&D Labs, where I work, we’ve been talking a lot about ‘smart content’, both in relation to advertising, search and news delivery. For the past 157 years (that’s how old the newspaper is) we’ve essentially delivered ‘dumb content’ to people’s doorsteps. You and I, irrespective of interests, location etc. have received the same newspaper on our doorsteps every morning. We’re beginning to explore ways to make content smarter, to understand what you’ve read, which device you’ve read it on and your micro level interests—making the most important news find you, instead of you having to find it.

This also changes the advertising model where ads become even smarter. Sure, ads are at about a 1st grade reading level now; with adsense and cookies, the ad networks have half an idea of what I’m interested in, but they aren’t exactly smart about it. Just because a friend sends me an email about a baseball game doesn’t mean I want to see ESPN ads in my Gmail.

So what does this have to do with a Twitter business model? Twitter, potentially, has the ability to deliver unbelievably smart advertising; advertising that I actually want to see, and they have the ability to deliver search results far superior and more accurate to Google, putting Twitter in the running to beat Google in the latent quest to the semantic web. With some really intelligent data mining and cross pollination, they could give me ads that makes sense not for something I looked at 3 weeks ago, or a link my wife clicked on when she borrowed my laptop, but ads that are extremely relevant to ‘what I’m doing right now’.

A quick perusal of my Tweets shows that I live in Brooklyn, NY, I work for The New York Times, teach at NYU/ITP, I travel somewhere once a month for work, I love gardening, cappuccinos, my Vespa , U.I./Design and hardware hacking, I’m a political news junkie, I read Gizmodo & NYTimes.com and I was looking for a new car for a while, but now have a MINI and I’m also friends with these people. That’s a treasure trove of data about me, and it’s semantic on a granular level about only my interests.

If I send a tweet saying “I’m looking for a new car does anyone have any recommendations”, I would be more than happy to see ‘smart’ user generated advertising recommendations based on my past tweets, mine the data of other people living Brooklyn who have tweeted about their car and deliver a tweet/ad based on those result leaving spammers lost in the noise. I’d also expect when I send a tweet saying ‘I got a new car and love it!’ that those car ads stop appearing and something else, relevant to only me, takes its place.

And it doesn’t have to be advertising delivered on their site alone. One of the great successes of Twitter has been their APIs and the wonderful applications and sites that users have built with them. Why not build out an advertising or search API that delivers the latest micro level tags or ad links of users interests? There’s a plethora of opportunities with this data, and if it’s done right it becomes enticing and engaging, not annoying, irrelevant and outdated.

tags: , , ,
  • http://twitter.com/t411 Den

    One of the possible business models – user-defined information systems on the top of Twitter: http://servletsuite.blogspot.com/2008/12/information-server-on-top-of-twitter.html

  • http://www.adelph.us William

    I want to begin by first stating the fact that Twitter is a close source application and as such members that give their content to Twitter so that Twitter can earn millions off it are locked in, and also more than likely since they are locked in will not be included in much share of any revenue that twitter will eventually earn on their content.

    Open Source can give a bit of a guarantees that members are included in the revenue discussions as well as revenue sharing because members have a choice and are not locked in. If members want to leave the service and create one that meets there needs they are free to do so. This puts members in control of the value they bring and create. Twitter is closed source so members will more than likely be outside of revenue generation discussions as well as any revenue share.

    I also want to understand why it is assumed that Twitter and its founders are the smartest guys in the the room. Are Twitter and its founders and managers the only people in the entire world that can build a micro blogging service with a revenue model. To me this is the arrogance of closed source.

    We have an Open Source Micro Blogging service that already has some of the features that you have mentioned in your blog.

  • http://woganmay.com/ Wogan

    It’ll require a bit of semantics, but something like this is workable. Essentially, every user ends up with a smart adserving bot – not a bad idea.

