Online exhibit commercials represents a large source of sales for online publishers. Because of its vital importance for publishers and advertisers, competition in the commercials surroundings is desirable. Yet, in the “programmatic” era, the sector is characterized by a high degree of opacity and a few of its segments seem like ruled by Google, with considerations being expressed that it may engage in anti aggressive concepts. Against this background, the intention of this paper is to explore the show commercials ecosystem and discuss applicable pageant law issues.
It first discusses market definitions and shows that Google may be dominant on several ad tech markets after which describes how programmatic advertising capabilities in practice. Finally, the paper identifies a few practices that can amount to abuse of a dominant position in breach of Article 102 TFEU. Since the first ever clickable banner ad for ATandT seemed on HotWired. com in October 1994,1 online commercials has developed into a major company, with an envisioned global turnover exceeding 260 billion dollars in 2018. 2 Online ads represents a serious stream of income not only for tech giants providing time-honored facilities monetized by ads, corresponding to Google, Twitter and Facebook, but additionally for hundreds of web site owners, from major online newspapers with millions of readers to blogs catering for specialised audiences. In 2017, online advertisements represented 98% of Facebook’s revenues,3 in addition to more than 86% of Google’s and Twitter’s income.
4 The same year, the New York Times Company, which owns the eponymous magazine, earned approximately one third of its total sales from online advertising,5 anything of its revenues being basically generated via subscription fees. While search ads represents a huge a part of the web advertisements industry, this paper focuses on what is known as show ads since, it represents a large, and sometimes the only, source of sales not only for giant tech corporations, but additionally for a myriad of publishers, large and small, which offer powerful content to Internet users. 8 But for online exhibit advertising, many such publishers doesn’t subsist, and the Internet can be impoverished. Display advertising also is important to advertisers, in selected once they seek to elevate “brand attention” among consumers. Because of its vital significance to advertisers and publishers, healthy pageant in the advertisements atmosphere is attractive.
Yet, regardless of the extraordinary growth of online display advertising, the picture is not completely rosy. In the “programmatic” era, where ad stock is sold through automated resolution making tactics managed by “ad tech” intermediaries, the web exhibit ads sector is characterized by a high degree of opacity, and publishers and advertisers have expressed issues about the so called “ad tech tax”, i. e. the large and opaque fees utilized by quite a few intermediaries. 9 For occasion, The Guardian revealed in 2016 that “in worst case scenarios, for each pound an advertiser spends programmatically only 30 pence actually goes to the publisher”, that means that ad tech intermediaries could extract up to 70% of programmatic sales. 10 Moreover, while the ad tech sector comprises a wide variety of intermediaries, its main segments are ruled by Google, with issues being expressed that it may engage in both exploitative and exclusionary techniques.
11It is thus not shocking that festival professionals are hunting closely at the competitive dynamics in online advertising. The French Competition Authority introduced in 2016 a sector enquiry in the net advertising sector, followed by a stakeholder session. On 6 March 2018 the Autority made public its opinion, “through which it analyses a very complicated market, characterized by a fragile aggressive equilibrium” the “FCA Opinion”. 12 The German Competition Authority announced on 1 February 2018 that it was launching its own sector enquiry into online advertisements, “ue to the great economic significance of this sector for advertisers and content material suppliers active on the Internet and in view of discussions concerning the challenging competitive atmosphere in this market”, 13 and launched a short paper on the same topic. 14 In the UK, the Select Committee on Communications appointed by the House of Lords noted in its 2018 Report the loss of transparency in digital advertisements and advised that the Competition and Markets Authority CMA “conduct a market study of digital advertisements to examine whether the market is operating fairly for agencies and clients. ”15 On 12 October 2018, the independent Digital Competition Expert Panel introduced a public consultation “to behavior an unbiased review of the state of pageant in the digital financial system.
” The questions to which interested parties are invited to reply fear, inter alia, competition in online ads. 16Of these various initiatives, the FCA Opinion is the only one that mainly makes a speciality of exhibit advertising, but it continues to be at a reasonably high level of generality. It has been reported, but it surely, that the French Competition Authority may initiate complaints towards express undertakings based on the findings of its Opinion,17 and on 8 November 2018 the Authority announced “the starting of litigation investigations on abusive data collection and processing in addition to access restrictions. ”18 Even so, at this stage, there’s little data in the public domain concerning the competition issues that can arise in the display commercials sector and we aren’t acutely aware of any scholarly paper dedicated to this area. Against this background, the aim of this paper is to discover the exhibit advertisements surroundings and talk about the pageant law issues that can arise during this sector because of Google’s manage of the ad tech value chain. We should at this point offer a word of warning.
As the readers will perceive, online display advertising is very complex as its mechanics contain multiple electronic techniques, adding real time auctions, carried out by computer systems in milliseconds. Thus, the technical parts of this paper constitute our greatest effort to describe these methods in a manner that is on the market to pageant law and economics practitioners, while the legal analysis seeks to apply EU festival rules to the ad tech surroundings. Our pageant analysis is tentative at this stage given the limited amount of publicly available data on some of the practices that create pageant concerns. Yet the stakes are high pondering the critical significance of reveal advertising for both advertisers and publishers, and it is hoped that the abovementioned initiatives taken by competition specialists will throw some light on a slightly opaque sector. Search ads refers to text ads displayed above or below the search effects of a search engine every time a user enters a search query that suits with a keyword on which advertisers bid. For example, when a user enters a search query in Google e.
g. “cars for rent”, Google will display in its Search Engine Results Pages SERPs along with and usually above so called “biological” search outcomes i. e. herbal effects that are displayed in line with the search engine’s algorithm “paid” search results, i. e. ad links.
19 Search ads is said to be most a success in terms of “conversion”,20 in that the user coming into the search query expresses her interest in a given product or service, and is thus much more likely to perform the favored action. Initially, online exhibit commercials was no more complicated than ordinary, offline commercials, e. g. in print media or TV. Publishers wishing to monetize their available ad space called “ad inventory” or just “stock” engaged in direct, bilateral negotiations with advertisers in order to sell ad space at a given price.
Such “manual” media buying had a few drawbacks. First, it was time ingesting and required a dedicated salesforce to conduct the negotiations. Moreover, publishers faced the “fill” risk, i. e. that they can be left with unsold inventory. Finally, the common use of Internet introduced with it the emergence of thousands of websites with accessible ad space, which couldn’t nearly be sold directly to advertisers.
In its most generic form, called programmatic real time bidding “RTB”, every time a user visits the site of a publisher, advertisers are invited to bid for the available ad space so as to show their commercial to the certain user called “ad impact” in a real time public sale. The maximum bidder wins the ad influence and gets to serve the ad that the user will basically see on the website. Remarkably, the complete process from the instant the user types in his/her browser the URL of the publisher’s web site until the ad is eventually shown lasts only fragments of a second, usually about 300–400 milliseconds. 22Initially, programmatic ads was used to facilitate the sale of “remnant” stock, i. e.
inventory that publishers had not controlled to at once sell to advertisers. Publishers would opt to sell their most expensive, high yield stock called “top rate”, e. g. the top of the home page of a web newspaper through direct sales. However, that is now not the case. Programmatic commercials, once linked to cheap ad inventory of dubious exceptional, is being more and more used to sell “premium” stock.
It is reported that by 2019, 67% of worldwide digital ads might be bought programmatically,23 while according to a report by eMarketer, greater than 80% of digital exhibit ads in the US will be bought programmatically in 2018. 24Programmatic advertisements has brought with it a few crucial adjustments. First, there was a shift from the context i. e. the content of the site to the user. Advertisers place less emphasis on where their advertisement can be shown, and instead base their choice in response to the express user that may be uncovered to the ad.
