Google Testing Exchange Bidding In Dynamic Allocation
Google is working on a lucrative new ad product but some people who’ve seen it think it’s a ‘secret tax’ and it ‘requires us to lie’
In April, Google announced a new ad product which most people in the advertising industry saw as a move designed to stamp out “header bidding,” one of the newest pieces of technology ad tech companies and publishers have been using in an attempt to carve away at DoubleClick’s monopoly.
Header bidding essentially lets Google’s rivals jump to the front of the line in the contest for ad slots, and that largely drives up ad revenues for publishers.
Needless to say, Google isn’t going to take this lying down.
Google’s DoubleClick has been running a test program that gives a few ad exchanges which compete with its own ad exchange, AdX, access to Google’s “Dynamic Allocation” product. The product is called EBDA (exchange bidding in Dynamic Allocation).
But people within the ad tech industry have also been calling EBDA a lot of ruder things, too.
We spoke around a dozen ad tech and publishing executives, the majority of whom broadly shared the same opinion: EBDA in its current form has a long way to go before it convinces the ad tech industry it benefits anyone except Google. There were five big concerns about EBDA, which we have outlined below. Most of the people we spoke to said its planned Q3 launch is likely to be delayed as a result.
It’s telling that almost everyone we spoke to asked us to keep their names anonymous. If publishers want EBDA then exchanges would still do well to participate in order to meet the demands of their customers. So ad tech companies don’t want to be seen publicly dismissing the program and running the risk of being thrown off. Nevertheless, they’re all talking about it in private.
A Google spokesperson sent Business Insider this statement: “We’re continuing to test exchange bidding in Dynamic Allocation taking our publishers’ and partners’ feedback into account as we refine the product. Our goal is to give publishers the features, flexibility and transparency they’re asking for while helping them grow their businesses. We’re really happy with the results so far and look forward to adding more exchanges and partners in the coming quarters.”
What on earth is header bidding and what does EBDA have to do with it?
Here’s a bitesize version.
Google’s DoubleClick has a huge ad tech monopoly. Virtually every publisher uses Google’s DoubleClick for Publishers to get ads served on their websites.
A popular function within DoubleClick for Publishers is called “Dynamic Allocation.” It allows Google’s ad exchange, AdX, to compete directly with publishers’ in-house sales teams to bid for ad slots.
The issue for the ad tech community was that this product was only allowing AdX to compete with these bids. Everyone else had to wait for a slot to become available after the direct deals were sold, greatly reducing their chances of winning in the auction that takes place for ad space as a user loads up a web page.
Ad tech companies figured out a hacky workaround: Header bidding! They asked publishers to insert a piece of code into the header of their webpages which sends out an ad request before AdX can get a look in.
The process looks like this:
Ari Paparo
Header bidding has proved hugely popular. A report from BI Intelligence published earlier this year suggested almost 70% of publishers have adopted header bidding technology, compared to almost zero two years ago.
That hurts Google, because it means other exchanges beat AdX to sell the best inventory. The sources we spoke to said header bidding has been found to improve yield (the amount of money publishers can make from their ads) from anywhere between 10% up to 70%, compared to using Google’s DoubleClick. AdExchanger reported that many programmatic-focused publishers who use header bidding saw Google’s share of revenue decline from 90% to just 40%-50%.
Google’s response was to launch EBDA, which allows select third-party exchanges to bid within the DoubleClick Dynamic Allocation product. (Google has been careful to never say publicly this was a direct response to the header bidding trend.)
Well, we say “launch” — the product is only in the very early stages of testing at the moment, piloting with exchange partners including Rubicon Project, PubMatic and Index Exchange and publishers including About.com, Hearst, Meredith, and Zillow.
BI Intelligence
Our ad tech sources remarked that the early-stage announcement was very unusual for Google, which usually waits until products are near-ready before it announces them. The sense is that Google rushed EBDA to market in order to thwart the rising header bidding trend: If you’re a publisher and you were considering implementing header bidding, hold off for now, as Google is cooking up a solution for you!
In an interview with AdExchanger in April, Google’s director of product management for DoubleClick, Jonathan Bellack, alluded to the fact that header bidding can make web pages load slower, which isn’t optimal for the publisher or the user. EBDA, on the other hand, moves all the bidding to the ad server, which speeds up the process. It’s also a lot easier for a publisher to implement Google’s ready-made solution than add a header tag.
Bellack said: “We are just trying to help publishers make more money without compromising user experience.”
