Daily Mail Australia Level 12, 207 Kent Street, Sydney 2000

Daily Mail Australia

Level 12, 207 Kent Street,

Sydney 2000

dailymail.com.au

25 February 2021

Daily Mail Australia’s response to the Australian Competition & Consumer Commission’s

Interim Report for its Digital advertising services inquiry

Daily Mail Australia forms part of one of the world’s largest English-language group of newspaper

websites, with more than 226 million global unique browsers.

Daily Mail Australia has a loyal

readership of 8.8 million monthly unique visitors, with an average time of 24 minutes spent per

person.

Our success is down to editorial excellence, dynamic and engaging content, and a picture-

led, easily navigable format available on any device.

On 10 February 2020, the Australian Government directed the Australian Competition & Consumer

Commission (“ACCC”) to conduct an inquiry into markets for the supply of digital advertising

technology services and digital advertising agency services (the “Ad Tech Inquiry”). On 10 March

2020, the ACCC published an Issues Paper inviting interested parties to submit views and

information (the “Issues Paper”),

to which Daily Mail Australia duly responded.

On 28 January 2021, the ACCC released an Interim Report within the context of its Ad Tech

Inquiry (the “Interim Report”),

finding that a lack of competition and transparency in the ad tech

supply chain is impacting publishers, advertisers, and consumers, and invited stakeholder feedback.

Daily Mail Australia commends the ACCC for its excellent work in navigating the complex

ecosystem of ad tech services and fully supports the findings in the Interim Report. The time has

now come for the ACCC to take swift action to ensure that the ad tech ecosystem, which has been

largely monopolized by Google, is truly competitive and transparent to the benefit of publishers,

advertisers and ultimately consumers.

Daily Mail Australia responds hereunder to the various questions raised by the ACCC throughout

its Interim Report. For the sake of convenience and completeness, we have included all the

Adobe Analytics, Jan 2021, Global. In March 2020 we reached a record-high 272 million global unique browsers.

https://www.nielsen.com/au/en/press-releases/2020/abc-news-websites-ranks-no-1/

Australian Competition & Consumer Commission, Ad Tech Inquiry, Issues Paper, 10 March 2020, available at

https://www.accc.gov.au/system/files/Ad%20tech%20inquiry%20-%20issues%20paper.pdf

Daily Mail Australia’s response to the Ad Tech Inquiry Issues Paper of the Australian Competition & Consumer

Commission (ACCC), 2 June 2020, non-confidential version available at

https://www.accc.gov.au/system/files/Daily%20Mail%20Australia%20%282%20June%202020%29.pdf

Australian Competition & Consumer Commission, Digital advertising services inquiry, Interim Report, December

2020, available at

https://www.accc.gov.au/system/files/Digital%20Advertising%20Services%20Inquiry%20-

%20Interim%20report.pdf.

questions as formulated by the ACCC (in bold), categorizing them according to the Chapters of the

Interim Report, but respond only to questions concerning publishers. We have not included

questions from Chapter 7 as these concern ad agencies.

Chapter 1 – The supply of digital display advertising in Australia

The ACCC invites further stakeholder views regarding the key points in this chapter

including, in particular:

• the extent to which video and non-video display advertising are substitutable

While marketers are best placed to respond to this question, our understanding is generally aligned

with the preliminary view of the ACCC. We as a publisher see video ads in two main formats: in-

stream video ads and out-stream video ads.

In-stream video ads play in line with content and command a higher CPM and often perform better

than out-stream ads, particularly if the content is click to play. We view in-stream video ads as a

distinct ad product that is often used by direct advertisers. Good examples of in-stream video are

YouTube and CTV supply.

Out-stream video ads are more similar to display ads. Video components are increasingly used

within traditional display advertising to improve the ad performance. Out-stream video ads are thus

a closer substitute to display advertising.