    In fact, come to think of it, I’m more likely to buy stuff if they’re advertised to me in that way. Now if only they could find someone to actually build it :)

  • Falafulu Fisi

    Nick said…
    Twitter, potentially, has the ability to deliver unbelievably smart advertising; advertising that I actually want to see, and they have the ability to deliver search results far superior and more accurate to Google, putting Twitter in the running to beat Google in the latent quest to the semantic web.

    I am afraid that you’re proposing a wishful thinking here. I don’t know twitter at all, since I am not a user, simply because I don’t like information overaloading, second it is not important to me to track what other internet users are saying or passing around because they’re not important to me (workwise or self education). I view them as simply useless and time-wasting.

    The kind of thing you’re talking about Twitter, I think is impossible for them to develop, and if they do, I think that Google, Microsoft, Yahoo etc, will develop something more superior to any sort of data-mining that twitter is doing.

    The big 3 had already done ground-breaking research in online advertising which they published those proceedings in ACM Journal:

    Data Mining and Audience Intelligence for Advertising (ADKDD – 2007).

    There is also a ADKDD – 2008 in December that the proceeding will be coming out early next year.

    My point is, unless Twitter researchers have invented something proprietary which they haven’t publish in the literature and also superior than existing techniques, then I think that they have a chance (in the short term). Mind you that the big 3 (Google, Yahoo, Microsoft) do have an army of PhDs that do nothing but R&Ds , ie, inventing new things and improving existing things (algorithms, etc). This means that whatever an advantage someone holds over a proprietary algorithm , it won’t before someone else eventually found out about it independently and this often happens. This is where an R&D teams can be advantage, because they evolve and keep inventing new algorithms. So, whatever twitter is going to invent will eventually be found out by researchers from the big 3 (or any active researcher out there) and this is independent if they’re aware of what twitter is doing or not.

    I doubt that Twitter has an R&D division.

  • Falafulu Fisi

    it won’t before someone else eventually found out about it

    meant to say:

    it won’t last before someone else eventually found out about it

  • http://absono.us whitneymcn

    I’m not certain there’s a real gold mine there once I start thinking about the numbers involved; possibly better than just dropping banner ads on twitter.com, but it doesn’t seem rock solid.

    For a rough model, assume that Twitter currently has 1,000,000 users (high, I know), and that about 60% of those users are within the US and have been active within the past 30 days.

    For any given ad serving day, we can split that into two groups:

    1. Those that have tweeted enough enough information that they’re “targetable” and tweet something that indicates ad-friendly intent (again overestimating, I’d go with 2%).
    2. Everybody else (98% on any given day).

    So if Twitter can get a $100 CPM on these “targeted, intent-driven” ads and a $0.50 CPM on ads to everyone else, sending one ad per day to each active user would get them a daily gross revenue of something like $1,500–around $550k of annual revenue.

    Given that events that are high value to advertisers (like your looking for a car example) occur relatively infrequently in people’s lives and they won’t *necessarily* tweet about them every time, I’d guess that the high-value end of things would actually be much smaller than 2% daily.

    It’s interesting to plug in different numbers and play around, but I’ve had a hard time coming up with any arrangement that makes a purely ad-driven revenue model work well for Twitter. Could be a nice supplement, but I’m not seeing a gold mine in there yet.

  • http://www.karimgargum.com Karim

    I agree with @Whitneymnc. I think that with a site that’s so heavily reliant on user-generated content, you have to flip the advertising model. Focusing on affiliate agreements and allowing users to tweet for profit (a la magpie offering).

    Perhaps there can be a virtual ‘job board’ with different products available for twitter users to promote.

    Companies that have interesting products/services would be better off creating their own twitter profile, and offering followers exclusive, alpha/beta access to new products and then encouraging them to spread the good word. I don’t think conventional advertising techniques will work here.