If the user is within the target group of the campaign set up by the advertiser, the latter could be willing to exhibit its ad even on websites whose content material bears no relationship to its product. For example, while luxury watch makers historically sought to affiliate their ads with certain types of content material e. g. the “how to spend it” page of the Financial Times, which points many luxury items, here is no longer necessarily the case as advertisers are now able to reach tailored audience segments that correspond to their crusade goals even with the web site they visit. Second, user data are more efficient than ever.
In order to target a particular user, it is crucial that advertisers purchase access to data about that user e. g. behavioural data extracted from looking historical past, sociodemographic data equivalent to age and gender or geographical data to which they may are looking to show their ad. The more and better user data advertisers have, the upper they are willing to bid for a user within their target group, most popular in precept to higher sales for the publisher. If, on any other hand, advertisers have limited data concerning the user, they’ll take a more careful strategy and bid lower the bid is “blind”.
At its most elementary, advertisers are purchasing access to ad inventory and publishers are compensated for granting such access. The problem is to figure out what occurs in between them. This is an important question, since it has been prompt that publishers may result obtaining as low as 30% of what advertisers pay,27 and there are reasons to agree with this would be due to a lack of pageant in the ad tech market. Even though the life of a number of actors could give the impact of a fragmented panorama with dispersed competitors, it has been urged that Google has managed to hang a stronghold, in that it is just about the market leader across all the steps of the value chain. But first it is effective to present the a number of actors and clarify their role.
The key actors are the following:28 Publishers e. g. online newspapers serve user content e. g. news articles it is monetized by promoting ad stock to advertisers. Advertisers e.
g. car manufacturers buy ad inventory on publishers’ webpages to sell their brand to targeted users. Publisher Ad Servers are tools that publishers use to administer their ad inventory. A writer ad server determines and information how ad stock is filled every time a user visit the publisher’s website. 29 Examples are Google’s DoubleClick For Publishers “DFP”, lately rebranded as “Google Ad Manager” after its integration with AdX see below,30 the OpenX ad server and the AdZerk ad server.
Advertiser Ad Servers are tools that advertisers use to administer their ad campaigns. An advertiser ad server performs two fundamental features: it a stores and offers the commercial called “inventive” in ad tech jargon and b helps advertisers video display and optimize their ad campaign by monitoring where ads are served and featuring precise reporting on their performance e. g. click via rates, etc. .
31 An instance is Google’s DoubleClick Campaign Manager, these days rebranded to “Display and Video 360”. Supply Side Platforms SSPs organize demand for ad stock and help the writer choose the main profitable ad to display. 32 Traditionally, SSPs were utilized by publishers to connect to ad exchanges to sell their inventory. However, through the years SSPs have advanced, with many now functioning as ad exchanges themselves, allowing publishers to connect directly to DSPs in preference to connecting via an ad trade. For this reason, ad tech experts often use the terms SSP and ad trade interchangeably. SSP examples are Google’s Ad Exchange “AdX”, AppNexus, PubMatic and One by AOL.
Demand Side Platforms DSPs manage the buying of ad inventory for advertisers via a single leadership interface. DSPs are used by advertisers to attach to an ad exchange/SSP and buy ad stock. 33 DSPs can even include data processing functionalities to assist advertisers find the top-rated impressions for their ads. Examples of DSPs are Google’s DoubleClick Bid Manager DBM, DataXu, MediaMath and Amazon DSP. Ad Exchanges are virtual marketplaces for ad inventory where supply and insist meet. Traditionally, publishers supply ad inventory through SSPs and advertisers bid in real time through DSPs.
Examples of ad exchanges are Google’s AdX, AppNexus, The Rubicon Project, OpenX and One by AOL. As noted above, SSPs and ad exchanges, while traditionally separate services, are more and more provided for as integrated solutions, such as in Google’s AdX, which has been lately built-in with DFP to form Google Ad Manager. Ad Networks pool ad inventories from a huge choice of publishers after which sell them in slices to advertisers. 34 Ad networks can buy and sell directly, buy and sell inventory on ad exchanges, or some combination of both. An example is Google’s “AdSense,” which permits small publishers “accomplice sites” to sell ads to Google demand assets. AdSense is accessed via AdWords, a programme that allows advertisers to create ads, which will appear on applicable Google search outcomes pages and Google’s community of associate sites.
Google companion sites form the Google Display Network GDN, which contains greater than two million websites and is said to hide over 90% of individuals active on the Internet. 35Data Management Platforms DMPs and data suppliers are guilty for collecting, storing, organizing and analysing massive quantities of information collected from a whole lot of sources first party and third party data developing unique user profiles, often across different contraptions. Examples of DMPs are BlueKai Oracle, Weborama and Adobe Audience Manager. DMPs are often linked to a DSP to help advertisers target their viewers. 36 Examples of knowledge providers include comScore and IAS. Supply Side Platforms SSPs arrange demand for ad stock and help the publisher choose the most profitable ad to demonstrate.
32 Traditionally, SSPs were used by publishers to connect to ad exchanges to sell their inventory. However, over the years SSPs have evolved, with many now functioning as ad exchanges themselves, enabling publishers to connect without delay to DSPs as opposed to connecting through an ad exchange. For this reason, ad tech specialists often use the terms SSP and ad exchange interchangeably. SSP examples are Google’s Ad Exchange “AdX”, AppNexus, PubMatic and One by AOL. Ad Networks pool ad inventories from a large collection of publishers after which sell them in slices to advertisers. 34 Ad networks should purchase and sell at once, buy and sell stock on ad exchanges, or some mixture of both.
An example is Google’s “AdSense,” which allows small publishers “companion sites” to sell ads to Google demand assets. AdSense is accessed through AdWords, a programme that allows advertisers to create ads, that will appear on applicable Google search effects pages and Google’s community of accomplice sites. Google partner sites form the Google Display Network GDN, which accommodates greater than two million internet sites and is expounded to cover over 90% of individuals active on the Internet. 35Now that the reader is accustomed to the actors in the display advertising environment, it is possible to discover and flag up some key pageant law issues coming up during this sector. We first investigate market definition and dominance Section A.
We find that there are causes to agree with that the ad tech markets as currently defined by competition authorities stay at too high a level of generality and will be disaggregated into more accurately described markets. We also find that facts suggests that Google may be dominant on some ad tech markets. We then identify a couple of Google conducts, that could produce exploitative and exclusionary resultseasily in breach of Article 102 TFEU Section B. The next query is whether the market for online ads should be sub segmented among search and non search i. e.
demonstrate advertising. Such contrast was regarded by the European Commission in Google/DoubleClick, 39 but the issue was indirectly left open. The same approach was followed in subsequent merger manage decisions. 40 In its 2010 opinion focusing on search ads, the French Competition Authority adopted a clear contrast among search and exhibit advertisements, citing the disparity in the selection of users of show and search advertising and the difference in capabilities of the two commercials types stemming from the restricted text nature of search ads. 41 The same view was expressed in its 2018 opinion. 42 On the other side of the Atlantic, the Federal Trade Commission FTC found in the Google/DoubleClick merger that search and demonstrate ads were not substitutes for each other.