• the extent to which advertisers use more than one advertiser ad server, demand-side

platform, or ad verification and attribution provider and the factors that inform their

decision, and

This question would be best addressed by marketers. Our understanding is that marketers generally

single-home when it comes to advertiser ad servers. While multi-homing is more prevalent with

respect to DSPs, our understanding is that this is limited to large advertisers; small, less

sophisticated marketers will use a self-service solution like Google Ads which at the same time can

provide them with access to valuable Google Search and YouTube inventory. But even large

advertisers will generally use one DSP for a single campaign; using multiple DSPs for the same

campaign can pose considerable challenges when it comes to applying frequency caps but also

accurately measuring conversions across DSPs.

• the extent to which publishers use more than one publisher ad server or supply-side

platform and the factors that inform their decision.

Publishers generally use only one publisher ad server as this is operationally more efficient. Multi-

homing among publisher ad servers can make it hard to evaluate performance, given that ad servers

have their own reporting systems and methods for measuring performance. In addition, as the

ACCC notes, switching publisher ad servers is an expensive and lengthy process, and there is no

real alternative to switch to.

Google is a de facto monopoly.

Interim Report, page 46.

We only use Google’s publisher ad server (Google Ad Manager or “GAM”, previously known as

DoubleClick for Publishers or “DFP”). We have never considered multi-homing across publisher

ad servers given the disadvantages this would entail.

On the other hand, publishers (at least the larger and sophisticated of them) generally multi-home

across multiple SSPs with a view to increase the eligible demand for their inventory and maximize

yield. Indeed, header bidding was developed as a technology to enable publishers to place various

SSPs in a unified and transparent real-time competition, thus making up for the inefficiencies of

the waterfall system (and the advantage of Google’s own SSP).

Once a publisher has implemented

a header bidding solution for its property, adding or removing SSPs is fairly straightforward. Of

course, in the case of client-side header bidding there may be some technical limitations in the

number of SSPs integrated in the header (as adding too many partners may negatively affect page

latency) but we are able to manage any latency risk through appropriate measures (e.g., limiting

the number of partners that we integrate, applying strict time-outs for the header bidding auction).

It is worth adding that publishers may engage in multi-homing across SSPs in order to ensure they

have access to important sources of demand that are exclusive to a certain SSP. For instance, as

explained below Google Ads demand (which is important for many publishers, including us) is

exclusively (or almost exclusively) available through Google’s own SSP, “AdX” (now part of

GAM).

Even so, the fact that publishers may multi-home across SSPs in no way means that Google’s AdX

is subject to sufficient competitive constraints. As explained below (see our response to Chapter 3

Questions about SSPs and ad networks) AdX is very likely to have market power in the market for

SSPs, not least because for years Google has used the leading position of its publisher ad server to

favour it over rival SSPs.

Chapter 2 – The role of data

The ACCC invites further stakeholder views regarding the impact of Google’s restrictions

on market participants’ ability to access data required for ad targeting and ad attribution

functions, including:

• Google blocking advertisers’ ability to access its DoubleClick ID

We think marketers are best placed to answer this question. However, our general understanding is

that Google’s move has placed pressure on marketers that buy at least some inventory through

Google’s buy-side tools (that is, practically the majority marketers) to concentrate their ad spend

with such tools in order to have an accurate view across the inventory they buy (e.g., for purposes

of applying frequency caps or measuring conversions). Not doing so leaves the marketers with two

Note that publishers multi-homed even under the waterfall system in order to make sure that their inventory would

not be left unsold. On the inefficiencies of the waterfall system and AdX’s advantages, see below our response to

Question 12 of Chapter 4. See also our response to the Issues Paper.

In 2018 Google integrated its ad exchange AdX and its publisher ad server DoubleClick for Publishers (“DFP”)

into Google Ad Manager (“GAM”). For clarity purposes we shall use the original names of Google’s ad tech

solutions for publishers. We shall thus use AdX when referring to GAM’s ad exchange functionality and DFP

when referring to GAM’s ad serving functionality.

fragmented landscapes that cannot be reconciled and independently verified; inventory bought

through Google’s buy-side tools and inventory bought through other DSPs. With Google removing

the DoubleClick ID from buy-side event level files, marketers can no longer independently manage

their own audience strategies and they must rely on Google to ‘tell’ them how best to optimise their

ad spend.