  • http://www.yourcompanynewsletter.com Scott

    I think semantics is the right way to go with this, but i also think simple text search/match could be a basis for a start. I see lots of people discussing a topic I’m interested in (newsletters), and have set up a search in TweetDeck. It sends me a list of potential prospects. Being able to target that list would be helpful… maybe even easy.
    Scott

  • bowerbird

    > putting Twitter in the running to beat Google
    > in the latent quest to the semantic web

    finally something worthwhile from the “semantic web”
    – a superior ad-server! let us all now make a parade!

    oh wait, that’s not worthwhile in the slightest, is it?

    ok, nice try, now back to the drawing board, guys… ;+)

    seriously, i’m reminded of dave winer’s position that
    a perfectly-targeted ad ceases to be an ad because
    it has become _information_ that the person _wants_.

    color me skeptical that twitter can _deliver_ that, but
    hey, i never thought twitter would be much more than
    a passing fad with minor compelling value long-term.

    oh wait, you mean it hasn’t really yet proven itself to be
    anything more than that? oh. well, call me when it has.

    -bowerbird

    p.s. this is not to say twitter won’t be with us forever:
    lovers always want to know what the other one is doing;
    but after the honeymoon phase, that usually goes away.

    p.p.s. not only did dave winer say this much better, but
    without confusing _it’s_ with _its_, _you’re_ with _your_, etc.

    > device you’ve read it on and you’re micro level interests
    > they could give me ads that makes sense
    > A quick parusel of my Tweets shows
    > mine the data of other people living Brooklyn
    > and deliver a tweet/ad based on those result
    > something else, relevant to only me, takes it’s place.
    > latest micro level tags or ad links of users interests

    good thing nick works in n.y.t. research, not copy-editing…

  • http://www.tnsmediagroup.com Trevor Wolfe

    Not sure where whitneyman got his numbers…

    Even the most conservative numbers from HP Laboratories, Hitwise, Comscore, TNS Compete, Nielsen Online have Twitter users over 4 million, and growing. Perhaps you meant 1.2 million “active users” http://blog.compete.com/2008/05/15/twitter-traffic-growth-usage-demographics/

    Not sure if twitter has the R&D to create the right ad-delivery system, but you’re right the data is pretty ripe. Not sure if advertisers will look at the targets from the purchase funnel, so I don’t necessarily think that consumers in their last stages of product research for a car will be their only options. Twitter users are early adopting, high earning, influential subset of American consumers. Getting the message in front of them might be appealing in of itself.

  • Anonymous

    “I would be more than happy to see ‘smart’ user generated advertising recommendations”

    That is the fundamental mistake in your reasoning.

    Advertisers have a long history of not being interested in what you want to see. Advertisers are interested in forcing you to see what they want you to see. Print, TV, phone, net; it’s all junk mail. Advertisers might be interested in more cost effective junk mail, which could include niche targeting, but assuming any kind of empathy or compassion for the victim is naive.

  • http://absono.us whitneymcn

    Hey, Trevor –

    In March TechCrunch gave these numbers as coming from “a source close to the company:”

    Total Users: 1+ million
    Total Active Users: 200,000 per week
    Total Twitter Messages: 3 million/day

    [ TC story here:
    http://www.techcrunch.com/2008/04/29/end-of-speculation-the-real-twitter-usage-numbers/ ]

    As has been noted there and elsewhere, Twitter has many, many more “registered” users and visitors. Also note the rather low “active” number there–I’m assuming a much higher active percentage in the comment above.

    It’s an interesting exercise regardless of the numbers you pick…try plugging the numbers into a spreadsheet and see what ad frequency and cost to advertiser you end up with to get to a good revenue number for the company. I couldn’t get anything that didn’t require pushing either growth and activity projections, ad frequency, or ad cost up to pretty bizarre levels, but I’ll admit I could have missed something.

  • http://blogs.timesunion.com/classconflict Brandon J. Mendelson

    Great article, but the reaction to Magpie and TwitAds indicates a knee jerk hostility to any sort of advertising on Twitter. So if they were to take this approach, perhaps it would be through a redesigned homepage and not in the stream?

  • http://www.thenetworkgarden.com Mark Sigal

    Here is an attempt I made to articulate a ’structured’ tweet service based on a consumer, small biz, NY Times and procter & gamble segmented universe.