43As noted above, advertisers may purchase ad stock either through the channel of direct sales or via that of intermediated sales. In Google/DoubleClick, the European Commission regarded that a separate market for intermediation in online ads can be defined in view of the fact that there is not any replacement for the provider offered by intermediaries for the sale of smaller publishers’ stock and for the sale of as a minimum part of the remnant inventory of larger publishers that also use the direct sales channel. 44The Commission seems to have protected ad networks and ad exchanges in the marketplace for ad intermediation. 45 Further subdivision between ad intermediation in search ads and ad intermediation in non search reveal ads was considered but left open. 46 The Commission maintained this strategy in subsequent merger decisions. 47In Google/DoubleClick, the Commission described a separate market for the supply of online reveal ad serving i.
e. facilities provided by ad servers and believed that this market can be added uncommon depending on even if such ad facilities are rendered to publishers or advertisers. 50 The French Competition Authority upheld the contrast between ad intermediation and ad serving in a 2010 resolution. 51 In its 2018 opinion it noted that there is indeed “some convergence between ad servers and technical intermediation amenities DSPs, SSPs, ad exchanges etc. ”,52 but it observed that “a similar statement was made in 2008 by the Commission”,53 concluding that “o information emerged from the session would call into question the conclusions of the Commission’s evaluation”.
54For instance, it is dubious that DSPs belong to the same market as ad exchanges/SSPs. DSPs form a definite market since they don’t compete with ad exchanges, but they take part in the auctions arranged by these exchanges/SSPs. Moreover, it sort of feels acceptable to phase the marketplace for ad serving era between ad servers for publishers and ad servers for advertisers. Ad servers for publishers fulfil substantially different needs than ad servers for advertisers and are targeted to alternative client groups. 55 It would seem not going that in the case of a price boom e.
g. of ad servers for advertisers, advertisers would switch to an ad server which is designed for publishers. It is telling that Google itself gifts its era solutions for publishers DFP and advertisers DCM as diverse products, as the French Competition Authority located in its 2018 Opinion. 56It is settled case law that the idea of dominance found in Article 102 TFEU refers to a position of financial energy loved by an undertaking, which enables it to avoid efficient festival being maintained on a relevant market, by affording it the power to act to an appreciable extent independently of its opponents, its purchasers and in some way of consumers. 57In analyzing even if a particular accomplishing holds a dominant position on a relevant market, regard is had to the market share of the engaging in and its competitors, in addition to to “other factors”, namely even if there are barriers to entry or expansion that hinder new competitors from coming into the market or current market players from increasing. 58 Thus, as relating to market definition, the evaluation of dominance is a fact intensive pastime.
It is usually advised that Google has a sturdy grip on the reveal ads atmosphere. For example, a 2015 Forbes article refers to a DFP crash affecting greater than 55. 000 internet sites as “a stark reminder of how a longtime player like Google has quietly executed dominance over the so called ‘ad tech’ industry”. 59 The author notes that Google “is now the biggest and/or dominant player” in each ad tech market adding SSPs, DSPs and ad servers. 60 In its 2018 Opinion, the French Competition Authority observes that during the ad intermediation and ad serving sectors, Google “has held a leading position since its acquisition of DoubleClick in 2008”.
61In fact, DoubleClick marked only the beginning of a chain of acquisitions, through which Google managed to become current in just about every phase across the value chain among publishers and advertisers. In 2010, Google expanded by acquiring AdMob, the most advantageous ad network for mobile. 62 The same year Google bought finest DSP Invite Media63 and in 2011 it acquired best SSP AdMeld,64 which it then integrated to AdX. 65 Google thus now offers the most advantageous ad server for publishers DFP, an ad server solution for advertisers DoubleClick Campaign Manager, an ad network AdSense which is part of the Google Display Network and is accessed by advertisers through AdWords, the optimum ad exchange/SSP AdX, the foremost DSP DoubleClick Bid Manager, in addition to its own mighty data management platform Google Analytics. The presence of Google across the value chain also implies that it could have a unique data advantage.
67 In its 2018 Opinion, the French Competition Authority observes many players pointed out that Google: only let advertisers who buy ad space via their buying platforms mine data generated from the amenities they publish. This implies that Google combines delivering its data and featuring intermediation amenities and ad servers for advertisers AdWords, the DCM ad server and the DBM DSP, which might appear to give it an virtue over its competitors. Advertisers can define viewers segments based on a couple of forms of data that only Google is capable of bring together. This consists of user data, Google’s first party data from the use of Google services, data on internet sites and third party inventories that Google sells during the Google Display Network, AdWords and DoubleClick AdX, and data from third party internet sites and applications that use DoubleClick and share data with Google. 68Finally, the French Competition Authority found that Google has an additional aggressive edge, in that it “is among the rare agencies to supply both demonstrate and search ads amenities to advertisers”. 69 That “allows for it to supply dual channel data analytics services”.
70only let advertisers who buy ad space via their buying platforms mine data generated from the amenities they post. This means that Google combines supplying its data and providing intermediation facilities and ad servers for advertisers AdWords, the DCM ad server and the DBM DSP, which might seem to give it an virtue over its opponents. Advertisers can define audience segments based on several forms of data that only Google is capable of bring together. This includes user data, Google’s first party data from using Google amenities, data on internet sites and third party inventories that Google sells through the Google Display Network, AdWords and DoubleClick AdX, and knowledge from third party websites and applications that use DoubleClick and share data with Google. 68Google’s place in the ad server market seems also protected by loads of elements. First, there’s the presence of switching costs.
An industry commentator notes that s a publisher, changing your primary ad server is not a trivial task. Think of it like doing a mid flight engine swap on an plane. Except that it’s your income engine. It’s hard to think about many publishers needing to take such a risk. 75Moreover, given that DFP is obtainable to publishers close to at no cost,76 rivals may find it harder to attract DFP customers since they cannot compete on price by undercutting DFP’s fees.
Finally, given the close connection in actual fact today, full integration between DFP and AdX, some consumers may be concerned that leaving DFP may affect their revenues from AdX. 77In this sub section, we discuss the mechanics of the actual time auctions that choose the advertiser that gets to display its ad to the user each time the latter visits an internet site. 79 For ease of exposition – but in addition because it is often the case in real world given DFP’s prominence – our example involves a publisher’s site using DFP, as it presently features and assuming that the publisher has enabled a contemporary DFP characteristic called “Exchange Bidding. ”80 We added assume that the writer’s web site is part of the standard Google Display Network “GDN” which is available to advertisers via AdWords. 81When a user visits the site, the user’s browser calls DFP which has an ad arbitration mechanism to decide which ad could be served. 82 As part of that mechanism, DFP first examines even if any directly sold ad is eligible to serve.
If there’s no eligible directly sold ad,83 DFP invitations Google’s AdX as well as any attached third party ad exchanges to submit a bid for the ad influence. 84 Google’s AdX will in turn run its own auction, inviting partaking DSPs/ad networks to submit a bid. 85 Each DSP/ad network will in turn run its own auction, inviting advertisers to submit a bid. In other words, DFP initiates a sequence of sequential auctions: in the first public sale, advertisers compete with each other within a selected DSP/ad network, e. g.
AdWords; in the second auction, DSPs compete with each other within a specific trade, e. g. Google’s AdX; and in the third auction, AdX competes with attached third party ad exchanges within DFP. As soon as DFP initiates the above technique, AdWords passes on information about the user to advertisers that have an AdWords account and invitations them to submit their bids to win the ad impact. 86 Advertisers use the data acquired to calculate how much they may bid – if they bid at all – and return their bids, expressed on a Cost Per Click CPC basis e.
g. the advertiser bids to pay 4 € for each time the user clicks on its ad. AdWords then selects the highest bid, which wins the auction. However, the advertiser doesn’t pay what it has definitely bid. Instead, the advertiser pays only what’s had to rank immediately above the second one maximum bidder, which can be 1 cent more. For that reason, the public sale is called a “second price public sale”.