• Google removing the ability for publishers to link bidding data from Google’s SSP

(Google Ad Exchange) to the impression-level data from Google’s publisher ad server,

and

Google provides us with information around the performance of our ad inventory via the GAM

query tool and the GAM Data Transfer files. This ad server data acts as an important source of truth

for our commercial operation, and almost all revenue-related advertising decisions rely on this data.

Primarily, the GAM query tool is used to pull information directly from the user interface.

Occasionally we will need to access more granular data than is available via the query tool, and so

we must use the Data Transfer files. The Data Transfer files give us insight into each ad auction

conducted on Daily Mail’s webpage. These files are important because they offer more granular

data than what is accessible via GAM query tool, for example, we can use Data Transfer files to

assess performance among discrete user populations.

We have access to seven impression-level datasets from Google: “Backfill Bids,” “Backfill Code

Serves,” “Backfill Impressions,” “Backfill Requests,” “Code Serves,” “Impressions,” and

“Requests.” The “Backfill” files include data from AdX and Open Bidding participants. The other

files include data from non-Google exchanges via header bidding, and direct sold activity.

The Data Transfer files are of most value if they can be linked together. In the past for example,

by linking the BackfillBids file to an Impression file through two fields – “KeyPart” and

“TimeUsec2” – publishers could compare losing Google bids to non-Google winning bids,

including the winning bid from client-side header bidding. That way, publishers could measure

the incremental value of a header-bidding partner compared to AdX and Open Bidding. As the

ACCC knows, this is no longer possible by reason of certain changes introduced by Google in late

2019. That has undermined publishers’ insight into bidding activity on a per-impression basis. This

creates opaqueness across our monetisation strategy, and further raises concerns around the Google

“black-box” bidding and decisioning processes across our audience. Google alleges it removed the

ability to join the “Backfill Bids” file in order to protect user privacy,

but we are skeptical that

privacy considerations can justify Google’s changes. In any event it is our, and not Google’s,

responsibility to protect the privacy of our users.

• Google’s proposals to replace third-party cookies on Chrome.

We are gravely concerned about Google’s proposal to replace third-party cookies in Chrome.

Without third-party cookies buyers are expected to reduce their investment into the open web due

Jason Bigler, “Rolling out first price auctions to Google Ad Manager partners”, 5 September 2019, available at

https://www.blog.google/products/admanager/rolling-out-first-price-auctions-google-ad-manager-partners/.

to a lack of targeting capabilities. This money will shift to large walled gardens such as Google and

Facebook where targeting and conversion measurement will be still possible at the user-level.

Buyers need to have an accurate mechanism to value and measure the media they purchase, and

publishers need a fair and non-biased system to deliver demand to their site with appropriate ad

serving controls.

We are concerned that the real purpose of Privacy Sandbox is not to allay privacy concerns – users’

data will still be exploited within Google’s walled garden just as it is at present – but to move the

control Google exercises over the ad market from their ad server to their browser, perhaps in order

to protect it from divestiture.

Beyond that, the particular concerns we have about Privacy Sandbox, as it currently stands, are the

following:

- Google claims Privacy Sandbox is an open-source project, but we doubt they will ensure

they have full market consensus prior to roll out. When the Accelerated Mobile Pages

format (“AMP”) rolled out many features were missing. In any event, while Google would

like to give the impression it seeks industry feedback, the reality is that they have never

committed to implementing the Privacy Sandbox only once they achieve sufficient market

consensus (or e.g., only if the Privacy Sandbox becomes a W3C standard).

Google has

only vaguely stated that it will not roll out the Privacy Sandbox if it judges that the Privacy

Sandbox proposals are “overall unsuccessful or insufficiently developed.”