    The idea is a tiered model whereby ‘basic’ twitter remains free but a premium structured service is offered for online brand builders and commercial businesses.

    Twitter-nomics: Envisioning Structured Tweets
    http://thenetworkgarden.com/weblog/2008/05/twitter-nomics.html

    Check it out if interested.

    Mark

  • http://Http://www.TheContentWrangler.com Scott Abel

    Thanks for covering this subject. Making content smart is a critical factor in delivering on the promise of XML. That’s why an intimate group of content pros are gathering in Palm Springs, January 29-30 at the first annual Intelligent Content conference: http://www.intelligentcontent2009.com.

    The event aims to help attendees understand what’s needed to create, manageband deliver intelligent content solutions that solve business challenges and provide positive return on investment.

    If you are interested in attending or becoming a sponsor, contact me at scottabel@mac.com or 760.969.3249.

    This topic will also be covered from a marketing angle at our Web Content 2009 Tampa Bay conference in Clearwater Beach Florida this February: http://www.webcontentconferences.com/tampabay. This event will explore how Web 2.0 technologies are impacting web marketing strategy.

  • http://www.twitterrati.com Mark Evans

    Interesting theory but how well Twitter can deliver relevant advertising seems to be based on how much information you disclose within your Tweets. What if you don’t disclose any information about where you live, what you do for a living, what you buy, or your interests? Then, how can Twitter deliver relevant ads?

    Mark

  • http://pagerush.com Adam from PageRush

    Twitter posts are no more semantic than email in Gmail. Google has the same or better advertising potential with Gmail, but anyone who uses Gmail will tell you how poorly targeted the advertisements are. This isn’t from lack of inventory, but lack of natural language processing algorithms.

  • http://khaitan.org/blog Indus Khaitan

    Good ideas around twitter monetization. It may be easier to analyze semantic data from 140-character messages (no markup, no presentation kludge, just pure text).

    However, the same could be ridden with challenges as most of the time in short sentences you would be able to discern only the entities and not come out with actual classification/categorization. (eg. A Mini cooper was side swiped by a hit-and-run http://twitpic/) Do you show ads about insurance or cars?

    The best advertising possibility is ads which are based on entities found in the messages rather than intent.

    Indus Khaitan

  • http://www.syntience.com Michael P. Gusek

    I do not feel this was written by someone who clearly understands what semantic technology actually is…

    If you could see what the market actually is, you would see companies cashing in big on technologies that do not work. Smoke and mirrors, folks.

  • Ads Zoooks

    Yes, the ads I want are all adverts directed to you.
    That way I do not get any unless I want them, then I will pester you. You seem to like ads.

  • http://sean.horgan.net Sean Horgan

    @Falafulu Fisi

    and there was an army of PhD’s at IBM, Microsoft, and Yahoo! before Google came along.

    It takes more than just a bunch of academics in any one company.

  • http://www.zerodaydefense.net David Scott Lewis

    Saying that the New York Times delivers “dumb content” is disparaging your employer.

    In fact, I read the New York Times (online, since I live in China and can’t even get the IHT on a timely basis) for its breadth, not depth. This is an important point NOT to be overlooked. In essence, I consider the Times an aggregator of the best content, even though it’s mostly original Times content. No publication beats the Times for it’s breadth (well, maybe the IHT, but this is pretty much the same thing as the Times, just a bit more filtered).

    I also find Arts & Letters Daily must reading, as well as Techmeme, the South China Morning Post, Good Morning Silicon Valley (from the San Jose Mercury News), and several trade-specific sources that also have — relatively speaking — broad coverage within their industry sector (e.g., the Solar “Knowledge Centre” section in New Energy Finance). Yet, nothing beats the New York Times for it’s breadth. I could do with less coverage of New York City and State of New York issues, but I generally skip these sections/articles, and read from just about every other section of the paper.

    So, now that I’ve rallied behind the Times, I’ll get to your main point.

