Credit: Maersk line

The budding Google-Facebook-Twitter war

Or, the tyranny of the ad tech plumbing

Fact: Google and Facebook (and soon Twitter) are involved in a war to define the value of advertising. The winner will own the yardstick by which billions in advertising are judged. That war is a pure vestige of the technology we’ve wrapped around advertising, and the contest is only just beginning. To understand the war, we need to understand (slightly) the technology.

The magic box

Any idea who Malcolm McLean was?

I bet not. But that one man changed our economy more than most.

McLean was the inventor of the intermodal container, those metal boxes one sees piled into immense heaps on the cargo ships coming from China. The genius of the container is that the entire workflow around transporting physical goods is standardized on the same 8'x 8' x 40' box. Manufacturers load goods onto eight-foot-wide palettes straight into the box. The box becomes a freight car when loaded on railroad wheels, and once arrived at the ship is directly hefted aboard via those immense cranes that dot every modern port. Piled like Legos on the ship, the boxes arrive at another port and are loaded on a truck frame this time,and are driven to their destination. And it’s universal: Whether the ship docks in Singapore or Oakland, cargo will be swiftly loaded and unloaded via the magic of containerization. Compared to the manual stevedoring that used to plague shipping, containers make our global supply chain possible.

What’s that got to do with ads?

Modern digital advertising has also settled on a containerized box — or, rather, a set of them. They’re known as the Internet Advertising Bureau (IAB) ad units for desktop ads, and the Mobile Marketing Association (MMA) ad units for mobile. They come in the standard sizes, measured in pixels — 728x90, 300x250, etc.—of every banner ad on the Web, with other sizes for mobile.

As in freight, the standard sizes means the ad that runs today on can also run on Yahoo! Finance or the New York Times tomorrow. It also means that much of the technical plumbing around advertising also works efficiently: the ad servers that feed you the flashing pixels in the ad, the same way web servers serve pages, all work with the standard sizes. The analytics software that slices and dices your data by things like publisher, ad size, and placement on the page, also assume the standard sizes. It’s containerization applied to paid media, and in general, it works.

Here’s where the trouble starts.

To continue the nautical theme, there are some ships in this analogy that consider themselves either too big or important to accept containers. Either because they don’t like the look of them on deck, or because they claim moving freight is really a side business and not really their mission in life, no containers are allowed. So if you want to move anything through them, you suddenly have to repack everything in whatever arbitrary, random form they have.

This situation is what we politely call ‘native ad formats’ in the ads business. Those ships are products like Google Search, Facebook, and Twitter. Either because they started as user-focused viral plays unconcerned with monetization (Facebook,Twitter), or because they very intentionally bucked a going standard, basically because they could (Google Search), the net result is that standard ad formats get no play on their eyeball jukebox.

Opting for native formats causes two immense frictions for those platforms, particularly to anyone trying to bolt a traditional ad exchange onto their inventory (as we did at Facebook via FBX, and Twitter is attempting to do now with MoPub).

Those two frictions are the obvious one of differing ads creative (i.e., the ad itself is graphically different), and a more subtle one that concerns how ads are delivered and counted in the ads world. The battles surrounding this latter point, I claim, will determine the ad empires of the future.

Ad servers: those abaci of the ads world

Why do ad servers matter?

At first glance, they seem inconsequential. To continue my (now perhaps stretched) nautical analogy, they’re merely the container cranes that shuffle boxes around. But they’re much more than that.

Look: The advertiser doesn’t trust their agency, the agency doesn’t trust their trading desk, the trading desk doesn’t trust the ads-buying software they’re using, and the ads-buying technology company doesn’t trust the exchanges. The only thing that keeps this dishonest world honest is the existence of an agreed-upon source of truth. That oracle is the ad server. If a marketer wants to reach a million people in the eastern U.S., showing each person no more than four ad impressions between the hours of 4-10pm on Thursday (a common buy for movies, incidentally, which always launch on a Friday), then that marketer will only be satisfied if the ad server report says that’s what he got.You see, the ad server isn’t merely a data server spewing forth pixels on demand; it’s also the accounting system that decides what got delivered when, to whom, how often, and where on the Internet.

