• http://www.rimmkaufman.com George Michie

    Really interesting post, Frost. Might this be one argument for why Google can/eventually will make DSPs obsolete? Google’s forte is creating order out of unstructured data. Giving advertisers the kinds of controls offered by DSPs (RTB, dynamic creative, etc) would not only cut a tremendous amount of cost out of media buying, it also seems like they could do this better than anyone. There is the trust factor, and the fact that Google doesn’t control all of the inventory…yet, but…Love to get your thoughts on that.

  • http://www.linkedin.com/in/larrykaul Larry Kaul

    Paid search marketers don’t treat each data item differently. Bids which can be the most critical factor to determine financial performance are set using average value methods or manual clustering techniques by most bid management software solutions. The best way to set bids is to treat each keyword at its most granular level as a unique entity. For complex programs the only way to do this is automatically using software. @optimineinc has cracked this code. As described decision making through tracking at a granular level is the key to marketing success.

  • http://www.simpli.fi Frost Prioleau

    George- IMO, any DSP that does not enable their advertisers to create order out of unstructured data will be at risk of being obsolete. While Google has fantastic engineers that are capable of lots of things, because of their size they may be slower to implement this capability than small compaines like Simpli.fi, where we are doing element level targeting today. The other key is implementing algorithms that can make decisions on the combinations of the most impactful data elements in the time allowed in RTB, which is often less than 30ms. Doing this at scale is another hurdle that will differentiate DSPs going forward.

  • http://www.simpli.fi Frost Prioleau

    Larry- It sounds like you are seeing the same thing in paid search that we see in display….that bids should be set and optimized at the keyword level.