    What you’re suggesting certainly makes a lot of sense, but to think that Twitter will either provide or enable the type of solution that you’re proposed is probably more along the lines of wishful thinking. Although semweb technologies — along with many related technologies and methodologies, such as automatic classification, k-means clustering, metaphor frameworks, source-based link analysis — are showing great promise, they’re not yet ready for prime time.

    See the recent ReadWriteWeb piece on the best semweb offerings in 2008. Well, even the semweb-specific services have a long, long way to go before being viable for a non-geek audience, an audience that the Times would likely target.

    Bottom line: Don’t get your hopes too high. What you’re suggesting is a long way from becoming a reality.

  • http://blog.mastermaq.ca Mack D. Male

    I don’t think Twitter has any more information about me than Facebook does, but that hasn’t helped Facebook in the advertising department! I can’t see what you’ve suggested happening anytime soon.

  • Nick

    David Scott Lewis: I didn’t say NYTimes content was dumb. I said we ‘deliver’ dumb content, meaning the content has no idea about the reader, or whether they are even interested in the topics at hand. You and I both get the same sections; Sports, Business, National, etc. delivered to our doorsteps, or computer screens, every morning irrespective of our particular interests.

    Ads Zoooks: Huh?!

    Adam from PageRush: Good point, but my Twitter posts are inherently more semantic because they are written by me not by someone else. My Gmail inbox is riddled with mail from other people about subjects I have absolutely no interest in, and in turn the Google ads are mostly inaccurate too.

  • http://valleyblogzine.blogspot.com valleyblogzine

    Doesn’t your facebook profile have the information you have stated as being available through your tweets?

  • http://aalaap.com Aalaap Ghag

    Solid. I like the sound of this. The fact that tweets show up more often when I search for something in Google only strengthens this point. This needs to happen now!

  • http://www.ichoosr.com Bart Stevens

    Pls have a look at VRM (in general) and a project by Adriana Lukas called The Mine, I think she can help you guys.

    Cheers

    Bart

  • http://konterkariert.tumblr.com/post/42355329/twitter-starts-web-data-mining-with-summize Christoph Möller

    read my article about Summize, Twitter investor Fred Wilson commented “great post”

    “Twitter starts Web Data Mining with Summize”
    http://konterkariert.tumblr.com/post/42355329/twitter-starts-web-data-mining-with-summize

    I also see a big potential in Twitter.

    Why did Twitter buy Summize?

    “Summize’s mission is to search & discover the topics and attitudes expressed within online conversations.”

    Jay Virdy: “At Summize, we assembled a small, quirky, but highly efficient and experienced team to build a powerful platform to extract user opinions from blogs and review sites. Dr. Abdur Chowdhury, our cacographic Chief Scientist, applied machine learning techniques to understand how users express sentiment using common words and polarizing phrases.

    Om Malik: “Just as AdSense serendipitously turned Google into a giant cash register, with Summize, Twitter can take the first step towards a business model.”

    See also the startup Peer39 – they are also doing Sentiment Analyis and their business model is Advertising – http://www.peer39.com/

    “The next revolution in advertising? Peer39 thinks it’s semantics”
    http://venturebeat.com/2008/06/27/the-next-revolution-in-advertising-peer39-thinks-its-semantics/

    Peer39’s firepower comes from a set of executives and board members with backgrounds in either semantics or advertising. Eytan Elbaz is on their board. He helped invent Google AdSense and sold Applied Semantics to the search giant.

  • http://www.hivesight.com Elad Kehat

    Good post. Mind that it’s not just twits that contain great information for semantic ads to build on – full length blog posts, occasionally profile wall posts, and emails too contain such info.
    Also, advertising isn’t the only application. A better understanding of consumer tastes applies to market research as well.

  • Michael Sander

    I’m with the other’s here… social networking sites like facebook have just as much information as twitter. And on facebook the data is better organized and the social graph tends to be of higher quality. On Twitter, users are more likely to follow and be followed by complete strangers. Whereas on Facebook most people are actually friends, or at least acquaintances, with everyone in their social graph.

    I dunno… but if I was a marketer, I’d think Facebook has got more targeted data.

