Easy as 1, 2, 3

So, the people I work with are probably sick of hearing me talk about the three levels of display optimization. For posterity, I’ll write them down here.

True optimization of display advertising spend must address three different, but related questions. I have dubbed them the three levels of display optimization:

Level 1. Bid Optimization – how much is this impression worth? (Equivalenty, how much should I bid?)
Level 2. Creative Optimization – now that I’ve purchased this impression, which creative variation will yield the most value?
Level 3. Landing Page Optimization – now that someone has clicked, what is the most effective sequence of pages/images/offers/whatever to seal the deal and generate revenue?

For reasons I’ve already gone through in a previous post, I believe levels 1 and 2 need to be part of the same operational stack.

Ideally, level 3 would also be part of the same stack, but I can see why advertisers would be loath to give up control over their landing pages to a third party. (Though it does happen occasionally.) My guess is that level 3 will most likely be loosely integrated, rather than optimized from within the same algorithm. This integration will probably occur by passing the “creative signature” – background color 6, with message 8, image 4, and call-to-action 2 – on the click URL. While the advertiser (and his or her landing page optimization algorithm) would almost certainly prefer the passing of this as well as the underlying targeting data, the passing of the latter might be complicated by the accompanying privacy-related conniptions of the FTC. (And don’t forget that conversion information must be passed back to both level 1 and 2, or an integrated level 1/2, to close the feedback loop(s).)

It’s interesting to note that most of the DSP-related buzz has centered around the denizens of Level 1, both the independents and the feverish efforts of the various agency holding companies to roll their own. What isn’t as well documented, though they’re starting to talk about it, are that the current residents of Level 2 and 3 are also starting to make noise on their own.

Level 2 Creative Optimizers include startups like Teracent, Tumri, Dapper, and a host of others, as well as the incumbent rich media players like PointRoll, Eyeblaster, etc, who have added (or are adding) optimization capabilities to their impressive frontend razzle-dazzle.

Level 3 LPO’s include a bunch of companies that I haven’t paid as much attention to, though I’ve been told they’re fairly numerous, especially if you count the site-side analytics guys. I think both Adchemy and Aggregate Knowledge started out here. (Someone confirm?)

Anyway, if you look closely at the good folks in Level 2 and 3, you’ll see that many of these companies are already well on their way toward moving into adjacent slots. Start reading the recent press for many of the companies named above, and you’ll see what I mean.

So who’s going to get the full stack first? Well, barring acquisition, I suppose the right question to ask is: which is hardest to build? Personally, I think it’s Level 2 and 3. (And, btw, have you noticed how similar Level 2 and 3 become if you think about landing pages as just another sequence of full-screen ads?)

15 thoughts on “Easy as 1, 2, 3

  1. First, great thoughts and a blog that’s now on my radar screen.

    Second, I think that the majority of the advertisers will not give up the full stack if it includes the landing page – though for the ones that do, will see tremendous return.

    The reason for this is because the process of media optimization, creative optimization, and landing page optimization generally is split across multiple parties. For big brand advertisers with many agencies, it’s going to be hard to get everyone to give up control unless the brand really pushes for this and that’s going to take a while.

  2. Peter – you could have said that this article was about Search media and nothing would have been different about it.

    Looking at Search it’s easy to echo Darren’s ascertation about landing pages and the complexities of service channel conflict.

    So what happens? Bid management gets commoditized, ad creative get little attention and landing pages are all but forgotten. All the budget goes to media.

    Smart marketers will always do well in either channel but they are few. In the end nothing changes here. We need new paradigms.

  3. For what it’s worth, I agree with Darren’s assertion. I say as much – or at least imply it – in the post.

    Actually I think that a big difference between display and search will be the creative.

    What’s the character limit on search creative? 125 characters? I like words as much as the next guy, but there’s only so much you can do with less than a tweet. I’m sure that others will object, but my gut is that there’s just not that much to optimize.

    Compare that to the sheer permutation loads that can be found in display advertising – particularly when you bring both rich media and dynamic creative generation into the mix. It’s no cliche to say the possibilities are endless.

  4. BT doesn’t seem to be going anywhere fast does it… where do you see this being important in your stack? All of it of course! Level 1 is pretty obvious – the better targeted the more its worth, but how do you adapt the creative/conversion?

