The little dirty secret about RTB and Programmatic

My first reactions on the whole RTB and Programmatic craze were of plain skepticism. You know, being working on AI and Business Intelligence since forever, I have had my share of overhype.  Watching everyone on the advertising industry going bonkers over it didn't help to overcome my grumpy attitude.

I must admit, the overall issue started getting interesting when I got inside the belly of the beast and started noticing that hey, it's not really so bad. To be honest, it's just another manifestation of the pervasive and crappy "algorithm as be-all and end-all disruptor on a concrete market" narrative that I will blow to smithereens another day. The idea of what humans can do could be replaced with an algorithm is very, very optimistic. To be honest, it took us 13 years from being able to win at chess to being able to win at Go. Do not stop the press yet. 

Well, back to our business. Problem is, RTB is making headlines and progresses, but it's not exactly setting the world on fire. Well, in fact, if you dig a little bit deeper and check wonderful in deep analysis of real campaign results as Thomas Baekdal do, you can find real horror stories under the glitz of Programmatic:

One thing that we will see over and over again, this year, are examples of advertising campaigns that just don't work. Let me just show you two that came to my attention within the past month.

The first one is from Fisku, a programmatic advertising agency, who did a study trying to prove that their ads are better than other ad solutions. What they found, though, was this:

The thing you need to look for here is the cost versus the income, and we very quickly see that there is a problem. The cost per impression (CPI) is higher than the average revenue per user (ARPU). In other words, these ads produced negative ROI.

Fisku then tries to muddle their way out of this by adding a note saying: "ARPU (Average Revenue Per User) being less than CPI (Cost Per Impression) is not indicative that overall results led to negative ROI."

I do agree that the actual impact on ROI shouldn't be calculated based on a single metric, but it doesn't change that, according to the number that we have, Fisku's campaigns produced negative results.

Talk about shooting yourself in the foot.

Also note the cost per purchasing user (CPPU). It's insanely high. The cost of attracting just a single person using display ads is almost $300... of which the average person only generates $0.15 in revenue.

That's insane.

And the same is true for audience network ads and News Feed ads. The costs might be lower ($43 for News Feed), but revenue is still terribly low ($1.63).

Of course, Fiksu, still not wanting to admit defeat tries to spin this terrible result by saying:"Increased revenue per purchasing user: Facebook's Audience Networks' revenue per user was 20x greater than on display and more than twice that of the Facebook News Feed. This is a clear indication of Facebook's Audience Network's ability to attract loyal, high-quality users."

And true enough. Revenue was 20x greater than with display, but the cost was also 26 times higher than the revenue. For every $1 earned, they lost $26.

It doesn't matter how much you spin this. It is a terrible result. And they are far from alone in this.

 

Uh-oh.

Baekdal then rambles on with something we already know: The current advertising scene is just about to crumble. I strongly recommend you to read the rest of his great work, but I'm going to extract some tidbits that encapsule the whole problem:

These two examples are just a few of the many similar examples we will see over the next couple of years, and every time they will serve as a very real, very scary wake-up call, forcing you out of the dream-like world where advertising just magically works.

But does this mean that advertising doesn't work? No, of course not. But it does mean that advertising has lost part of its effectiveness due to both the abundance of everything and the sheer amount of it that people see every day.

We have to realize that only some forms of advertising will work, and only for some type of activity or products. The old idea that advertising works for everything is no longer true, and probably hasn't been true for a long time.

Ok, those are harsh but very needed truths laid on the table. Baekdal then starts dwelling on how you really need a proper strategy in place and knowing inside out your clients and their customer journeys if you really want your ads to be effective. No longer you can throw your shitty briefing on the table and let the advertising boys do their magic: A lot of work, and specifically, customer-centric strategic work, must be performed before you search help from the guys of Madison Avenue.

And sorry to inform you, RTB and programmatic won't save the day. Why? Beacuse they have an ugly, dirty little secret hidden under their glitzy appearance.

The DMP, the core piece that would allow Programmatic to be able to be "customer centric along the funnel",  as the above capture from a Mediacom presentation shows, the container of the customer information needed in order to make the ad relevant, is usually... empty.

As Xavier Rius, a friend, states on this Linkedin conversation (Translated from above capture):

Hey everyone. We are in 3.0 since a year ago, but getting quality data is very complicated if we do not coordinate ourselves all. We need to exercise teamwork. The current existing data is ridiculous, and if we test it it does not work. DMPs are wonderful tools, but they are empty, gentlemen.

Hitting the nail with the proverbial two-ton hammer.

The key clog of the whole Programmatic/RTB engine is having the proper data on the DMP (Data Management Platform), a key component of the Programmatic architecture. And the needed data, plain and simply, it's not there. And sorry to throw salt to the gushing wound, Xavier, but no amount of coordination will bring the kind of data you need in order to make this work. 

In order to obtain the data you need, you need to perform tasks which are out of your current league, in terms of both scope and reach, which prevents the whole RTB architecture from delivering the magical, elusive, "perfect ad"

Reading this aloud myself just makes me feel like Darth Vader stating the obvious regarding the Death Star. There is a much, much bigger force underneath that dwarfs the power of your "technological terror"

And this force is called Personetics. Which is, among other things, the science of identifying and obtaining (usually non present) purposeful data.

It's a real pity the fact that people from Mediacom and the rest of the advertising industry have no clue about it, and what makes it work. As this wonderful slide from Mary Meeker's slide deck proves year after year:

I can predict the gap being same for 2015 (Hey, I'm no genius, it has done nothing but grow since 2011, first year I have of the slide deck). Because the advertising industry has no clue on how to be relevant on a hyperpersonal space.

So, in a nutshell: Advertising companies and advertisers are putting millions on technological terrors with empty fuel deposits. And the science that could get them filled, it's completely absent from their current skillsets, and worse even, under their radars. I assume won't be like this forever, but until them, the dirty little secret is that RTB and Programmatic won't provide the elusive, magical solution than the Advertising industry desperately seeks, because they have bought a beatiful race car which lacks fuel. And sorry again to pop your last available survival bubble, but Big Data won't rescue you either.

Personetics, folks. Get accostumed to the word before Gartner throws it around. then it will be available to everyone who can pay them.