Comments on: Whenever ad networks talk about their “targeting” remember the Netflix prize http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/ Essays on viral marketing, freemium, and social gaming Tue, 22 May 2012 16:59:00 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Adrian Bye http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2812 Adrian Bye Mon, 05 Oct 2009 00:01:04 +0000 http://andrewchenblog.com/?p=1243#comment-2812 A lot of ad inventory is driven via relationships. Therefore some ad networks will have significant competitive advantage simply due to the relationships they have developed. A lot of ad inventory is driven via relationships. Therefore some ad networks will have significant competitive advantage simply due to the relationships they have developed.

]]> By: adrianbye http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2226 adrianbye Sun, 04 Oct 2009 17:01:04 +0000 http://andrewchenblog.com/?p=1243#comment-2226 A lot of ad inventory is driven via relationships. Therefore some ad networks will have significant competitive advantage simply due to the relationships they have developed. A lot of ad inventory is driven via relationships. Therefore some ad networks will have significant competitive advantage simply due to the relationships they have developed.

]]> By: dherman76 http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2222 dherman76 Thu, 01 Oct 2009 00:30:52 +0000 http://andrewchenblog.com/?p=1243#comment-2222 Andrew, great posts as always. You nailed it on the head about the algorithms- it's all about the data that the algorithms are processing. Garbage In, Garbage Out. Andrew, great posts as always. You nailed it on the head about the algorithms- it's all about the data that the algorithms are processing. Garbage In, Garbage Out.

]]> By: huang http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2207 huang Sat, 26 Sep 2009 04:40:49 +0000 http://andrewchenblog.com/?p=1243#comment-2207 <p> Frozen winter is coming,do you like to buy ugg to preventing cold and keeping warm.As a pursuer for the most stylish ,is there any ugg boots you have in mind?I think you have no excuse anymore if you don't have a pair of <a href="http://www.realugg.com/ugg-bailey-button-c-52.html" rel="nofollow"><strong>UGG Bailey Button Boots</strong></a>.The <a href="http://www.realugg.com/ugg-bailey-button-c-52.html" rel="nofollow"><strong>Bailey Button UGG</strong></a> can be worn as a <A href="http://www.realugg.com/ugg-classic-short-c-28.html?zenid=ae999140b41bfc67c4fd4f9ec46cb39e" rel="nofollow"><STRONG>UGG classic short Boots </STRONG></A> or <A href="http://www.realugg.com/ugg-classic-mini-c-33.html?zenid=ae999140b41bfc67c4fd4f9ec46cb39e" rel="nofollow"><STRONG> UGG classic mini Boots </STRONG></A>. Hottest pop style.If you like the<a href="http://www.realugg.com/ugg-tall-boots-c-30.html" rel="nofollow"><strong> UGG tall <a href="http://boots%3C/strong%3E%3C/a%3E%2Cwww.realugg.com" rel="nofollow">boots</strong></a>,www.realugg.com</a> can provide high quality of <a href="http://www.realugg.com/ugg-tall-boots-c-30.html" rel="nofollow"><strong>Ugg Tall Boots</strong></a> in low price,Are you straitened for <a href="http://www.realugg.com/3-pairs-ugg-boots-insoles-p-2147.html" rel="nofollow"><strong>Ugg Boots Insoles</strong></a>,don't worry ,we have the best <a href="http://www.realugg.com/3-pairs-ugg-boots-insoles-p-2147.html" rel="nofollow"><strong>Ugg Boots Insoles</strong></a> on sale ,many of which are hand made by craftspeople.RealUGG,Inc came up with some new-styles-<a href="http://www.realugg.com/ugg-lo-pro-button-c-57.html" rel="nofollow"><strong> Ugg Lo Pro Button , </strong></a> <a href="http://www.realugg.com/ugg-womens-rainier-c-53.html" rel="nofollow"><strong>Ugg Women's Rainier</strong></a>, We offer a wide range of UGG Boots in different color and styles. Browse the website, we are sure you will find <a href="http://www.realugg.com/" rel="nofollow"><strong>Real UGG</strong></a> to tempt you! </p> Frozen winter is coming,do you like to buy ugg to preventing cold and keeping warm.As a pursuer for the most stylish ,is there any ugg boots you have in mind?I think you have no excuse anymore if you don't have a pair of UGG Bailey Button Boots.The Bailey Button UGG can be worn as a UGG classic short Boots or UGG classic mini Boots . Hottest pop style.If you like the UGG tall boots,www.realugg.com can provide high quality of Ugg Tall Boots in low price,Are you straitened for Ugg Boots Insoles,don't worry ,we have the best Ugg Boots Insoles on sale ,many of which are hand made by craftspeople.RealUGG,Inc came up with some new-styles- Ugg Lo Pro Button , Ugg Women's Rainier, We offer a wide range of UGG Boots in different color and styles. Browse the website, we are sure you will find Real UGG to tempt you!