- How can Google be prevented from giving itself an advantage with targeting, pricing and

attribution (as it has done within its DoubleClick ecosystem)?

- How can Google be prevented from manipulating FLoC audiences to benefit itself, or to

reduce monetisation to specific domains (we have experienced this in both AdX and

Display & Video 360 or “DV360”)?

- If the validity or accuracy of FLoCs is called into question (or is unknown), how can we

prevent ad demand gravitating to logged in environments, in which Google has a very large

market share? We should note that Google is the first-party domain across AMP pages

served by the Google AMP Cache, so presumably Google does not need to use the Privacy

Sandbox to track users. If AMP monetisation relatively improves compared to mobile web

once third-party cookies are phased out in Chrome, this could force publishers to further

Note that current discussions of the Privacy Sandbox proposals take place within the context of W3C groups that

have no standardization power, namely the W3C Improving Web Advertising Business Group and the Web

Platform Incubator Community Group. It is Working Groups that promulgate W3C standards.

Competition and Markets Authority, Online platforms and digital advertising, Market study final report, Appendix

G: the role of tracking in digital advertising, 1 July 2020, available at

https://assets.publishing.service.gov.uk/media/5fe49554e90e0711ffe07d05/Appendix G -

Tracking and PETS v.16 non-confidential WEB.pdf, paragraph 322.

We have experience of Google applying seemingly random classifications to our domains. These classifications

are then targetable within Google buying tools. The exclusion of our inventory from specific segments (in this

case brand safety segments) can have real monetisation impacts. We worry that similar issues could occur with

FLoC.

- If the Privacy Sandbox is rolled out, it must be as independent and interoperable as possible,

so other non-Chromium browsers can adopt it. In this way publishers could compensate for

any deficiencies of the Privacy Sandbox by improving yields on Safari and Firefox.

Proposal 1: Measures to improve data portability and interoperability (page 79)

The ACCC is considering measures aimed at increasing data portability and interoperability,

to reduce barriers to entry and expansion and promote competition in the supply of ad tech

services. Any such measures would require safeguards to ensure that consumers have

sufficient control over the sharing and processing of their data.

We agree with the ACCC that measures should be taken to reduce data-driven barriers to entry in

ad tech services and thus unlock competition. Data portability and interoperability measures have

in principle the potential to lower data-related barriers to entry by increasing data mobility.

However, in the context of digital advertising, we are sceptical as to whether data portability

measures would prove effective; it is not clear that consumers would have the incentive to share

their data for advertising purposes. Rather, it is more likely that consumers would use data

portability tools in order to switch to a rival user-facing service (e.g., a rival social network or

search engine) without losing their data. As a result, any impact on digital advertising would be

indirect, in that publishers competing with the walled gardens of Google and Facebook could

potentially increase the volume and variety of data they hold and thus improve their advertising

offering. As for the privacy implications of such solutions, we consider that these would not be

significant, since data portability tools would be activated at the consumer’s request and under the

latter’s control.

On the other hand, data interoperability measures (which, as defined by the ACCC, would not

depend on the existence of a consumer request) would have a greater potential to increase data

mobility and reduce the data advantage of Google. We share our thoughts below when discussing

the ACCC’s proposals for a common transaction ID and common user ID for attribution and

verification purposes.

More generally, we think measures to increase interoperability generally (i.e., not only data

interoperability) have an important role to play in rejuvenating competition in ad tech.

Interoperability needs to cover both demand, supply, and technical features. Examples of lack of

interoperability which should be addressed include: Google Ads demand is not accessible through

non-Google SSPs to any substantial degree; Google have combined their exchange and their ad

server into a single product such that Google AdX demand cannot be properly accessed without

adopting Google’s ad server; YouTube supply is closed off to non-Google ad buying and

measurement technologies; Google Analytics and Google ad server are integrated to share data

while non-Google Analytics systems cannot integrate with Google’s ad server.