87 An instance may also help illustrate this form of public sale. If advertiser A bids 3 € CPC, advertiser B bids 4 € CPC and advertiser C bids 2 € CPC, the winning advertiser B can pay 3. 01 € CPC. The AdWords public sale is over, but that does not always mean that advertiser B will get to serve its ad on the page visited by the user. The reason is that AdWords isn’t the only platform connected to AdX. There are other ad networks/DSPs hooked up to AdX, which, similar to AdWords, invite advertisers that have an account with them and run their own auctions usually second price and come up with their highest bid.
These ad networks/DSPs will now compete with AdWords for the ad impression in a new auction, organized by Google’s AdX. Publishers should forever benefit from such festival, since it is feasible that a competing ad community/DSP might offer the next bid than AdWords. In this auction, bids are expressed on a different basis, namely on a Cost Per Mille “CPM” basis i. e. the fee paid for each thousand impressions of the similar ad, hence the name. This auction is again second price.
For instance, if DSP1 bids 10 € CPM, DSP2 bids 11 € CPM and AdWords bids 12 € CPM, the winner, AdWords, gets to pay 11. 01 € CPM, i. e. just a little more than the second one maximum bidder. The second auction hosted by AdX is over.
Under the so called waterfall system, publishers using DFP could connect the latter with a few exchanges, so that they would avoid any risk of relying on just one trade and ending up with unsold stock as one ad trade might not value the impact and not bid. However, the a number of exchanges does not compete with each other. Instead, they would be ranked in keeping with their regular historic yield i. e. how much cash they’d made on common for the writer in the past in a waterfall like sequence. 89 Each time an ad impression was accessible, DFP would give priority to directly sold ads.
90 Once there were no more eligible without delay sold ads, a bid request for the ad impact can be sent to the trade ranked first in the waterfall: If the 1st exchange bought the ad impression, the exchanges lower in the waterfall does not be invited to bid and the ad preference would be completed. If the 1st trade did not buy the ad impact, the latter would be provided on the market to the trade immediately below in the waterfall at a discount. That technique would continue until the ad influence could be ultimately sold to an exchange. The “deeper” the ad impact would cascade into the waterfall, the lower the associated fee at which it was provided for sale. If no one expressed attention in buying the ad effect, the “fallback” option for the writer can be to fill the ad space with an ad advertising its own business so called “in house” ad. In 2014, Google launched a feature in DFP called dynamic allocation,91 which enabled AdX to act in a “dynamic” manner and disrespect the waterfall.
As described above, publishers using DFP would assign each ad exchange an expected CPM price based on ancient data, thus rating ad exchanges in a waterfall, in accordance with which they could be called to bid if an impact was accessible. However, after the introduction of dynamic allocation, when an ad impression was accessible, DFP would select the highest envisioned CPM price of an ad exchange in the waterfall after which send that estimated price to Google’s AdX. AdX would then run a real time public sale to see if it could offer a slightly higher price, e. g. 1 cent more. 92 If it may, then AdX would get to serve the ad.
Therefore, dynamic allocation granted AdX two distinct benefits over other ad exchanges: First, AdX could run a real time auction for each ad impact, while other ad exchanges were “stuck” with their expected prices, never getting the chance to submit a real time bid the “real time demand” virtue. That means that DFP sheltered AdX from real time competition from other exchanges, which can thus allow AdX to buy impressions at artificially low prices. Second, AdX would use the highest estimated price of the ad exchange at the top of the waterfall as the cost floor for its own public sale. That means that in observe AdX could always beat any trade in the waterfall, provided it may submit a a little higher bid. AdX had always the “last look” on the ad impression, and that’s the reason why industry commentators referred to this virtue as the “last look” advantage.
93An industry commentator summarizes the considerations brought on by dynamic allocation as follows: Google made the demonstrate panorama less aggressive by launching Dynamic Allocation in 2014, which enabled its exchange AdX to insert a real time bid into DFP for every impact. Thus AdX could enter accurate pricing while other partners were stuck with their standard tags, even though their bidders could potentially cite a better price. Theoretically, Dynamic Allocation could enable AdX bidders to pay less for impressions than other partners could be willing to, for this reason starving the publisher of earnings. This apparently unfair setup spurred the adoption of header bidding. 94Header bidding is just another kind of auction.
There are, however it, key variations among header bidding and the third auction run by DFP. First, header bidding happens before the user’s browser asks DFP to serve the ad hence it also is called a pre public sale. Second, the public sale is run by the browser of the user, not DFP. It is the browser, not DFP, that acts as the auctioneer, inviting interested parties to bid for the ad impression. This form of header bidding is called “client side” header bidding.
Third, and most significantly, the browser invites demand partners e. g. ad exchanges/SSPs to submit bids for the ad influence concurrently in a unified auction. There is no waterfall, i. e.
demand partners will not be prioritized. Although Ad Exchange still had the “last look” and will outbid the successful bid from the header bidding public sale, header bidding however allowed publishers to have access to real time demand from quite a lot of ad exchanges and thus get a correct insight of their inventory’s value. 97 AdX could no longer rely upon the anticipated bid from other exchanges which can be much less than the particular, real time bid to win the public sale. Header bidding thus uncovered AdX to a couple degree of competition from other exchanges in that it undermined AdX “real time advantage”. Header bidding also supplied advantages for buyers since they can bid for each ad impression – even premium stock – and not just for the influence that had “cascaded” down the waterfall.
98It is thus not shocking that publishers imposing header bidding saw a significant boom in their ad sales, occasionally up to 60%,99 encouraging its common adoption and the emergence of application answers offered by plenty of corporations that help publishers prepare their demand companions in header bidding called “wrappers”. 100 Google, on the other hand, was less passionate about this development. 101 An AppNexus director for example stated that “Google sees this as a large threat to their dominance, and has no attention in having this adopted by the IAB ”. 102A ability draw back of header bidding, nevertheless it, is that it may boom page latency, i. e.
the website of the publisher may take longer to load. In order to address page latency, some publishers turned to server side header bidding,103 where the pre public sale happens in a remote server instead of the user’s browser. While page loading time is more desirable, publishers generate lower earnings, partly as a result of for technical motives, buyers have less data about the user and thus do not bid as high as they in a different way would. 104 Furthermore, because the auction occurs in a server owned by 0,33 party e. g.
Amazon, there is a loss of transparency. Exchange Bidding is Google’s answer to header bidding,105 announced as a function of DFP in 2016 and made commonly available for publishers in 2018. 106 Exchange Bidding allows publishers using DFP to attach third party exchanges so called “yield companions” to Ad Exchange via a server to server connection. 107 Each time an ad impression is accessible for sale, all competing exchanges submit their bids simultaneously in a unified public sale hosted by DFP. This is the third public sale that was described in our instance above.