  • http://konterkariert.tumblr.com Christoph Möller

    YES, Twitter is a gold mine, that is what I’m saying, too

    please read my blog post about Twitter and their acquisition of Summize and Sentiment Analysis

    http://konterkariert.tumblr.com/post/42355329/twitter-starts-web-data-mining-with-summize

  • zuni

    Yay? I am not a fan of advertising so forgive me for not sharing your enthusiasm. The technology might be applied in other areas but ask yourself, what does advertising do for you? Pull out your wallet and pay for some cleverly thought up product or service. I know, it makes the world go round but these days the world needs to slow down and work together to solve some serious problems. Enough time has been wasted on a how to convince people to want the newest new that they can’t afford. (I am mainly speaking of the level of advertising that is the most pervasive in our culture/world and in my opinion is the lowest form of serious thought and intention. Manipulation through the mind by the mind)

  • Dan Gellert

    Interesting article and a concept that I am sure has been discussed at Twitter numerous times. However, I am betting that they run into 2 walls each time: First wall – the lack of scale of the Twitter ecosystem will make it uninteresting for advertisers. The whitneymcn math (which I agree with) makes sense – and it actually is even worse from the advertiser perspective. Mini Cooper doesn’t want to buy an ad to Nick Bilton – they want to buy 1MM impressions to 1MM NIck Biltons – something Twitter does not yet have the ability to offer.

    Wall #2 is pure technology. Twitter and natural language processing (or semantic analysis) are far away from each other in terms of technology requirements. It is extremely challenging to build a robust, scalable, 100% accurate NPL engine (one that understands the different between Tropic of Cancer and Cancer for instance) and to say that Twitter should just start doing semantic advertising around UGC is a far leap. An advertiser is only going to advertise on Twitter if they know for certain that the user wants a Mini Cooper and is not bitching about his Mini Cooper.

    Interesting thought, but I hope that Twitter goes another monetization route.

  • http://byjoeybaker.com Joey Baker

    Great, if not new idea. Yes, advertising that directly targets the user is the holy grail of online advertising. But what you’re suggesting for Twitter is something that Google, Facebook, MySpace, etc … should all be able to accomplish – easier.

    All of these companies have tons of your data – more than twitter can learn from brief tweets. Facebook ads have been more-or-less successful, google somehow hasn’t really been able to leverage gmail and other google apps as that datamine (yet).

    The problem with the proposal is that a) it isn’t unique. Not saying it won’t work, but it has been done before. b) its really, really, really, hard to do well. If facebook and google have been trying, it’s not likely that twitter, which has far less money to spend on R+D will be able to implement a system out of a more complex data pool.

    @Nick: I’d challenge you with this: what ads would you bring to yourself based on the data you shared? Oh, and then go try and sell those ads. Oh, and once you’ve done that, be sure to have a enough variety of ads to target *everyone’s* ad interests.

    It’s not an easy thing to do.

    Again, I’m not arguing that you’re wrong, just that you’ve identified a nearly impossible goal for twitter.

    IHMO

  • http://thegoldenspot.blogspot.com thegoldenspot

    Google ads are effective at selling products on their search page when I am searching for the same specific product they are advertising. How much more semantic(as in contextual) can you get? How can you beat that? Twitter’s gold mine is going to have to be something different than semantic advertising to beat Google at making money from advertising dollars (if that is your idea of winning). You are on to something when you write about using their api; I think you mean there are plenty of opportunities to use this service, not ‘data’. Like I said, no ad is more effective than when I am looking to buy something.

  • Martin Higham

    It doesn’t matter how smart the targetting is at the end of the day you can only push adverts about stuff that I might be interested in and too many of these will just annoy me.

    Google is in a much stronger position for advertising stuff that I want to buy. When I’m searching for for ketttles there’s a good chance that I want to buy one. Just because I regularly talk about kettles doesn’t mean I’m in the market for a new one.