But lots of people in the ad tech community are unconvinced and think EBDA only really serves to give Google a business advantage. Here’s why.
Issue 1: The price to take part is unclear as Google is making everyone sign non-disclosure agreements
When Google first announced EBDA, Ari Paparo, the CEO of ad tech company Beeswax (which isn’t participating in the trial) said it was “pretty good news” for ad tech companies with exchanges and demand-side platforms (DSPs) as they would get more access to more inventory without having to pay an “AdX” tax.
But there are some indications there might be a tax involved.
Some people we spoke to said Google is asking ad tech companies to pay a small fee — one person said a 15% revenue share, while another said the fee was equal to a “mid-to-high single digit” percentage on each bid — in order to participate. Others said it’s the other way around and the publisher has to cough up a commission cost.
Neither side knows what’s going on as Google is making both ad tech companies and publishers sign non-disclosure agreements (NDAs) preventing them from talking to each other.
One ad tech executive told us they were uneasy about signing: “It’s a secret tax and it requires us to lie.”
The fee issue is one that will likely get ironed out ahead of release. One ad tech exec conceded that a one-time fee — because it does cost a publisher to support their ad server — is rational, but a continuous fee is not. It essentially depletes the price of the exchange’s bid every time they come to auction, but could favor AdX, which is exempt of the fee (as it’s unlikely Google would be charging itself to participate).
On the NDA side of things, most publishers are now accustomed to signing contracts that forbid them from revealing prices — particularly when it comes to header bidding — so it’s unclear why this has caused so much consternation among ad tech vendors.
Issue 2: Google pays publishers the revenue, not the exchanges that generate it
One source told us: “It puts their system in the position of the middle-man to dis-intermediate the direct relationship publishers have with exchanges. When you sell something on eBay, you don’t have to run the money through Verizon — you have chosen to do business with eBay.”
Several of the ad tech executives told us Google has created a “black box” issue where they have no idea how or if their bid is being handicapped because they don’t know how much Google is paying the publisher.
A publishing executive with knowledge of EBDA told us they also had concerns about the data they will be receiving from Google: “I can’t see what AdX bids at. They have bid-level reporting, but it’s aggregated.”
In other words, Google isn’t sharing enough data that would allow a publisher to work out whether it was prioritizing AdX or the actual highest bid in the auction. That’s not to say Google is doing anything underhand, but it is not transparent either. More clarity is needed from both sides on the mechanics of the auction.
Jonathan Bellack, Google DoubleClick director of product management.Twitter
Google’s Bellack told AdAge in April: “Everyone in the industry, whether buyer or seller, wants more transparency when it can benefit them. There are a lot of areas where transparency is a big question here. One of the things when it came to these client-side technologies was trying to improve the transparency of what publishers were actually being paid. When you open your ad selection process to a third party claiming what they are going to pay, you want some assurance they will really pay that. Part of what we are trying to achieve with the exchange bidding solution is giving publishers confidence that their reports and earnings matchup.”
Issue 3: Exchanges can’t bring in their own demand
Ad tech companies participating in the program have been told they can integrate their sales exchanges directly — but they can’t let their buyers bid directly through their own demand-side platforms (DSPs).
The only way those ad tech companies can bid via EBDA is if they go through a third-party exchange.
That’s problematic: Ad Tech Company A would have to pay Ad Tech Company B a fee in order to bid through their exchange, so by the time a bid filtered through the system it would be handicapped against AdX by the extra fee. Routing through different exchanges also takes speed out of the process, which again could put the bid at a disadvantage versus AdX.
“There’s something problematic about representing this as direct bidding, because it’s not,” an ad tech exec told us.
It’s worth bearing in mind that Google didn’t create EBDA for ad tech vendors. EBDA is a way for publishers to attempt to get the highest bid for their ad space between a variety of exchanges, without the lag of adding lots of extra code for each separate exchange to their websites.
And Google already has a product that allows DSPs to bid on DoubleClick: It’s called AdX.
Issue 4: Google decides how long the auction stays open for, not the publisher — and Google gets “last look”
In the current iteration of EBDA, Google decides how long the auction stays open for before the ad slot is awarded to the highest bidder.
When it comes to monetizing their web pages, a publisher has to play a delicate balancing act between maximizing yield (by keeping the auction open for longer) and minimizing page “latency,” the time it takes for a page to load. Many of the ad tech executives we spoke to said it’s the publisher’s website, so it should be the publisher that decides when to call time out.