Ad Exchange has no longer the “last look” virtue and faces real time pageant from these hooked up exchanges. 108 There is a caveat however: the removal of the last look advantage concerns only the third party exchanges that have accredited to integrate with AdX during the server to server connection. Thus, AdX retains its advantage for any ad trade that does not participate in Exchange Bidding. Even though Exchange Bidding is arguably Google’s effort to persuade publishers that there is now not a need to use header bidding, commentators are involved concerning the transparency of Google’s answer, and express fears that Google could still favour its AdX in subtle ways. For occasion, DFP may pass unique information to AdX concerning the audience that could be uncovered to the ad, allowing it to solicit higher bids from advertisers than connected exchanges with “less” perception. 109 And in any event, AdX keeps its virtue over exchanges refusing to take part in Exchange Bidding e.
g. out of distrust. Thus, publishers should opt to engage in client side header bidding, which regardless of its latency problems, is transparent and promises high yields. What may be problematic, but it, is if Google tried to evade publishers from attractive in client side header bidding. However, a couple of features of the AMP standard make it applicable to our analysis of reveal advertising. First, AMP is designed in such a way that it is incompatible with traditional header bidding, i.
e. client side header bidding. 111 Publishers of AMP compliant websites should engage in server side header bidding, but it is dubious even if the latter can be viewed as an alternative to client side header bidding. The reason is that server side header bidding is characterized by a lack of transparency. As an industry expert notes: Server side header bidding calls for teamwork in a non transparent environment what occurs on the server is invisible to both the writer and the buyers. It’s feasible that auctions can be conducted in a way where one demand associate gets preference or a final look.
Or data can be leaked or hidden fees be taken. 112There is, nonetheless it, a further difficulty posed by the AMP elementary, which is that it is making it harder for publishers to compete with Google in offering focused on services to advertisers. As has been noted, the upward push of programmatic commercials resulted in advertisers valuing user data and the targeting probabilities they unlock greater than ever. Some publishers with wide readerships, comparable to most advantageous newspapers, have tried to build their own unique proprietary datasets about their viewers so as to offer focused on services without delay to advertisers, removing the need to resort to intermediaries akin to Google. 114 However, such efforts are considerably undermined in relation to AMP. When the user visits an AMP compliant page, the content of the page is fetched not from the publisher’s servers, but from Google’s servers, where it’s been “cached”.
The result’s that Google collects large troves of data linked to the users’ interactions with the publisher’s website. Google shares such data with the writer in a format that forestalls cross site matching, i. e. the writer cannot match users vacationing various internet sites which belong to the similar publisher. Publishers are thus unable to collect the necessary data to create longitudinal user profiles they deserve to offer appealing concentrated on services. Of course, Google could claim that publishers do not have to agree to the AMP simple.
But, in actuality, publishers, especially news content providers, must be AMP compliant, as otherwise they would lose the Internet traffic generated by Google searches. The reason is that Google only allows AMP compliant webpages designated as such with a lightning bolt icon and an “AMP” label to seem in its News Carousel. 115 Moreover, mobile websites that don’t comply with the AMP basic will figure lower on Google SERPs, since as of July 2018 page speed has become “a major ranking factor for mobile searches”. 116 Compliance with the AMP essential is thus with ease obligatory for publishers given the significance of Google search as a source of referrals. For occasion, data indicates that more than half 53% of all referral traffic that virtual publishers accept comes from Google search.
117What prompts us to look into potential exploitative practices are the a variety of considerations that the classic actors, i. e. publishers and advertisers, have expressed concerning the opaqueness of the sector and the charges charged by the operators that intermediate among them. 118 Publishers and advertisers have restricted visibility into the specific functioning of the show commercials ecosystem. 119 There are widespread concerns in the ads and publishing industry concerning this lack of transparency and the so called “ad tech tax”, i.
e. the charges utilized by loads of middlemen among publishers and advertisers. 120For occasion, IAB present in a 2014 report that ad tech corporations cumulatively seize 55% of programmatic sales, the rest 45% going to publishers. 121 WARC has envisioned that in 2017 the “ad tech tax” accounted for 55% of all programmatic spend, leaving below 36% for publishers, if ad fraud is taken into account. 122 As noted above, The Guardian discovered in 2016 that during a worst case scenario ad tech intermediaries could extract up to 70% of programmatic sales. The Guardian filed a lawsuit in 2017 in opposition t ad trade Rubicon Project over alleged undisclosed buyer fees,123 however the events settled.
124 The Select Committee on Communications appointed by the House of Lords, noted in its 2018 Report that based on a U. S. study, publishers end up receiving only 29% of programmatic revenues. 125 At the similar time, commentators comply with that it is sort of most unlikely to decide accurately the charges charged by ad exchanges,126 while DSPs it seems that charge hidden fees. 127 It is thus not stunning that transparency is the #1 worry for retailers in 2018.
128Besides the charges which are charged by intermediaries at every corner of the ad tech stack, industry commentators have also diagnosed a particular feature of programmatic commercials that may be used by intermediaries to interact in arbitrage and thus take advantage of publishers and advertisers, which is the lifestyles of consecutive second price auctions. 129 The issue might at first glance seem irrelevant from a competition law attitude. However, if a dominant company were found to have interaction in such a tradition, it can be considered as a variety of exploitation in breach of Article 102a TFEU. In any event, it is worth exploring how there may be any arbitrage from the lifestyles of consecutive auctions. Assume, for example, that there are three advertisers in the AdWords public sale: Advertiser 1 bids 10 € CPM, Advertiser 2 bids 12 € CPM winner, and Advertiser 3 bids 11 € CPM. 133 Since the AdWords public sale is second price, the successful Advertiser 2 can be charged 11.
01 € CPM. According to Google’s aid manager website, if Google Ads wins the public sale, the advertisers in the successful ad unit will pay no greater than what is needed to rank higher than the following advertiser, on a CPC basis, when a user clicks on the ad or completes an alternative valid event in reference to the ad. 134However, it’s not necessarily the amount of cash that the writer will accept. The reason is that there’s an extra second price public sale, organized by AdX, where other DSPs/ad networks compete with AdWords. For instance, DSP 1 bids 7 € CPM, DSP 2 bids 5 € CPM, and AdWords bids 11. 01 € CPM winner.
As the Ad Exchange public sale is again second price, the writer may be paid a little above the second one maximum bid, i. e. 7. 01 € CPM. 135 In Google’s assist supervisor website, it is stated that “he writer could be paid the maximum of the second highest bid value in the Ad Exchange public sale or the minimum CPM”.
136 That would allow the intermediary, during this case Google, to extract the difference between what the advertiser was charged and what the writer gets at the tip of the chain. Although such a practice could only be proved by analysing bidding data, it may be one of the reasons why publishers only capture a fraction of the costs paid by advertisers to purchase their ad inventory. The risk of exploitation would for sure be less likely to arise if the ad tech market was aggressive. In a aggressive market, Google would be dissuaded from attractive in any form of exploitative behavior, as publishers and/or advertisers could discipline AdX by not trading through it. Moreover, excessive competition between AdWords and competing ad networks/DSPs within the auction arranged by AdX would cut down the margin accessible for arbitrage. The example above illustrates that the chance for arbitrage is greater when the first auction yields a comparatively high bid137 and the second one highest bid in the second one public sale is materially lower.
138As illustrated by the Google Shopping resolution of the Commission, festival complications may arise when a firm that owns a dominant platform Google Search competes on a downstream market comparison shopping facilities with other firms that need to have access to the dominant platform to supply their facilities. 140 In that determination, the Commission found that Google abused its dominant place by systematically giving well known placement to its own comparison shopping carrier in its search effects, while demoting rival comparison buying groceries amenities in these effects. The abusive behavior diagnosed by the Commission has been labelled as “self preferencing” in that Google used its dominant platform to provide a aggressive advantage to its comparison shopping facilities over rival services. A related fear seems to have led the Commission and the German Competition Authority to these days launch a preliminary investigation of Amazon’s e commerce platform. 141 While little is famous about that investigation, it sort of feels to be focused on Amazon’s dual role as a competitor, but in addition host, to third party retailers, which sell goods on Amazon’s internet sites.