Proposal 2: Data separation mechanisms (page 81)

The ACCC is considering the extent to which data separation mechanisms, such as data silos

or purpose limitation requirements, may be effective in levelling the playing field between

large platforms with a significant data advantage and rival ad tech providers. To promote

competition by levelling the playing field in relation to the data advantage of large digital

platforms, the ACCC is considering measures directed at mandating data separation within

companies in limited circumstances.

While we recognize that data separation measures may impose regulatory burdens on the

companies concerned, we think that in certain cases they may be the most effective way to

rejuvenate competition. When it comes to online advertising, Google’s unique data advantage

stems in large part from its ability to combine data collected across its numerous user-facing

services and across third-party properties to inform its advertising offering. Competing publishers

and rival ad tech vendors are at a clear disadvantage. It should be noted that this extensive internal

sharing of data may very well be questionable under data protection legislation – Google will

typically engage in such practices by default, without obtaining active, opt-in user consent, which

in certain jurisdictions is required by the law (e.g., the GDPR in the EU).

Among the various data-related proposals of the ACCC, we think that data separation measures

have the greatest potential to unlock competition by levelling the playing field. An additional

benefit is that such measures can considerably improve user privacy. It is questionable, to say the

least, whether Google needs to combine data sets across its multiple products in order to provide

its services to users. Independent ad tech has operated for many years without combining huge data

sets against personally identifiable and non-resettable identifiers.

It should not be overlooked that the greatest harms to privacy happen not necessarily in the open

web, but within the walled gardens of players like Google and Facebook which build intrusive real-

world user profiles. Privacy concerns with respect to third-party cookies seem to pale in comparison

to the first-party privacy concerns across Google and Facebook. For many years Google had

promised to keep separate its users’ personally identifiable information captured via its consumer

products from its DoubleClick advertising business. This changed in 2016 when Google updated

its privacy policy to combine its user data sets to help power its ad business.

In any event, we think some form of data separation measures is justified and necessary as Google’s

Privacy Sandbox begins to roll out. Google Chrome will soon block third-party cookies, thus

reducing the data advertisers are able to utilise to run their ad campaigns. Many advertisers have

spent considerable resource to build first-party customer data sets but unfortunately these are going

to become much less useful post-Chrome’s cookie changes as they will not be able to target against

them at scale, unless they do so via Google or Facebook. It is important to ensure that all parties

are utilising the same data and functionality within the Privacy Sandbox, and enforcing data

separation remedies may assist with this.

At the very least, Google should be prohibited from tracking users across sites for advertising

purposes once third-party cookies are deprecated. Technically speaking, Google may engage in

such tracking without relying on third-party cookies or the Privacy Sandbox; e.g., it may use its

Julia Angwin, “Google Has Quietly Dropped Ban on Personally Identifiable Web Tracking”, ProPublica, 21

October 2016, available at

https://www.propublica.org/article/google-has-quietly-dropped-ban-on-personally-

identifiable-web-tracking.

first-party status on AMP pages served from Google AMP Cache or Chrome’s sign-in mode.

Allowing this practice to continue while everyone else will be technically restricted from tracking

users across sites would grant Google an unjustified advantage over its rivals.

Chapter 3 – Industry structure and competitive conditions

Industry structure (page 88)

• In addition to the examples identified in Appendix E, the ACCC welcomes stakeholder

comments about whether there have been any other notable entry or exit in the

provision of ad tech services.

There was an increase in M&A activity in the second half of 2020, and early 2021. Notable deals

include:

- Tapad was acquired by Experian in November 2020;

- Beeswax (DSP) was acquired by Comcast’s FreeWheel in December 2020;

- WhiteOps was acquired by Goldman Sachs in December 2020;

- Magnite (SSP) acquired SpotX (SSP specialized in video) in February 2021.

• The ACCC also welcomes stakeholder comments about possible reasons for exit and

consolidation.

