Nothing is certain except death and taxes — and the annual frustration over how inefficient digital advertising actually is. In this edition of Digiday’s Confessions series, in which we exchange anonymity for candor, it’s the turn of an ad exec who has a more sobering take than usual on the matter: Most activation in the ecosystem is inefficient and profits from that inefficiency, they said.
The exec detailed exactly how (and why) this happens.
This conversation has been edited and condensed for clarity.
How entrenched are these inefficiencies into the way online advertising is done?
OK, so normally an advertiser would start with a list of people they want to reach based on a customer list or even one they’ve acquired from another source. To reach those people online, the advertiser has to put the audience list through a whole bunch of filters and transformations. First, the list has to go through an onboarder after which the advertiser will typically lose 50% of that list. Then the onboarder has to move what’s left to the delivery platform. And that’s where another 50% of entries on the list is usually lost, whether that’s a result of cookies or some other targeting mechanism. Which is to say the advertiser probably doesn’t have the scale they’re looking for anymore.
So the better the original audience list is to begin with the more this process hurts?
Simply put, yes it is. But the good news is that if your original audience list is bad to begin with then the dilution of it doesn’t hurt you as much. That said, it also means the baseline quality of that audience is pretty low to begin with. You’re never improving the audience as you go through this process.
Why do marketers continue to do this?
What’s the alternative? That’s the sad thing about all of this.
Why don’t the platforms help?
There is not a single platform that will tell you when you push an audience to them, which of that audience they can actually reach. Let’s say you push an audience of a million to either one of the larger platforms. They will tell you that they can match 500,000 of that amount. But they won’t tell you which 500,000 that is. What this means is that marketers have no way of knowing whether they’re able to reach that original million. They have to push it to five different platforms and hope that between them that they cover the whole thing.
It’s the walled garden model in effect: create silos of spend that prevents dollars from moving too quickly or rapidly between platforms.
Exactly. And it’s not just the typical walled garden platforms that do this. Even the regular DSPs employ these tactics. They’d rather have a chance to reach all of them. A DSP wants to get the whole audience list because it gives you more chance to pry more budget out of the advertiser.
Can’t advertisers insist to platforms and ad tech vendors that they get the feedback loop in exchange for the original audience?
They can do — and some have tried, but it never works in my experience.
Is that because qualitative transparency is hard to come by in online advertising?
You could say that. There’s a great deal of transparency on media costs and the sites on where the media is delivered — well sort of. But there’s much less transparency on the performance and execution of the advertising.
And even once the campaign has been done, the walled gardens mark their own homework.
Same as ever. Those walled gardens don’t provide any feedback. Sure, they provide aggregate numbers but they don’t share any log level data that will let the marketer identify any redundancy.
Then again, even if the platforms did provide that log level data there’d be a challenge in figuring out how they reconcile those disparate data sets.
True. Getting that data is just the start. The advertiser still wouldn’t know if it’s the same person they’re reaching across the platforms because there’s no common identity space.
Is solving this issue the impossible job in online advertising?
There’s this notion that they can’t because it’s in some way a violation of the user to tell the advertiser that they received an impression. That’s something I’ve never been able to get my head around. If a media owner is concerned about cross-site tracking then provide the log level data back without the site information. That would address the issue.
This sounds grim. Are there are any signs of encouragement?
It’s certainly frustrating. These platforms, whether they’re walled gardens or ad tech, have to say the right things to us behind closed doors because we’re representing the client. I always measure the intent to change the situation on whether these companies actually try and do something. So far, we’ve seen a range of responses. At least one platform we’re talking to currently is on track to provide the feedback loop we talked about earlier. This is certainly not a trend. It could become one though.
Really, we’re asking for something as straightforward as this: we send a platform over a million IDs in a file, the platform sends us back that same file with a yes or no next to each one so we know whether we can target those audiences there. It gets to a point where it becomes hard to duct too much. Granted, client pressure would make that even harder.