“There are early hints that [Google is] giving us a time out that is not achievable by any speed limit,” one ad tech exec told us.
Google’s AdX also maintains “last look” at every auction. If it wanted to, AdX could see that the highest bid for a particular user from all the exchanges was $2.00. By having the “last look,” AdX could simply bid $2.01 to win the auction. The other ad exchanges would simply bid their highest price, without being able to see the other bids in the auction.
In Google’s defense, this is how most header bidding solutions work: The vendor that operates the header bidding code also gets the last look advantage — so Google isn’t necessarily doing anything new, but it’s yet another complaint from ad tech companies who feel they are getting a raw deal.
Issue 5: Ad tech companies feel Google is difficult to work with
Google and ad tech companies have always had a natural resistance to working with one another. They fiercely compete, but every now and then they are forced to work together too. It’s a classic “frenemy” relationship, as advertising mogul Sir Martin Sorrell famously described it.
One ad tech executive told us: “They technically want everything their way, in Google style. It’s hard to convince them or get them to accept a different way of thinking.”
This chimed with the opinion of an executive at another ad tech company: “The Google response is not super flexible. ‘This is the way it’s going to be’. We are not embracing [EBDA] whole-heartedly but we are expectant to see what potential lies in it. We are aligned with publishers and we will only do this as long is it is effective for our ability to drive revenue for publishers. If it depletes that, then we won’t do it.”
In May, Google hired Sam Cox, the former VP of global partnerships at ad tech company MediaMath as group product manager for AdX buy side and policy. The ad tech executives we spoke to were hopeful this move will help open Google up to their way of thinking.
Sam Cox, Group Product Manager, AdX BuySide and Policy at Google. YouTube
“There’s a sense of fairness that he brings to the table and he will be looking at how the marketplace will work,” one ad tech executive told us.
Some people think the header bidding/EBDA saga might need further outside intervention too.
One source said: “All this calls for is for the [internet advertising trade body] IAB to step in and help a group of us determine what is a fair treatment in this situation. The header bidding problem is a perceived conflict of interest between the container and the submitter … the IAB needs to get involved as we need a definition.”
The opposing view: “I’m thrilled that Google did it.”
Mike Smith, SVP of advertising platforms and core audience at Hearst. LinkedIn
One person did agree to speak with us on the record. Mike Smith is the SVP of advertising platforms and core audience at Hearst and SVP of revenue platforms and operations at Hearst Magazines, one of the world’s largest publishing companies, home to titles such as Cosmopolitan, Harper’s Bazaar, and Esquire. Hearst is involved in the alpha testing of the EBDA product and is a close partner with Google on a number of projects.
Smith told us while he doesn’t share the concern of others about EBDA, he can understand there’s skepticism about a bias in any sort of auction technology that’s controlled by Google because it owns the largest ad server that most publishers use and the largest ad exchange. However, he believes EBDA is a “great initiative” for several reasons.
“The software that links the ad server and the exchange in Dynamic Allocation has always been a very critical yield management process and the fact that that’s been opened to other suppliers of advertising and their demand partners is a wonderful, giant step forward,” he said.
Hearst runs around six different header bidders today, which makes the company a lot of money, but Smith is hopeful that simplifying the administration of auction technology will remove current inefficiencies, which should produce higher yield for publishers in the long run.
“I’m thrilled that Google did it, thrilled. It advances the simplification of programmatic and that’s important because if it continues to be complicated to configure and operate, it will slow its growth,” Smith added.
Where do we go from here?
Road closed for EBDA? Unlikely. FlickrCC/Yuya Sekiguchi
Mentioning the word “Google” around most ad tech executives or publishers usually results in an emotional response, so it’s hardly surprising there are a number of vocal critics of the program.
There have even been some suggestions in the ad tech community that EBDA is just “vaporware” that will never actually launch, and that Google made the announcement simply to slow down header bidding adoption. This is a pretty far-fetched suggestion, considering the product is in active testing and Google has built teams to support it.
Ultimately, most of the people we spoke to are seeking to ensure that EBDA will treat all exchanges equally. If Google can provide ad tech companies and publishers more transparency and assurances that it isn’t rigging the game, then the chorus of criticism about the product will likely die down.
Most of the sources we spoke to said Google had been aiming to get this product to market in the third quarter of this year. None of them believed that was likely.
Business Insider, Tuesday, August 2
(62)