  • http://adnerds.be Bart Muskala

    Interesting article, but as stated by many people already: not an easy task for the Twitter folks. Semantic advertising is – in a sense – great. It is unobtrusive for the consumer and it is direct and targeted for the advertiser. And an advertiser with a long term focus, might be interested. It’s like an adwords campaign, you don’t run them for a week or a month. Yet, this one is better…

    Next to that, I think the real beauty lies not in Twitter, but in the connections one can make with other social networks. Single sign on (like OpenID and – even better – Facebook Connect) will make it clearer what people want. What they talk about (on their blogs, facebook status, twitter, friendfeed comments, what pictures they shoot and put on Flickr, what cities they visit on dopplr or on the CitiesIVisited app on Facebook, what they bought on Amazon etc).

    Each and every company is trying its very best to bring advertising that is as close as possible to one’s needs, but all of them fail. OK, I heard the stories of people that found a vacation place while mailing friends about it on gMail, but most of the time it fails. 90% of the ads on Facebook are pure crap and totally untargeted.

    Suppose every person could be ‘recognized’ while surfing on the web (if he allows this?) enabling advertisers to address him with relevant ads and offerings. As Don Pearson explained to Jeremiah Owyang (Forrester) a few weeks ago: it would be nice that people shout somewhere on the web what they want, and that advertisers can react to that in a smart way. This is good for both the consumer as well as for the advertiser. But this requires OR an interconnected web with links between each and every initiative/site/company (Facebook Connect, FriendFeed, Twitter, …) OR it requires a smart browser that knows where to get its relevant ads, since the browser could ‘remember’ everything a user does. If I recall correctly, Firefox was already in some sort of testing phase with its semantic initiatives…

  • http://byronmiller.typepad.com Ron Milller

    This is a great post because you get that it’s all about targeting information to meet the needs of the reader, not dumping content in a one-size-fits-all fashion and assuming it will be relevant for everyone. I wrote a post today about the information overload we are facing on the internet today, and the semantic web, targeted content, offers us a way out. Yes, we are only at the very beginning of this journey, but what you are saying makes complete sense, and to me, is the only way to bring this avalanche of content under control to give us meaning and context. Great post.

    Ron Miller
    By Ron Miller Blog
    http://byronmiller.typepad.com

  • http://dbawill.org William Harris

    Brilliant piece, and most definitely the future of advertising. I’d gladly sit through a 45 minute infomercial on a Jeep SRT8 or Fender Bass guitars, but this LCD stuff revolts me and makes me hate the product being shoved down my throat. I hope more ad people are forward thinking like you are, it will mean the difference between a surging and successful marketing sector, and a complete failure.

  • http://www.germaise.com Scott Germaise

    “putting Twitter in the running to beat Google in the latent quest to the semantic web”

    Umm… Google has Googlebytes of search history. A lot of it user specific. And ads are generated based on keywords. Yes, we can call keywords semantic and talk about linguistic concepts like facets and tactics like query expansion, but at the end of the day, (and I mean TODAY), Google’s got actual user intent. Sometimes user intent is just lame use of the search bar for navigation. But the rest is for information. Sometimes for personal research, work, school, etc., but other times indicating purchase intent. Google, (and for that matter Yahoo, Live and Ask), have known about and been optimizing ad results for this for years. (I won’t even go into other aspects of behavioral advertising for which these companies and others have solid historical data.)

    Now, take Twitter. There may be a variety of ways for them to monetize. But getting meaning out of 140 characters will be hard. Especially given a penchant for odd abbreviations even worse then in IM. This is not necessarily an easy problem from a computational linguistics perspective. Engines need more to bite on to infer meaning. Some kind of advertising is clearly a way to go. But I’m not sure it’ll be anything really special; at least compared to what can be inferred and delivered based on user search. Twitter may come up with some other brilliant thing. Should be interesting.

    Oh, sorry for the long wind, but there’s one other factor. Twitter users are at this point still early adopters. The few mill unduplicated monthlies are Web 2.0 folks. They may be less likely then the general mass market consumer to click on ads. (At least, some of my personal experience with SEM seems to indicate that’s possible.)