  5. A couple of follow on thoughts to your comment (and thanks for getting this dialog started)

    The creative aspect of search is incredibly undervalued. Your gut is similar to many others but search creative has huge influence. There is plenty to optimize on both conversion & CTR which influences Quality Score and ultimately bids and ROI. I’ve done a large number of multivariate tests on search creative and 1) it has a huge impact on conversion 2) that influence can vary by different elements – title, d1, d3, url 3) performance can vary greatly. Here’s one case study I presented at SES three years ago: http://bit.ly/2VFjqw

    The endless possibilities of dynamic display are actually not so great for testing. The optimization has to been done on click through rates that are really low are rarely factor in network/publisher data (where the ad was running, what position on the page, what time of day, etc) thus bringing lots of noise into the result sets. This is not to say I’m not a fan of dynamic display – i’ve worked quite a bit in that area – but it’s not a panacea. You can get amazing results from a 3×2 MVT on a 300 x 250 gifs but how many people even do that? Hardly anyone.

    One last point to keep in mind — and this is just a function of cross testing and based on lots of personal experience. The ad generally is the most influential factor in conversion rates no matter what you are testing in the stack however the bid and the landing page ultimately control the overall performance results. Ain’t that a trip? Happy optimizations!

  6. @Chico:
    BT, along with all other types of targeting data, is critical in every level of the stack. The typical way to use it is through automated multivariate optimization. Essentially, letting an algorithm figure out which types of data are predictive of performance when faced with thousands of creative permutations and even more discrete targeting buckets. Drop me a line, and I can explain this in more detail. :)

    @JMendez:
    I’ll take a look at the case study, thanks!

    “The endless possibilities of dynamic display are actually not so great for testing. The optimization has to been done on click through rates that are really low are rarely factor in network/publisher data (where the ad was running, what position on the page, what time of day, etc) thus bringing lots of noise into the result sets.”

    Fortunately, there are other options than straight CTR optimization. While CTR does factor in, most of the clients I’ve seen prefer to optimize on some hybrid of (rich media) interaction rates and/or directly on conversion. (Conversion events can be so statistically sparse, that practically speaking you do need other measures of performance as proxies.)

    Endless possibilities of display do provide a challenge, but the right way to address is to combine dynamic creative with automated optimization algorithms that relieve much of the quantitative and analytical burden of the otherwise daunting permutation counts.

    “This is not to say I’m not a fan of dynamic display – i’ve worked quite a bit in that area – but it’s not a panacea. You can get amazing results from a 3×2 MVT on a 300 x 250 gifs but how many people even do that? Hardly anyone.”

    I’m a fan of dynamic display too – particularly when it’s paired with automated optimization algorithms to address some of the very issues you highlight. In fact, that’s my day job – working with these types of technologies is pretty much all I do. :) And you might be surprised by the strides in adoption made by these technologies in the last 6 months, and not just with 3×2 MVTs – but with far more complex setups that often entail literally millions of permutations. (Again, only with automation can complexity like this be realistically handled, but it’s just beautiful when applied correctly.)

    Great comments all, thanks!

  7. Hi there I am so grateful I found your blog page, I really
    found you by error, while I was searching on Aol for something else, Anyhow I am here now and would just like to say thank you for a incredible post
    and a all round entertaining blog (I also love the theme/design), I don’t have time
    to read it all at the moment but I have saved it and also added in your RSS feeds, so when I have time I will be back to
    read more, Please do keep up the fantastic jo.

    My webpage – http://www.161989up.com

  8. This design is steller! You definitely kknow how to keep
    a reader amused. Between your wit and your videos, I was almost moved too start mmy own blog
    (well, almost…HaHa!) Great job. I really loved what you had to say, and more than that, how you presented it.
    Too cool!

  9. Automobiles be doctors, engineers, chemists, architects,
    teachers, etc. Give the banks attorney the file number and the judge’s discover.

    There are often attorneys who usually be on the pry for the news of an lock up.

  10. Usually I do not learn article on blogs, but I would
    like to say that this write-up very compelled me to try and do it!
    Your writing style has been surprised me. Thank you,
    quite great post.

Leave a Reply

Your email address will not be published. Required fields are marked *