We broadly agree with the list of possible reasons cited in the Interim Report. The ad tech value

chain is largely monopolized by Google, with rival providers of ad tech services having a marginal

presence. In certain segments of the market (e.g., publisher ad serving) exit seems to be a direct

consequence of Google’s anticompetitive conduct. For instance, Google’s tie-like conduct,

whereby Google Ads is tied to AdX and AdX is tied to DFP means that it is extremely hard for

rival publisher ad servers to operate on a standalone basis; in order to challenge Google’s position

they have to enter multiple markets at the same time (that is, the market for SSPs and potentially

the market for DSPs). This might help explain certain acquisition activity aimed at achieving

greater vertical integration; even though it should be added that this strategy does not seem to have

made any real difference as no single operator has ever come close to challenging Google’s position

in the ad tech ecosystem.

Overall, increasing pressures on independent ad tech from Google and Facebook, combined with

data regulation and compliance costs, the deprecation of third-party cookies, and price competition

may be making it unfeasible to run a smaller ad tech business.

Allison Schiff, “Telenor Sells Tapad To Experian For $280 Million”, AdExchanger, 19 November 2020, available

at https://www.adexchanger.com/privacy/telenor-sells-tapad-to-experian-for-280-million/

Ryan Joe, “FreeWheel Buys Beeswax”, AdExchanger, 7 December 2020, available at

https://www.adexchanger.com/digital-tv/freewheel-buys-beeswax/

Goldman Sachs Acquires White Ops; Roku To Pass 100M US Users In 2020, AdExchanger, 22 December 2020,

available at https://www.adexchanger.com/ad-exchange-news/tuesday-22122020/

Allison Schiff, “Magnite Acquires SpotX For An Eye-Watering $1.17 Billion”, AdExchanger, 5 February 2021,

available at https://www.adexchanger.com/digital-tv/magnite-acquires-spotx-for-an-eye-watering-1-17-billion/

A lot of the M&A activity in 2020 was focused on identity and CTV, which are likely going to be

the two main growth areas in 2021. Perhaps the Covid-19 pandemic has contributed to accelerating

industry changes, and shifting consumer habits across digital, which has led to the high level of

exit activity in 2020.

Nature of competition (page 93)

• The ACCC is continuing to consider the factors that ad tech providers compete on.

The ACCC welcomes stakeholder comments on the importance of the factors identified

in this Interim Report, and the importance of any additional factors.

We fully support the ACCC’s preliminary view. In our view the ACCC lists all the relevant factors

that determine our choice of ad tech vendors. Performance of ad tech services (and ability to

measure such performance), pricing, ease of use, access to data and demand are the most important

factors we consider when selecting our ad tech partners. We also refer the reader to our response

to Question 17 of the Issues Paper listing the factors a publisher considers when assessing the

effectiveness of its SSP partners.

Access to advertisers, publishers and ad inventory (page 94)

The ACCC is continuing to consider the benefits of having an integration with another ad

tech service. Specifically, whether an integration is able to provide particular access to

advertisers, publishers or ad inventory.

• The ACCC is interested in receiving stakeholder views on the difficulty of establishing

integrations with other ad tech services (e.g. the cost and time involved), and whether

ad tech providers and users have experienced issues in establishing integrations with

other ad tech services.

We consider ad tech vendors are best placed to respond to this question.

• The ACCC is interested in receiving views on the role of integration service providers

such as BidSwitch.

We consider ad tech vendors are best placed to respond to this question.

Advertiser ad servers (page 99)

Stakeholders have indicated that ad tech providers compete on factors such as:

o performance of ad tech services,

o access to advertisers, publishers and ad inventory,

o ease and reliability of integration with other ad tech services,

o access to data and ad targeting capabilities,

o price and other fees, and

o ability to measure and verify the performance and quality of ad tech services.

• The ACCC is continuing to consider the competitive dynamics in the supply of

advertiser ad server services and the degree of competitive constraints faced by

Google as the major provider of these services.

• The ACCC is particularly interested in:

o limits to advertisers multi-homing,

o advertiser switching costs,

o importance of integrations with DSPs and other ad tech services for

advertisers,

o the types and sizes of set up and maintenance costs (including regulatory

costs), and

o recent entry and exit of advertiser ad servers.

