Is AI friend or foe to audience creation?

I’m Alan Chapell. I’ve been working at the intersection of privacy, competition, advertising and music for decades and I’m now a regulatory analyst writing for The Monopoly Report.

Our latest Monopoly Report podcast is out with legal analyst Megan Gray. Megan and I talk about remedies in the DOJ search trial taking place against Google - to help everyone get up to speed in advance of Judge Mehta’s ruling in August.

A way too brief history of audiences in the ads space

Prologue: I first put my shingle out as a privacy attorney in late 2003. To say it took me a while to build a book of business would be putting things kindly. It literally took years! For the first year or so, I spent a good deal of time writing for industry trade pubs - as I didn’t really have much else going on back in those days.

In 2004, I wrote an article for an outfit called iMedia Connection on this new-ish concept called behavioral targeting that seemed to be taking up a good deal of oxygen at the time.

It’s been just over two decades of audience creation here in ad land. Looking back, I view the audience space as having a number of defined eras. I recognize that I’m over simplifying - and that this entire exercise can come off as a bit subjective. (Almost as thorny as rating music.)

Audience Creation 1.0 - The Tacoda / Revenue Science years

When I was at DoubleClick in 2001, there was product called Intelligent Targeting. Nobody really understood how it worked. And while there was certainly interest in the product, it wasn’t exactly flying off the shelves. But when Dave Morgan and Omar Tawakol began evangelizing the concept of behavioral targeting, it really started to take off.

In this first era, audience companies were mostly pulling context off a single web page visit and using that to target an ad a few clicks away. The technique kinda worked… but it also required that one overlook the IP concerns.

Audience Creation 2.0 - The LiveRamp / BlueKai / Demdex years

Eventually, there was growing sentiment around the idea that taking audiences from publishers without compensation was considered impolite. There were three separate innovations that ushered in Audience era 2.0:

1. LiveRamp enable advertisers to onboard their CRM files.

2. BlueKai created a marketplace that attempted to place a commercial value on audiences.

3. Demdex introduced the concept that an advertiser (and later, a publisher) could manage their own audiences in a way that maintained a level of interoperability with the rest of the ads ecosystem (also, we used the term ecosystem a LOT back then).

Era 2.0 did not replace the techniques from Era 1.0 entirely. But over time, the above three tools became a larger and more significant part of the audience creation mix.

Audience Creation 3.0 – The data exhaust years

Some may argue that this era is the precursor to what’s coming with AI. But it’s worth noting that Audience 3.0 traces its roots back to the early days of social media. Companies like Friendster and MySpace had shifted the way Internet Users were consuming digital and that necessitated a shift in the very definition of the term audience.

Companies like Media6Degrees (now Dstillery) and 33Across figured out that patterns of digital media consumption could be indicative of buying behavior – even if you didn’t know or care about the actual pages or digital media properties visited.

These techniques didn’t replace what came before. Rather, they were additive. And respectfully, I believe that a good deal of what’s being held out as AI targeting today is closer to a variant of the 3.0 Audience techniques.

Audience Creation 4.0 – The AI years

Some will argue that we’re squarely in this era already. As noted above, I’m not so sure. If you’re doing interesting stuff in this space and are inclined to share, please reach out to me.

Meanwhile, a few thoughts come to mind:

1. Does AI represent the next wild west? - My sense is that there is a fair amount of the business community that believes that AI’s injection of secret sauce into audience creation will usher in a new era of doing WTF we want. The idea behind that sentiment – exploit the grey areas - has existed within ad tech culture for a long time. After all, if nobody understands the data processing decisions that underpin AI audience creation, we've got a few years before privacy laws and regulators catch up, amiright?

   Alan’s View – I’m not sure how to break this to you, but regulators are much more sophisticated than many around here want to believe. Perhaps more importantly, inferences attached to personal data like a UID are still personal data in the minds of regulators. Accordingly, if you can’t explain how it works, but it just works… well, that is a bug… not a feature.

   Using all that AI firepower simply to find new ways to ascertain that some teenager in the mid-west is pregnant before her parents find out doesn't necessarily seem like a step in the right direction.

2. Relevant AND Privacy safe? - There are a few in the ads space who believe (or at least hope) that AI infused contextual solutions are (or will be) as good as interest based or other targeted ads. And the hope is that AI will eventually be able to tell you enough about a particular URL to get to a reasonable level of relevancy. In other words ,some in this space believe that AI will (somewhat ironically) be a tool to address the privacy issues in the ads space.

   Alan’s View – AI works better for brand safety than audience creation as of today - but that will change rapidly. For now - and for at least the near term - we will continue need more information than what you might get from the URL to drive relevance that hopes to compete with big tech, such as: (i) information about other Users who have visited that URL, (ii) information about the previous URL visited, and/or (iii) information about that particular User.

   Some of the above is difficult to do effectively without a UID somewhere in the mix - which probably isn’t great news for the ID less crowd…. or for those who are hoping AI will solve privacy in the near term.

3. AI is Irrelevant as it isn’t fixing the underlying issues - What is the ROI to optimizing a system that we all know is pretty broken? Looking more broadly, its great that you can (to varying degrees) create campaigns, build creatives, protect brands and create relevancy with minimal human involvement. But those “improvements” are meaningless because they don’t address enough of the big problems. This is certainly a more cynical view of the impact of AI.

   Alan’s View - I agree that we need to fix what's broken about the ads space, and optimizing via AI without a huge reimagining or reinvention will only exacerbate the underlying problems. I’d love to be proven wrong.

__________________________________________________________________________

If there’s an area that you want to see covered on these pages, if you agree/disagree with something I’ve written, if you want tell me you dig my music, or if you just want to yell at me, please reach out to me on LinkedIn or in the comments below.