And that’s just for fairly brainless brand marketing buys that want to make a big, broad splash.

For marketers who track their performance, the tyranny of the ad server yardstick is even greater, since it defines the units, in addition to the accounting, of marketing dollars.

Let me explain.

This is getting wonky and your eyes might start glazing over, but really, defining ads metrics is as core to the Internet as the definition of the meter is to the International System of Units. Boring, but at the core of a universe of phenomena.

We define our metrics, and then our metrics define us

Have you ever driven on a highway and seen those roadside billboards? You know: MCDONALD’S, FOUR MILES ON RIGHT! What if you removed one of the billboards? How many burgers would go unbought as compared to when the sign was there? It’s small, but measurable.

Whether we’d like to believe it or not, even if we don’t click or mentally process the image assaulting our retinas, our subconscious, that cradle of desires both holy and unholy, has seen it. And four miles down the road, when you see those Golden Arches appear, you will be more inclined to pull over for that prefab greasepuck than otherwise.

That sign four miles away was not the Golden Arches storefront that lead to your pulling over, but it contributed an infinitesimal probability to your acting on the prompt. As such, if we’re to be smart marketers, we should assign credit to that lone billboard, if only incrementally.

In the online world, seeing that sign miles aways is known as a viewthrough: the physical seeing of an ad, but with no resulting immediate action. Turning off the highway when you see the Golden Arches is known as a clickthrough: an immediate action following a marketing prompt. The relative value of viewthroughs versus clickthroughs is a subtle measurement question that needn’t slow us down here. But appreciate for a moment what privileging one or the other means for someone trying to decide whether to buy more billboards or spend more on venues and storefront bling.

In a world where we know if people are hungry and have our choice of venues to erect a McDonald’s, we’ll invest in rent and storefront bling as we already know we’re catching somebody at a weak, hungry moment (say, at a baseball stadium or a mall’s food court). We don’t need signs all over the stadium or mall. We know they’re hungry and where they congregate; we just need to pay the price of being there.

In a world where we’re not sure if people are hungry and advertising is as much inducement as information, we’ll plaster the world with billboards in the hope of getting consumers to a store, and not worry so much about happening to catch someone hungry who’s walking by. We’re manufacturing necessity, rather than exploiting it, which is the situation on most highways. Ergo, we have billboards every other mile but stores thirty miles apart; we’re betting on inducement, not pre-existing intent.

Pints and pounds, liters and meters

Google comes from the search world, in which a person literally types in a desire for something. All Google has to do is get them to the storefront. The value of Google’s clickthrough is immense and expensive (which is why Google prints money). The value of a viewthrough to them is nil. In fact, it’s so nil that Google opts not to even show you an ad if your query doesn’t seem commercial, literally choosing to show you a blank billboard much of the time (roughly half of Google queries don’t generate ads next to them—test it yourself). To them, the claims of viewthrough value seem bogus—an attempt to justify wasteful media spending. As a result, their ad server only really counts clickthrough. The manual literally states ‘clicks trump impressions’, and they drive that point home in every demo.

Facebook, on the other hand, worships viewthrough value. Since Facebook has no real intent data (e.g., they don’t know at all what you want), they have to manufacture desire rather than exploit it. Hence in their accounting, viewthrough is valuable indeed, and they would very much like credit for that splashy image (or soon, video) they inserted in your feed, even if you didn’t click on it.

(Want to send a Facebook marketing or sales person into fits? Tell them that ‘clicks trump impressions’ and that Facebook doesn’t measure up to Google on a click-to-conversion basis. Enjoy the sputtering and whipping out of PowerPoints.)