We think marketers are best placed to respond to this question. However, our understanding is in

line with the preliminary view of the ACCC; the market for advertiser ad servers is largely

dominated by Google, with rival operators such as Flashtalking or Sizmek having a marginal

presence. Sizmek’s collapse (which eventually resulted in its acquisition by Amazon) is telling of

the state of competition in this market segment. We also understand that marketers generally single

home, at least when it comes to the crucial issue of campaign performance measurement, which is

likely to further strengthen Google’s position. The advertiser ad server is intended to function as

the marketer’s source of truth for its various marketing campaigns (e.g., measuring campaign reach,

frequency, and conversions). Multi-homing would make it harder to achieve a uniform and

consistent view over their campaigns. Finally, much like is the case with publisher ad servers,

switching advertiser ad servers is a lengthy and expensive process, as the marketer will need to re-

tag all of its properties and migrate to the new ad server.

Demand-side platforms (page 105)

• The ACCC is continuing to consider the competitive dynamics in the supply of DSP

services and the degree of competitive constraints faced by Google as the major

provider of these services. The ACCC is particularly interested in:

o the importance of exclusive access to ad inventory,

o prevalence of single-homing versus multi-homing,

o switching costs,

o the importance of access and use of data,

o vertical integration,

o whether there are any substantial differences between Google Ads and

Google’s Display & Video 360, and

o set up and maintenance costs associated with the establishment of a DSP

service. The ACCC is also interested in stakeholder views about the

substitutability between DSPs and ad networks.

While marketers would be best placed to answer this question, our understanding is aligned with

the ACCC’s preliminary view. Google is the leading, and very likely dominant provider of DSP

services. The exclusive access of Google’s DSPs (DV360 and Google Ads) to valuable YouTube

inventory creates a very strong incentive for marketers to use Google’s DSPs (plus, Google buy-

has given preference to its own SSP, in that rival SSPs are charged a 5-10% revenue share

and are prohibited from leveraging demand from their affiliated DSPs.

Through such

features Google has distorted free market operations, depriving rivals of impressions and

thus indirectly harming its customers.

- The fact that publishers may multi-home across SSPs in no way means that Google is

subject to competitive constraints. Publishers and rival ad tech vendors devised header

bidding precisely in order to make up for the inefficiencies of the waterfall system and

mitigate AdX’s advantage, but Google has always retained an advantage by reason of

controlling the publisher ad server, the ultimate arbiter in the ad selection process.

Subsequent features introduced by Google such as the minimum_bid_to_win information

(which is not shared with header bidding partners) or the Unified Pricing Rules continue to

tilt the balance in favour of AdX. Another example is Open Bidding, whereby publishers

have the ability to multi-home across SSPs and place them in competition against AdX.

Yet as the CMA has already found, Open Bidding was not intended to level the playing

field. Open Bidders are charged a 5-10% fee which places them at an obvious disadvantage

compared to AdX and are subject to additional restrictions.

As regards the substitutability between ad networks and SSPs, we note that ad networks act as the

single intermediary between publishers and agencies (or advertisers), whereas SSPs will run spend

via DSPs and often do not hold any meaningful relationships directly with agencies. But we have

seen a trend of ad networks either building in-house DSP functionality, or utilising SSPs to

purchase inventory from publishers that they do not hold relationships with. In any event, as

mentioned above, ad networks do not really command much media spend anymore, and ad network

revenue is not a significant part of our business. It is thus hard to see why would switch from an

SSP to an ad network.

Publisher ad servers (page 113)

• The ACCC is continuing to consider the competitive dynamics in the supply of

publisher ad server services and the degree of competitive constraints faced by Google

as the major provider of these services. The ACCC is particularly interested in:

o the links between Google’s publisher ad server and its SSP

o the presence of other non-Google publisher ad servers

o pricing of publisher ad server service

o set up and maintenance costs

(by SSP2) and 20% on the second impression (by AdX). In a world with dynamic revenue share the publisher is

charged 9% on the first impression (by AdX) and 31% on the second impression (again by AdX).