Because of this dual role Amazon has access to effective data on the provision, prices, return rates and acclaim for opponents’ products, which it could potentially use to stimulate its own retail activities at the rate of third party dealers on its marketplace. The fact that Google is, as we have seen above, both the organizer of the ultimate public sale in DFP and participating in the public sale in the sort of AdX gives rise to similar kinds of considerations as those diagnosed in the Google Shopping resolution and the Amazon preliminary investigation. This problem is not new. Already in the context of the Google / DoubleClick merger in 2008, stakeholders had expressed issues that Google could use DFP to favour its own intermediation services, e. g.
by tweaking the auction mechanism in favour of AdSense, Google’s ad network, thus depriving competing ad networks and exchanges from the important scale and liquidity they deserve to be sustainable. 142 At the time, the Commission rejected these arguments, repeating that Google would have the motivation to act neutrally vis à vis competing intermediaries, as a loss of neutrality can cause clients switching. The difficulty is that meanwhile Google has in large part monopolized the ad tech value chain, and that the type of constraints identified by the Commission in 2008 not exist. In addition to this difficulty, concerns have been expressed that Google might use the data gathered by DFP to favour AdX. As expressed by an industry observer: Google depended on the informational virtue DFP + AdX integration to “cherry pick” stock in mysterious, but decidedly underhanded ways. According to an ad tech government who wished to stay nameless, “AdX always won the impact if the user happened to be at the top of funnel stage in a purchase journey, basically stealing attributions from other exchanges.
On paper, it went on to show advertisers that DBM DoubleClick Bid Manager with AdX inventory gave them better effects than any other systems”. 143These observations remain relevant even after the creation of Exchange Bidding. The reality is that DFP possesses vast amounts of historic data regarding the bids submitted for selected impressions by competing ad exchanges and the cost at which the impression is ultimately sold, since there could be millions of impressions being sold every day and DFP is absolutely by far the most usual ad server solution. The informational advantage could thus be still current, and it isn’t possible to video display even if AdX may use such old data collected by DFP to calculate the acceptable bid to win the auction. Commentators have also taken issue with the fact that Exchange Bidding lacks transparency, a reason why competing ad exchanges are reluctant to take part in Exchange Bidding.
An author notes that: Demand partners often take pause at leaping into an S2S connection controlled by a person else, particularly when that “a person else” is a competitor. In managing the server side connection, Google indirectly makes a decision what data goes into EBDA . There are issues in S2S related to ID syncing among buy and sell sides, and from publisher to writer. Google might find an virtue for itself in those ID issues, not only because it’s managing the server side connection, but as a result of of its exceptional scale. So, EBDA still comes out looking anything like a black box, unified public sale or no.
144Demand partners often take pause at leaping into an S2S connection controlled by someone else, especially when that “a person else” is a competitor. In handling the server side connection, Google indirectly decides what data goes into EBDA . There are issues in S2S related to ID syncing between buy and sell sides, and from publisher to writer. Google might find an advantage for itself in those ID issues, not only because it’s managing the server side connection, but as a result of of its exceptional scale. So, EBDA still comes out looking anything like a black box, unified auction or no. 144Interestingly, Google has abandoned any attempt to give the influence that DFP might deal with AdX on an arm’s length basis when it introduced, in July 2018, the combination of AdX and DFP into Google Ad Manager, providing a “truly unified platform”.
145 This led an industry commentator to comply with that: my guess is that the top goal is to try and wean publishers off of header integrations and get them addicted to EBDA demand. It seems funny to me that there’s now not even a pretense of separation between ad server and SSP/trade. The name change re emphasizes that Google will leverage its near monopolistic manage of the writer ad server market to shoo away other demand sources—whether or not that’s good for the writer or the advertiser. It’s an alternate effort to squeeze out pageant and keep publishers and advertisers sucking at the Google teat. 146As mentioned above, in response to DFP’s dynamic allocation and so as to stimulate true competition between ad exchanges, publishers resorted to header bidding, which uncovered AdX to real time pageant from hooked up exchanges.
Google replied to header bidding by launching Exchange Bidding, which allows all attached exchanges to compete in a unified auction hosted by DFP. However, it also seems that Google undermined header bidding in the course of the advancement of AMP. The reason is that, as noted above, client side header bidding is not feasible in AMP compliant websites for technical causes. Moreover, by requiring that each one AMP pages are loaded on its servers, Google doesn’t only allow itself to assemble all the data associated with the users’ interactions with writer, but it also makes it harder for these publishers to procure access to this data by limiting access to it. Google’s technique is not unlike the conduct at stake in the Android resolution where the Commission considered that Google had illegally tied its Search and browser Chrome apps to its app store the Play Store, thus without difficulty coercing Android device makers to preload the Search and Chrome apps on their devices. 147 In that case, while Android device makers were theoretically not bound to preload Google’s suite of functions to develop an Android device, a refusal to do so would have made these devices commercially unsalable as they would were disadvantaged of the Play Store, which is a “should have” for all Android users.
148 In its decision, the Commission found that Google’s behavior breached Article 102 TFEU and condemned Google to a significant fine. Online display advertising is a sector of important importance to both advertisers and publishers. But for their reveal ads sales, even the world’s most well known newspapers does not be commercially viable. While online display ads was initially not very alternative from its offline equivalent as most inventory was sold via bilateral negotiations among publishers and advertisers, the rise of programmatic ads has had profound implications on the industry. Programmatic advertisements has been a source of opportunities for advertisers and publishers, but the fees charged by intermediaries are opaque, hence amounting to what’s perceived as ad tech tax.
While ad tech markets are populated by an expansion of actors, Google appears to hold a dominant place on a couple of such markets, and a number of of its conducts raise exploitative and exclusionary concerns potentially in breach of Article 102 TFEU. The French and German festival authorities are searching closely at the web demonstrate ads sector and, given their investigative powers, they should be able to assemble the information, adding bidding data, required to further explore the festival issues present in the sector. Other countrywide competition gurus may follow suit under the force introduced by advertisers and publishers. If a number of investigations are initiated at the Member State level, it may be ultimately alluring for the European Commission to intervene to keep away from the adoption of incompatible selections and remedies. The user types in its browser the URL of the publisher webpage e.
g. the webpage of the Wall Street Journal. In our instance we assume that the user is John, lives in France, is 30 years old and is attracted to cars. The browser calls the content server of the writer website. The content material server sends the content of the website, which has a pre defined empty space to be crammed with an ad.
The web server “tells” the browser to name DFP for the ad that can fill the available space. In case the publisher has resorted to header bidding, a pre public sale will ensue before the page starts to load and before DFP is named by the browser. In client side header bidding, the browser contacts at once all the demand companions ad exchanges/SSPs the writer has configured and runs a simultaneous auction. In server side header bidding, the browser contacts only one demand accomplice, which in turn contacts any other demand partners and runs the public sale on its server. In both cases, the successful bid could be sent by the browser to DFP where it could be matched with a remnant line item. The browser contacts DFP sending an ad tag.
An ad tag is a snippet of code normally HTML contained in the writer website that contains data about the ad space it truly is up for sale and the user that will be uncovered to the ad. DFP examines the acquired information and finds the road items which are compatible. DFP ranks line items according to bound criteria. Guaranteed line items rank ahead of remnant line items. 149 DFP selects the highest ranking guaranteed line item and the maximum ranking remnant line item which can be the profitable bid from header bidding, if such pre auction has taken place. DFP assigns the assured line item a cost called “transient CPM” that does not necessarily coincide with the particular CPM of the road item.
