How it all started: The DoubleClick acquisition

The Monopoly Report is a new free newsletter from Marketecture covering everything relating to Google’s ad tech antitrust trial.

How it all started: The DoubleClick acquisition

It seems appropriate to kick off this new newsletter with the story at the beginning of all of Google's problems, the 2008 acquisition of ad tech behemoth, DoubleClick for $3.1 billion. By reviewing the rationale of the deal and the mechanics of how the integration worked we can set the stage for the alleged abuses of monopoly power that are being charged against the company.

Dynamic Allocation and the DoubleClick Ad Exchange

It is often overlooked in the discussion of Google's behavior that prior to the acquisition of DoubleClick, that company had by 2007 already launched an exchange which included a fundamental competitive advantage called “Dynamic Allocation.” This feature, built deeply into DFP (what is now GAM), allowed the DoubleClick exchange to compete in real time against publisher direct sold inventory without redirect, waterfalls, or other inefficient techniques. This gave the new exchange a huge advantage in gaining access to premium publisher inventory over external SSPs like Rubicon (now Magnite) or ad networks like Undertone, BlueLithium, etc.

In the late aughts when programmatic was first being innovated by Right Media (Yahoo) and AdECN (Microsoft), it was seen almost entirely as a solution for monetizing remnant inventory. Publishers would implement a remnant solutions, be they exchanges, SSPs, or ad networks in priority order based on estimates of their monetization potential. In the pre-header bidding era, the ability for DoubleClick’s ad exchange to instead “jump the line” and grab inventory when it had demand was a huge advantage, as well as an efficiency for publishers.

By launching an exchange with a unique advantage built into the ad server, the seeds were sewn for using that monopoly to benefit the exchange’s liquidity. Even in 2008 it could be argued that DFP was a monopoly or close to it. With the notable exception of some very large publishers like MSN, Yahoo, and the New York Times, virtually every other major publisher around the world relied on DFP for ad serving. The DOJ’s complaint claims that at the time of the acquisition DFP had 60% market share.

AdSense, GAM, and YMAG.xls

I recently had Gokul Rajaram, sometimes called the “Godfather of Ad Sense,” on my podcast and he told the story of how he, Sergei Brin, and Susan Wojcicki built that product. I’m going to drastically simplify, but here’s the logical thought pattern that went into AdSense, and eventually DoubleClick:

1. Tens of thousands of advertisers are buying search ads using AdWords (now “Google Ads”).

2. What if we took that demand, and used it outside of search, on web pages? If we analyze the context of a webpage, it’s almost like a set of search terms. Thus, AdSense.

3. Wow, we have tens of thousands of publishers running AdSense tags on their pages, what if we ran image ads in those slots? Thus, the Google Content Network and image ads in AdWords.

4. Wait, why can’t we get the big budgets for our image ads? It seems like agencies and publishers all use this weird Web 1.0 company called DoubleClick to buy and sell ads. We have to do something about this…

5. Let’s build our own ad server to compete with DFP, we’ll call it Google Ad Manager. Thus, GAM (the first version).

6. Wow, this is way too hard, let’s buy it instead. Thus the $3.1 billion price tag.

While Google was a motivated buyer of the business, there was also some fortuitous timing on the sell-side that lined up the DoubleClick business for sale.

DoubleClick had been taken private in 2005 by private equity firm Hellman & Friedman, and new management took over led by David Rosenblatt (now CEO, 1stdibs) and Neal Mohan (now CEO, YouTube). In short order this group led a masterful turnaround of a moribund business, launching the above-mentioned ad exchange, and disposing of numerous slow-growth assets. According to a source I spoke to who was close to the sale, the executives and the board determined there was sufficient interest to run a process to sell the company after less than 18 months in private ownership.

The entire impetus of the DoubleClick sale process was to give preferred access to the publisher inventory within DFP based on that product’s ubiquity. How do we know this? Because there was a spreadsheet, known as YMAG.xls, now lost to history.

Neal was put in charge of explaining and justifying the synergies of a deal to the prospective buyers. Neal realized that while the various software businesses of DoubleClick were valuable and generated probably $400 million per year in revenue (my guesstimate), the real value was in the proprietary access to inventory, and the resulting high margins that could be had by anyone filling that inventory.

Economically, the exchange business was also much more attractive. ad serving rates had plummeted to $0.01 to $0.03 CPM while the an exchange could reasonably charge 10 to 15% of media sold — an order of magnitude more money per impression.

Who exactly had a large liquid pool of demand? Only four companies: Yahoo, Microsoft, AOL, and Google. Thus, YMAG. What did the spreadsheet calculate?

• DFP has X billion impressions running through it each month

• Assume X% of those are remnant or unsold

• Assume your liquid pool of demand could fill, say, half of those remnant impressions

• Assume your margin on those impressions is X%

• Therefore, if you own DoubleClick, you will make incremental BILLIONS of dollars.

This was the whole pitch. And it worked, because there was a bidding war for the company, that ultimately left Google as the owner. And thus, our antitrust story begins.

A side note: I mentioned that the spreadsheet in question is lost. If anyone has a copy of it please send to me anonymously and I will honor your confidentiality as a journalist.

Unique demand

Even today, the easiest way for any exchange to get access to supply is to have unique demand. There is plenty of non-unique demand — DSPs will bid their demand into any one of the undifferentiated pipes of supply they integrate with. But unique demand, that a publisher will only be able to earn from your differentiated pipes, is a huge advantage in the programmatic market.

Google Ads is literally the largest single source of unique advertising demand in the world. The company doesn’t disclose the number of accounts it has but it is almost certainly in the tens of millions globally. Various statistics show that half of small businesses have accounts, and surely every single large advertisers uses Ads.

This sets up the compounding competitive advantage that Google has enjoyed in the display advertising market, which is nicely detailed in the DOJ complaint (pages 6-8). It goes like this:

1. DFP is a sticky monopoly — publishers do not want to switch.

2. AdX is deeply integrated into DFP to give an inventory advantage over other exchanges through Dynamic Allocation.

3. Google Ads demand is only available through AdX, so in addition to having preferential supply access, it also monetizes better than other exchanges.

A decade of cat-and-mouse

By 2010 or so Google had integrated its unique supply into the exchange and relaunched as “AdX”. Competing exchanges had to fight with two hands tied behind their backs — they couldn’t compete fairly on supply or demand. That is, until the advent of yield management, and header bidding. In future newsletters we’ll be covering how the larger ad tech ecosystem worked to dull Google’s advantages, and how Google used programs with baroque names like Bernanke, Project Bell and Jedi Blue to keep its advantages in place.

Subscribe if you want to hear how it plays out.