Recency and frequency are factors in search but not in the same manner as display. This distinction, along with the larger role of context are three areas I believe can help improve display. The main differences with Search are twofold. 1) they are being calculated not only at the user level but in aggregate and 2) their analytic foundation is not the impression but actions (queries & CTR), price & (for Google) match type. Looking forward to more of your thoughts on the mess.

Charlene: I love your perspective. It's sparking quite a few thoughts, especially where semantic web comes into play. Not to mention having done many tests in "personalization" I have strong feelings as to its efficacy. Scale: I just don't think Social scales on a personal data level. We need thicker slices to sell, and make targeting possible to optimize. We simply can not optimize based on a single cookie or IP with so many segments associated to it. Well, maybe Google can but most of us can't crunch that much data. Discovery: I think content and social are already intertwined. We just have not been able to pull them together effectively with old technology. That failed technology (BT) was already based on the visitor. So was the "one-to-one" personalization technology like epiphany back in 1.0 that never lived up to the promise. Semantic technology of course will solve this targeting problem - but semantic has to be about the content. Data: I think the data is manageable now - at scale. :) Google has proven that content rules irrespective of personal data is enough to deliver high degrees of ad targeting relevance. More than enough for most marketers to be very successful. Regardless, I think personal and social context is often defined in content -- semantic web is perfect to bring these "content correlations" together, or "socialize" them. Last, we often express our affinity in the content we create, share and view. Companies like Glue, Inform, Dapper and others are at the start of something big and its will be gloriously mashedup and indeed social. We always must start with the assumption that the content the visitor is on is relevant to them. Targeting to this content the only way to provide a win for the person, the publisher and the advertiser. Not just the person. Happy holidays! Looking forward to your thoughts on semantic web when you get around to posting them.

Hi Charlene, Enjoyed your list. I love the idea of understanding demand on a social level. That's huge and untapped. However, I have a differing prediction on targeting behavior and influence based on issues with scale, user discovery and data. Scale: Display is all about scale, as it should be. It just doesn’t cost that much to buy 50 million impressions. Micro-targeting, with the exception of geo-targeting is generally not a successful segmentation strategy in display. Discovery: The influence exerted by a third party is not as important or measurable as the first party interaction with the content. Simply, my opinion is always the most influential to my performing an action. The implicit data might make me aim better, but it’s not the best arrow to shoot. My money will always be on content targeting to outperform behavior and influence. Data: Social interaction data might be interesting and there is lots of it. However unless better filters emerge it will be littered with so much noise as to make the data worthless. Unless of course you are micro-targeting. ☺ We still have so much headroom in 2009 to improve marketing to a person’s attention. Social recommendations have never proven to be more valuable and are certainly less numerous than personal discovery. There is certainly an opportunity here but it is probably coming from the semantic technology, based on temporal metrics and in 2010. That’s my prediction! Best wishes to you and Altimeter in 2009!