    I like Twitter and hope they can work it out.

    Scott

  • Flooth

    Mr. Bilton, you’re idea of a twitter ad engine sounds like hell on earth.

  • http://www.openlinksw.com/blog/~kidehen Kingsley Idehen

    Nick,

    Twitter certainly has the inside track based network density. But I don’t know if they necessarily see the goldmine as being interlinked with exploitation of RDF based Linked Data.

    BTW – The same applies to the New York Times, since they can play in this realm too.

    The business model is all about “Data as a Service”, one entity provides the Lookup and the other the details. Here is a simple example:

    DBpedia [1] is a Lookup database derived from Wikipedia. The New York Times has a rich collection of Facts associated with the entities in DBpedia which produces a symbiotic relationship of immense value to others. The resulting Entity oriented databases simply expose data in a myriad of formats (as requested by the data consumer).

    Back to Twitter, the data in Twitter can obtain routes to structured data meshes (exposing demographic data in structured form amongst other things) from DBpedia. Also the “#” tags could be linked to structured data descriptions in DBpedia (Faviki & Zemanta already offer something in this ballpark).

    Links:

    1. http://dbpedia.org
    2. http://faviki.com
    3. http://zemanta.com

    Kingsley

  • http://headup.com Mike Darnell

    Cool article and interesting comments.
    I think that the ideas presented are pretty much the basis for our existence.

    As a semantic web addon for Firefox we are definitely hoping to capitalize on our ability to offer contextual, personalized content to our users.

    For example: offering a user a “heads up” that he should be gift shopping for his fiance’s birthday & reminding him how much she loves Amy Tan’s books is something we are already doing.

    Is this a bad thing?
    I don’t think so – I think it’s a decent way to offer value and make some profit while doing it.

    Personalization is entirely opt-in and users can revoke the permission they grant us to access 3rd party services at will.

    TO get a feeling of how this works check us out at http://headup.com : )

    Cheers,
    Mike

  • http://lifeanalytics.blogspot.com Themos Kalafatis

    Twitter can offer a wealth of information about how people behave, think and what they like. Some examples :

    Clustering the thoughts of twitter users

    Sentiment Mining for Amazon’s Kindle

    Know your customers – The Twitter way

    Semantics is the next big thing.

  • http://messel.typepad.com/ Mark Essel

    Other services are quickly updating their information streams similar to the microblogging in twitter (facebook, friendfeed beta).

    What we can expect as consumers, is that all (or most) of these microblogging platforms will need the assistance of an intelligent data mining service for targeted advertising.

    Google may still provide the best solution. If they don’t someone else will, and make a great deal of money doing it.

  • http://www.mccollam.com/jakeblog Bill McCollam

    Looking at driving ad revenue from twitter is very 2.0. Think real disruption… think changing the very way people exchange value…
    http://www.mccollam.com/jakeblog/2009/04/twitter-economics/

  • http://www.rebekahgreen.com Rebekah

    Whether the revenue-generating model that twitter comes up with is a game-changer or not is yet to be seen, but I love how interested and excited we all are to see what happens. At this point, we are witnessing game-changing events and consciousness happening on an almost monthly basis, which makes it an incredibly fascinating time to be alive. I firmly believe that even though there may be smarter people in the room on how to create the best new ad model attached to this exploding movement/company, it doesn’t really matter. That has always been true about smarter people not being able to play in the game. But the fact that the term “twitter” is now a globally-recognized household word, in what seemed like a two month period, means they will attract the best people they possibly can to develop the strongest and most user-accepted money-making model we’ve seen yet. Global name recognition is Everything in today’s world, and right now they really do have the world in the palm of their hand. It will be fun to see what develops. http://www.rebekahgreen.com

  • how to sell gold tampa

    I firmly believe that even though there may be smarter people in the room on how to create the best new ad model attached to this exploding movement/company, it doesn’t really matter. That has always been true about smarter people not being able to play in the game. But the fact that the term “twitter” is now a globally-recognized household word, in what seemed like a two month period,