]]> By: Barry Chu http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2184 Barry Chu Tue, 22 Sep 2009 16:40:31 +0000 http://andrewchenblog.com/?p=1243#comment-2184 Hey Andrew - totally agree, as you no doubt remember from working in behavioral targeting, it's become readily apparent that quality of inventory trumps all targeting. The best targeting "algorithms" are the ones that can transfer value from a known high-value entity (such as site with high purchase intent) by identifying those users as they move from site to site. Even then, the value degrades as you shift the context of the user. I would caution anyone looking at investing in or joining an ad network to inquire about their quality and quantity of unique inventory first before worrying about their technology. Hey Andrew – totally agree, as you no doubt remember from working in behavioral targeting, it's become readily apparent that quality of inventory trumps all targeting. The best targeting “algorithms” are the ones that can transfer value from a known high-value entity (such as site with high purchase intent) by identifying those users as they move from site to site. Even then, the value degrades as you shift the context of the user. I would caution anyone looking at investing in or joining an ad network to inquire about their quality and quantity of unique inventory first before worrying about their technology.

]]> By: Andrew Chen http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2183 Andrew Chen Tue, 22 Sep 2009 16:22:04 +0000 http://andrewchenblog.com/?p=1243#comment-2183 There are about 300 ad networks, and people know it's a massive $500B<br>industry (offline+online). So there's a lot of very smart PhDs and geeks<br>chasing it, probably thousands of talented people total - I'd argue the ad<br>network industry is far more efficient than Netflix, no matter how smart<br>they are :-) There are about 300 ad networks, and people know it's a massive $500B
industry (offline+online). So there's a lot of very smart PhDs and geeks
chasing it, probably thousands of talented people total – I'd argue the ad
network industry is far more efficient than Netflix, no matter how smart
they are :-)

]]> By: Andrew Chen http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2182 Andrew Chen Tue, 22 Sep 2009 16:21:01 +0000 http://andrewchenblog.com/?p=1243#comment-2182 There are about 300 ad networks, and people know it's a massive $500B<br>industry (offline+online). So there's a lot of very smart PhDs and geeks<br>chasing it, probably thousands of talented people total - I'd argue the ad<br>network industry is far more efficient than Netflix, no matter how smart<br>they are :-) There are about 300 ad networks, and people know it's a massive $500B
industry (offline+online). So there's a lot of very smart PhDs and geeks
chasing it, probably thousands of talented people total – I'd argue the ad
network industry is far more efficient than Netflix, no matter how smart
they are :-)

]]> By: Andrew Chen http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2181 Andrew Chen Tue, 22 Sep 2009 16:20:06 +0000 http://andrewchenblog.com/?p=1243#comment-2181 There are about 300 ad networks, and people know it's a massive $500B<br>industry (offline+online). So there's a lot of very smart PhDs and geeks<br>chasing it, probably thousands of talented people total - I'd argue the ad<br>network industry is far more efficient than Netflix, no matter how smart<br>they are :-) There are about 300 ad networks, and people know it's a massive $500B
industry (offline+online). So there's a lot of very smart PhDs and geeks
chasing it, probably thousands of talented people total – I'd argue the ad
network industry is far more efficient than Netflix, no matter how smart
they are :-)

]]> By: mattmaroon http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2167 mattmaroon Tue, 22 Sep 2009 00:05:36 +0000 http://andrewchenblog.com/?p=1243#comment-2167 This is a touch disingenuous. BellKor didn't get a 10.5% increase over random recommendations, or ones made by me (especially since my recommendation to anyone who liked Michael Bay movies would be to off themselves as painfully as possible). They got a 10.5% increase over a recommendation engine built by highly-paid and probably highly-talented Netflix engineers. I'm guessing that Netflix's Cinematch was so state-of-the-art that a 10.5% improvement over it is quite impressive. I'd be surprised if any ad-networks, which generally aren't as engineering-focused as NFLX, are starting from such a baseline. This is a touch disingenuous. BellKor didn't get a 10.5% increase over random recommendations, or ones made by me (especially since my recommendation to anyone who liked Michael Bay movies would be to off themselves as painfully as possible). They got a 10.5% increase over a recommendation engine built by highly-paid and probably highly-talented Netflix engineers. I'm guessing that Netflix's Cinematch was so state-of-the-art that a 10.5% improvement over it is quite impressive. I'd be surprised if any ad-networks, which generally aren't as engineering-focused as NFLX, are starting from such a baseline.

]]> By: jamesshamenski http://andrewchenblog.com/2009/09/21/whenever-ad-networks-talk-about-their-targeting-remember-the-netflix-prize/comment-page-1/#comment-2164 jamesshamenski Mon, 21 Sep 2009 18:45:37 +0000 http://andrewchenblog.com/?p=1243#comment-2164 Hi Andrew,<br>I agree that disruption is bound to happen around inventory and data. <br><br>One more point I'd add is around online branding campaigns. Today, most ads online are valued on conversion. The other side of advertising is around branding campaigns. With the right approach, those branding dollars being spent in Magazines and TV will come online. This represents the largest portion of the advertising market which is yet to be disrupted. Hi Andrew,
I agree that disruption is bound to happen around inventory and data.

One more point I'd add is around online branding campaigns. Today, most ads online are valued on conversion. The other side of advertising is around branding campaigns. With the right approach, those branding dollars being spent in Magazines and TV will come online. This represents the largest portion of the advertising market which is yet to be disrupted.

]]>