That process opens the assured line item to competition from AdX bids, so as to maximize writer revenues. 150 The reason that DFP assign a temporary CPM is to be sure that the birth of the assured line items agreed among the publisher and the advertiser will not be compromised. DFP assigns a short lived CPM that reflects the development of assured line item’s birth: if the assured line item is behind schedule, the next brief CPM is assigned to enhance its probability of profitable and being delivered. If the guaranteed line item is close to reaching its beginning goal, the temporary CPM assigned may be lower. DFP sends bid request together with data derived from the ad tag to AdX to solicit bid responses that will compete with the guaranteed line item and the remnant line item chose. The higher of the transient CPM of the highest guaranteed line item and of the CPM of the top remnant line item is set as a cost floor on the auction run by AdX.
AdX has thus “last look”, i. e. it can beat any line item if it solicits a slightly higher bid. If the publisher has enabled Exchange Bidding, the publisher may connect AdX with third party ad exchanges called “yield partners” that will compete with AdX in a unified public sale. In such a case, Google’s AdX sends the bid request to competing ad exchanges through a “server to server” connection.
AdX has no “last look” advantage vis à vis these hooked up third party exchanges, but keeps it vis à vis other exchanges. Google’s own AdX and third party ad exchanges run auctions to determine the bid each of them will submit for the particular ad influence. This process the truth is carries a number of auctions. For instance, AdX is connected to a few DSPs and ad networks, adding AdWords. Each DSP/ad network will run its own public sale to choose the bid it will submit to the public sale organized by AdX. The data concerning the ad slot and the user help DSPs gauge how much they are willing to bid.
For occasion, a DSP that manages the crusade of a car brand focused on young men living in France will decide to bid higher. Once DSPs have submitted their bids, AdX runs a second price auction and selects the DSP with the maximum bid. Since the public sale is second price, the winning bidder can pay not what it basically bids, but just a bit more e. g. 1 cent than the second maximum bidder. For instance, DSP1 bids 2.
10 € CPM, DSP2 bids 2. 50 € CPM and DSP3 bids 1. 90 € CPM. The profitable bidder, i. e.
DSP2 will the fact is pay 2. 11 € CPM, not 2. 50 € CPM. Once AdX and the competing ad exchanges have run their very own auctions, each of them submits its maximum bid. In our example, AdX will submit a bid for 2. 11 € CPM.
DFP then hosts a unified public sale, where the bids from competing ad exchanges and AdX compete with the higher of the CPM of the highest remnant line item and the transient CPM of the highest guaranteed line item. The highest bidder wins and gets to serve the ad. Once the maximum bidder is determined, DFP contacts the browser and tells it to fetch the artistic content material that will fill the ad space from the ad server of the advertiser that won. The browser calls the ad server of the successful advertiser and serves the artistic content material on the webpage of the writer. In case the writer has resorted to header bidding, a pre public sale will occur before the page starts to load and before DFP is named by the browser.
In client side header bidding, the browser contacts without delay all the demand partners ad exchanges/SSPs the writer has configured and runs a simultaneous public sale. In server side header bidding, the browser contacts only one demand companion, which in turn contacts the other demand partners and runs the auction on its server. In both cases, the winning bid can be sent by the browser to DFP where it will be matched with a remnant line item. DFP assigns the assured line item a value called “transient CPM” that does not always coincide with the particular CPM of the road item. That procedure opens the assured line item to competition from AdX bids, so as to maximize writer revenues.
150 The reason that DFP assign a temporary CPM is to be sure that the start of the assured line items agreed among the writer and the advertiser will not be compromised. DFP assigns a brief CPM that reflects the progress of assured line item’s beginning: if the assured line item is behind schedule, the next transient CPM is assigned to enhance its probability of winning and being delivered. If the guaranteed line item is close to achieving its delivery goal, the transient CPM assigned could be lower. This procedure the fact is carries multiple auctions. For example, AdX is hooked up to a few DSPs and ad networks, including AdWords. Each DSP/ad network will run its own auction to choose the bid it will submit to the auction arranged by AdX.
The data regarding the ad slot and the user help DSPs gauge how much they are inclined to bid. For instance, a DSP that manages the campaign of a car manufacturer focused on young men living in France will decide to bid higher. Once DSPs have submitted their bids, AdX runs a second price auction and selects the DSP with the highest bid. Since the auction is second price, the profitable bidder will pay not what it truly bids, but just a little more e. g.
1 cent than the second one maximum bidder. For example, DSP1 bids 2. 10 € CPM, DSP2 bids 2. 50 € CPM and DSP3 bids 1. 90 € CPM. The successful bidder, i.
e. DSP2 will as a matter of fact pay 2. 11 € CPM, not 2. 50 € CPM. 7 Targeting is an advertisements technique that contains customizing promotional content material added to users on the idea of standards akin to their shopping behaviour or pursuits behavioural focused on, the theme and content material of an internet site contextual targeting, the geographical location of an individual geographical concentrated on, their social, demographic and economic features, reminiscent of age, gender, income, etc. sociodemographic targeting, or the time, day or week time focused on.
The definition is derived from the Opinion no. 18 A 03 of 6 March 2018 on data processing in the web ads sector accessible in English at 121. 9 A. Bruell, ‘Inside the Hidden Costs of Programmatic’ AdAge 14 September 2015 ; S. Gatz, ‘Publishers and the Hidden “Ad Tech Tax”’ AdExchanger 1 April 2016 ; N.
Neumann, ‘Ad Tech Transparency and the Question of Market Manipulation’ AdExchanger 1 May 2017 ; M. Zawadzinski, ‘Why a Lack of Transparency is Killing the Potential of Programmatic Buying’ The Clearcode Blog ; M. Sweeney, ‘Transparency in Ad Tech: The Problems, Fallouts and Solutions’ The Clearcode Blog ; ‘Quality, Transparency of Inventory Top Programmatic Buying Fears’ eMarketer 1 November 2016 . 21 For a wonderful advent to the programmatic revolution, see M. Sweeney, ‘The Colorful History of Advertising Technology in Just 63 Slides’ The Clearcode Blog, 12 May 2015 ; M. Sweeney, ‘How Real time Bidding RTB Changed Online Display Advertising’ The Clearcode Blog, 8 January 2015 ; I.
Simpson, ‘Real time Bidding RTB and Programmatic: One and the Same?’ The Clearcode Blog, 13 April 2016 ; M. Zawadziński, ‘Understanding RTB, Programmatic Direct and Private Marketplace’ The Clearcode Blog, 13 August 2018 . 26 An industry commentator notes in a contemporary article that ‘f you count the third party pixels operating on any publisher’s site, you’ll automatically see how confusing and convoluted the once simple technique of putting an ad on a online page has become’, see R. Lala, ‘Is It Too Late For Publishers To Take Back Control?’ AdExchanger 11 July 2018 . See also Bannister, ‘Has Sell side Ad Tech Become too Complex?’ AdExchanger 16 March 2018 noting that ‘omplexity is here to remain The question for plenty of publishers is whether they can navigate this minefield of complexity and find companions that help them simplify things and maximize their revenue concurrently’; I. Simpson, ‘Complex Relationships in Digital Advertising’ The Clearcode Blog, 14 April 2016 .
