1Bud Caddell

Noah,

Total agreement. I actually just pushed out a deck about digital media not being mass media for cheap. I feel like there's still a fundamental disconnect about the purpose and the power of the web as an altogether different beast.

Check it out if you want, http://www.slideshare.net/bud_caddell/digital-media-isnt-mass-media-for-cheap

March 10, 2009

2Matt Daniels

Love the idea to hyper-target--but what's the incremental cost for versioning the banner ad for all of those blogs and sites? Perhaps $2K a pop?

March 11, 2009

3Steven Kalifowitz

Completely agree. I was just talking with a friend about how it's no wonder NYTimes can't turn a profit. I've had an account on their site for about a decade... they still don't place contextual ads - neither to the article content, nor the viewer (me).

I just visited the NYTimes for kicks while writing this, and randomly clicked on this article "Preparing a Fresh Batch of Chefs" in the Television section; there were 5 ads:
3 for NYTimes offers
1 for the Paul Taylor Dance Co
1 for "Our Town" (I'm assuming this is a play)

WTF!?!? Is it so hard to place an ad for a cooking show?!?!? a cooking school, something related to the content...

And if contextualizing ads to the content is too hard, at least know that I wouldn't care about theater or dance, as I've never visited either of those sections of the NYTimes...

And Matt, this isn't about making a unique ad for each blog ($2k is high, but not far off), it's about serving up the right ad for the right viewer... in as many venues as possible.

March 11, 2009

4Steven Kalifowitz

Sorry - want to adjust my last comment to Matt. I may be wrong, it might be about making many versions of an ad. But that cost is off-set by the higher quality audience. You don't need to place as many ads if you're going after a smaller niche (fewer people to make impressions on) where each impression is way more valuable. I believe Noah has written before about how the whole system of measuring KPI's - and even determining what those are needs to radically change. But doing that would put many people out of work (who are currently in control) - so that's not happening soon.

March 11, 2009

5Joshua

Hallelujah! I've been bantering on about this for years....display ads completely suck and have NEVER worked. Existing online display formats are nothing more than a lazy attempt at monetising the internet. Wacking 'print model' ads onto the internet was doomed to fail. We've reached a stage where we think 0.2% CTR is good - in fact almost twice as good as average. Whilst this is an incredibly sad state of affairs - it does lend itself favorably to innovation. As the pressure continues to mount from above and budgets are squeezed, there will be a renewed vigour for accountability, performance and ROI. Let's face it - online is the only channel with more than a modicum of accountability - it's about time we used it. New online display models are in the works that are fit for purpose. Once usch format is the Open IMU. Among other marvellous things, it's an ad fomrat that knows where it is and serves contextual to the placement and the user preference. There are a few other tricks to it - but the results are what counts. CTR rates on average of 10%+...this is just the beginning.

March 11, 2009

6Noah Brier

@Matt: Yeah, I don't know exactly the answer, though I suspect part of it is that you put your ads on fewer sites.

@Steven: While I generally agree (they should know someone about you and be able to roughly target), I don't know where I fall in the targeting camp (specifically, I don't believe that we'll have a day where it's one-to-one). With that said, I do think the NYTimes had one of the better display ads I've seen recently: regular state & rollover.

@Joshua: Thanks and never heard of "Open IMU" will check it out ...

March 11, 2009

7Seni Thomas

This is a comment I posted in response to this:
http://www.darrenherman.com/2009/03/08/white-paper-disintermediation-of-online-display-advertising/

But, it's relevant here as well so my 2 cents.

"Good stuff man.

There are a few points worth expanding on, but I wanted to focus on the false sense of effectiveness that 'targeting data' sometimes affords.

Quite simply, if I went to network X and paid $15 CPMs for 5 targeted segments, but targeted them all with the same creative (non-unique message) there is really no reason why I shouldn't have just paid $3 CPMs for Run of Network as it would have included the "targeted segments", which is much more cost effective and there is still usually a layer of optimization that occurs at the RON level. This is usually the case when it comes to digital media plans and is where the false sense of effectiveness creeps in for the following reasons:

1. Everyone is addicted the notion of targeting
2. As a planner you need to prove performance and the CTR might be 0.7% in the targeted segments because the number of impressions served was lower, and .09% for the RON campaign even if the same people clicked on the ads. With RON you just get a ton of extra impressions that could be driving awareness, recall, etc.
3. The CTR metric can be a tricky beast, as can all data, and can lead to many inaccurate conclusions

Thus, I believe that the next generation of data based targeting must reside within a larger framework, where it is only a component of the solution, not the end game.

A + B || C

( || means piped into... Geek reference back to Unix/Linux)

Put another way,

Targeting + Dynamic Creative || Dynamic Web Experience

Note: I prefer the term Targeting as data is useless without an effective algorithm making sense of it

This framework accounts for the full user experience with the brand and allows each component to self optimize, while simultaneously optimizing at an aggregate level . For example you can feed various targeting segments across your purchased inventory, feed hundreds of different iterations of creative to each segment through a dynamic unit, and optimize based on the clicks, engagement, etc.

At this point you are creating a number of optimized pairs say A3 (for Targeting algorithm #3) and B4 (Creative iteration #4).

These optimized pairs when piped into the dynamic web experience create customized sites for specific types of people that clicked on a particular ad.

For example, if I'm an "In-Market TV Buyer", you can deliver a number of banners that depict different TV models, features, calls to action, etc. so that when activate a given banner the landing page highlights the information displayed on the banner I clicked.

It all sounds costly and time consuming, but many companies out there today provide these types of solutions for reasonable rates, and many times it takes less time than actually building traditional creative, as you only need to use a template and define the variables. For example, Adisn runs a dynamic solution for Buy.com that pipes in 150,000+ images and info from Buy.com's database. All it took was a template and an XML feed.

This is still a work in progress as no one has figured it all out quite yet; however, I do advocate that we all take a step back and look at this problem from a higher level.

Cheers all,

Seni"

March 11, 2009