Let’s go further. What’s a Facebook ‘like’ really worth? Sharing a page post? Commenting on something? Those all have value. Like the view through impressions we mentioned before, maybe incremental value, but value nonetheless. And if your site generates a mountain of likes (rather than clicks), you need an ad server whose accounting reflects that. Which is why Facebook bought Atlas, the second most popular ad server on the Internet, from Microsoft in a deal announced earlier this year.

The ads battle between Google and Facebook is akin to whether we use English or metric units, and this battle has just started.

Titillating insider anecdote: the Atlas acquisition

When the Atlas deal was in play at Facebook, all of the product managers involved ran to the IT Help Desk to get a Windows laptop to check out this thing we were supposedly buying. Believe it or not, Microsoft’s Atlas only ran on specific versions of Internet Explorer (in case you’re too young to remember, that’s a browser). Unfortunately, unlike vintage clothing, user interfaces never get so old that they’re fashionable again. It was creaky and ancient and possessed a flow so convoluted that even professional ads people got lost in the mess. But it had 20 percent marketshare.

Many non-Facebook people discreetly expressed disbelief about the acquisition to me, questioning how we could buy such a neglected piece of legacy technology. I don’t think it was such a bad move, necessarily. For a modest sum, Facebook jumpstarted a presence in the display ads world in which we were an afterthought.

Of course, with acquisitions it’s not what you pay, it’s what it costs you that matters. Facebook will have to feed and maintain a mountain of crap code (and the crappy coders who created it) for a long time to come. But hey, Facebook now has an abacus that people use.

The meat of the thing

Alright, enough about containers ships and greaseburgers, what’s this mean to Facebook and Twitter?

Whether they like it or not, social media platforms like Facebook and Twitter will have to play ball with the display ad server world. Facebook has already figured this out and invented what’s called ‘view tags’ to deal with it. How does it work? The ads you see on Facebook are always served from a Facebook domain: One can only imagine Zuck’s conniptions if ads on Facebook were ever to load slowly thanks to third-party ad serving. Instead, a tiny 1x1 pixel from Google, or whomever else, gets loaded on Facebook ads. So you’re loading a little piece of Google every time you hit Facebook, whether you realize it or not. That’s enough for an outside ad server to touch the user’s browser and note the ad impression: the accounting benefits of the ad server, without actually serving the ad. Note, this is totally public and visible on any browser. In fact, it’s documented in Facebook’s developer docs here.

Effectively, there are two sets of books being kept: the ads analytics that Facebook provides and no one uses except pure-play Facebook advertisers who don’t spend money elsewhere (a relatively small piece of the pie), and then real analytics being crunched inside the ad servers, which is what large marketing budgets that spend money far and wide use. The latter is what matters to any social media platform which wants budgets to expand beyond their pre-existing spend (which is everybody).

Twitter + ad exchange = TWIX?

You don’t escape the ad accountant in this world. The best you can do is provide your own accountant, who uses the most favorable units, and which hopefully become the accounting standard of record.

Twitter hasn’t figured this out yet, but I bet it will. Soon enough you’ll see pixels firing to Google from

What’s a retweet worth? What’s a Twitter ‘favorite’ worth? What’s a follow worth? Those questions will be asked once the Twitter/Mopub integration happens and there’s money flowing through. After dropping a million dollars on Twitter (a million that once went to traditional display or, even better, Facebook), marketers are going to want to know what they got in return. Twitter salespeople, like Catholic priests discussing the Eucharistic miracle, will recite canonical homilies about social engagement transubstantiating into sales and brand lift. The marketers will be rather more skeptical and demand proof. Twitter will have to create an ad server to answer those questions.

You can escape the ad format by going native, but you can’t escape the accountant’s ledger. The big question is whose accountant will win out in the end, and whether he lives in Mountain View, Menlo Park, or San Francisco. As of now, I predict it could go any which way.