40 Decision of 18 February 2010, COMP/M. 5727 Microsoft/Yahoo!Search Business, para 75; Decision of 18 February 2010, COMP/M. 5727 Microsoft/Yahoo!Search Business, para 75; Decision of 4 September 2012, COMP/M. 6314, Telefonica UK/Vodafone UK/Everything Everywhere/JV, para 151; Decision of 4 September 2012, COMP/M. 6314, Telefonica UK/Vodafone UK/Everything Everywhere/JV, para 151; Decision of 30 October 2014, COMP/M.
7217, Facebook/WhatsApp, para 76; Decision of 6 December 2016, COMP/M. 8124 Microsoft/LinkedIn para 161; Decision of 21 December 2016, COMP/M. 8180 Verizon/Yahoo, para 25. 49 Id. 5: ‘There are two types of ad intermediation merchandise: ad networks and ad exchanges. Ad networks and ad exchanges are alike in that they both mixture ads stock.
Ad networks are intermediaries that aggregate or buy commercials stock from a group of websites and sell this stock to advertisers or ad businesses, taking a share of the earnings from each sale. Ad exchanges differ in that they aggregate stock by featuring structures for advertisers and publishers to list and bid for inventory. The proof shows that the market by which ad networks and ad exchanges compete is relatively nascent, dynamic, and highly fragmented. ’58 In its Guidance Paper on Article 102 TFEU, the Commission notes that “he assessment of dominance will keep in mind the aggressive structure of the market, and in particular the following elements: constraints imposed by the existing gives from, and the place in the marketplace of, actual competitors the market position of the dominant conducting and its rivals, constraints imposed by the credible threat of future growth by actual rivals or entry by capacity opponents enlargement and entry, constraints imposed by the bargaining strength of the conducting’s customers countervailing buyer power. ” See Communication from the Commission – Guidance on the Commission’s enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct by dominant undertakings Text with EEA relevance 2009/C 45/02 para 12. 89 For a glorious description of the waterfall process, see M.
Zawadzinski, ‘What is Waterfalling and How Does It Work?’ The Clearcode Blog, 1 September 2016 ; P. Bannister, ‘As Header Bidding Rises, It’s More Important Than Ever to Understand The Waterfall’ AdExchanger 10 February 2016 . Publishers would set the waterfall within DFP by atmosphere remnant line items for the numerous ad exchanges and assigning them an envisioned bid based on their normal ancient yield. It is essential to notice that this does not come with any connection between such exchanges and AdX. Such connection occurs only in Exchange Bidding and provided that the competing trade has permitted to attach to AdX. 92 G.
Sloane, ‘WTF is Dynamic Allocation?’ Digiday 14 April 2016 ; noting that based on Alex Magnin, CRO of Thought Catalog, a new media writer ‘Dynamic allocation allowed Google’s exchange to cherry pick one of the best ad impressions as they came during the Google owned ad server, DFP’. ’ See also P. Dinodia, ‘Everything You Need to Know About Dynamic Allocation’ adpushup blog, 17 November 2017 ; S. Sluis, ‘The End Of Header Bidding?Google Opens Up Dynamic Allocation to Outside Demand’ AdExchanger, 13 April 2016 . 96 For an excellent description of header bidding, see Maciej Zawadziński, ‘What is Header Bidding and How Does it Work?’ The Clearcode Blog, 2 August 2016 ; M. Zawadzinski, ‘What’s the Difference Between Waterfall Auctions and Header Bidding?’ The Clearcode Blog, 22 September 2016 ; Ratko Vidakovic, ‘The Beginner’s Guide To Header Bidding’ AdProfs 30 March 2017 ;109 L.
O’Reily, ‘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”’ Business Insider 2 August 2016 ; J. Hercher, ‘Google’s Answer to Header Bidding Is Now Generally Available’, supra note 108 noting that according to an AppNexus’s director, Exchange Bidding trades on the transparency provided by header bidding integrations while maintaining publishers ‘locked in the AdX black box’; S. Sluis, ‘Google Removes Its ‘Last Look Auction Advantage’, supra note 95 noting that “Google will retain one additional virtue in the public sale: It knows more concerning the user than it passes on to any other exchanges” and that “the shift to server side answers which make auctions run faster, also threaten to make the auctions lose transparency again and reflect the last look advantage”; B. LaRue, ‘Last Stand for Google’s “Last Look”: What’s Next?’ Admonsters 31 March 2017 noting that Exchange Bidding ‘still comes out hunting something like a black box, unified public sale or no’. 115 S. Whang, ‘Google News is Getting Its Own Carousel of AMP Stories, and other AMP Features in the Works’ 20 April 2016 noting that ‘he Google News headlines carousel will contain only AMP articles.
’ The News Carousel is a box acting at the top of Google’s search leads to mobile, that displays news articles relevant to the user’s query. Users can swipe left or right to navigate during the articles in the with no need to scroll down on the page to view search results. 118 The CMO of PandG, one of the world’s maximum spending advertisers, famously said in a 2017 IAB assembly that ‘we’re all wasting way too much time and money on a media supply chain with poor requirements adoption, too many avid gamers grading their own homework, too many hidden touches, and too many holes to permit criminals to rip us off We have a media supply chain this is murky at best and fraudulent at worst. We serve ads to consumers via a non obvious media supply chain with spotty compliance to common requirements, unreliable measurement, hidden rebates and new inventions like bot and methbot fraud’. See L. Handley, ‘Procter and Gamble Chief Marketer Slams “Crappy Media Supply Chain”, Urges Marketers to Act’ CNBC 31 January 2017 .
137 That doesn’t look like a problem, given AdWords’ prominence as an ad network. It also is possible that Google takes virtue of its prominence on search advertising and uses a variety of ‘status quo bias’ to artificially create more competition among advertisers on exhibit advertisements, thus foremost to higher prices. AdWords is also the ‘gateway’ for search advertising on Google’s SERPs. However, an advertiser bidding for a campaign in AdWords is by default and unless she opts out bidding for both search ads in Google’s SERPs and for reveal ads in the Google Display Network. That drives up demand for Google Display Network, even when the advertiser doesn’t comprehend it. See D.
Pratt, ‘7 Default Settings in AdWords that Lower Your ROI’ AdHawk 18 June 2018 . 138 Again, that could be the case because e. g. AdWords has more data concerning the user compared to other ad networks/DSPs. According to Google, when AdWords is used to buy stock on Ad Exchange, there is minimal cookie matching loss from Ad Exchange to AdWords, to the effect that ‘there is a better likelihood AdWords will find impressions that meet the concentrated on standards of advertisers, creating more beneficial auction pressure and demand for the writer’s stock. ’ See .
Moreover, it has been prompt that pageant in the auction organized by AdX is way weaker than one would assume: see C. Cummings, ‘Google’s Busted Auctions’ PubNation Blog, 22 June 2016 noting that on average there have been only six bids per effect on Ad Exchange and that the distance among the profitable bid and the second one highest bid could be ‘huge’, up to 70% off the winning bid. 139 C. Ballentine, ‘Google Facebook Dominance Hurts Ad Tech Firms, Speeding Consolidation’ The New York Times 12 August 2018 See also C. Ballentine, ‘Investment in Ad Tech Grows Increasingly Scarce, With Forrester Predicting a 75% Drop in Venture Capital’ Adweek 7 November 2018 noting that ‘art of the fear among investors is the consolidation of ad spend on structures such as Facebook and Google.
Jay Friedman, president of Goodway Group, defined to Adweek that the historic opaque business models of many ad tech agencies have caused media buyers to be more prudent.