See also the discussion in the Competition and Markets Authority, Online platforms and digital advertising,

Market study final report, Appendix M: intermediation in open display advertising, 1 July 2020, available at

https://assets.publishing.service.gov.uk/media/5fe495c28fa8f56afaf406d4/Appendix M -

intermediation in open display advertising WEB.pdf (“CMA Final Report, Appendix M”), paragraphs 447 et

seq.

These would be: the prohibition to leverage demand from their own DSP as well as ad targeting restrictions (Open

Bidding is a server-side solution, hence rival SSPs have to engage in cookie syncing twice, once with Google and

then with their own buyers. This makes it harder to identify users and thus bid accordingly). See CMA Final

Report, Appendix M, paragraphs 451 et seq.

o the prevalence of single-homing, and

o switching costs.

We fully agree with the ACCC’s preliminary analysis. As explained in our response to the Issues

Paper, the market for publisher ad servers is highly concentrated, with Google capturing the lion’s

share. The ACCC’s findings confirm this and are well in line with the findings of overseas

regulators. Rival publisher ad servers are fringe competitors and do not pose a significant

competitive constraint to DFP.

As explained above, publishers generally single home when it comes to ad servers. This is line with

the publisher ad server’s role as a holistic yield management tool for the publisher; using several

ad servers adds complexity and inefficiency. This strengthens DFP’s position as the default

publisher ad server in Australia.

In addition, Google’s position is protected by considerable switching costs. Switching publisher ad

server is a lengthy and expensive exercise, typically lasting several months, and any migration error

risks costing thousands of lost revenue in terms of unsold impressions. As stated in our response to

the Issues Paper, and as confirmed by the CMA, Google has used Google Ads and AdX in order to

raise switching costs for publishers using DFP. Google has refused to participate in header bidding.

As a result, publishers may place AdX demand (and in turn, valuable Google Ads demand) in real-

time competition with other SSPs only if they use DFP as their ad server. This dissuades publishers

from switching to rival ad servers and further solidifies DFP’s position as the de facto ad server for

publishers. Please see also our response to Question 12 below with respect to the ACCC’s

preliminary findings in Chapter 4 of the Interim Report.

Chapter 4 – Vertical integration and conflicts of interest

Ad inventory integration (page 129)

1. How important is access to YouTube ad inventory to advertisers in Australia?

While marketers are best placed to answer this question, our understanding is aligned with the

preliminary view of the ACCC. For most if not all advertisers YouTube is a “must have” source of

ad inventory by reason of its unique scale and reach among consumers; the fact that YouTube

generated $ 15 billion in fiscal year 2019 is indicative of its popularity among marketers.

addition, as explained already above, in-stream video ads (the prime example being pre-, mid- or

post-roll video ads shown on YouTube) are not close substitutes to display ads and typically

command higher CPMs. This grants YouTube a unique role as the largest pool of video inventory.

In turn, this makes Google’s DSPs (which have exclusive access to such inventory) “must have”

products for most if not all marketers.

2. Do advertisers consider that multi-homing is a viable option for DSP services?

James Hercher, “Alphabet Reveals YouTube Revenue – $15B in 2019 – And More Granular Data”, AdExchanger,

3 February 2020, available at

https://www.adexchanger.com/investment/google-reveals-youtubes-ad-revenue-

15b/.

As explained above, while advertiser multi-homing among DSPs is technically possible, it comes

with significant technical limitations with respect to cross-DSP frequency capping and conversion

measurement. We understand this acts as an incentive for marketers to single home.

We also agree with the ACCC that smaller, less sophisticated marketers are very unlikely to multi-

home across DSPs, which will typically require dedicated staff and certain expertise. Such

advertisers are more likely to use self-service tools like Google Ads, which is relatively easy to

user and has no minimum ad spend requirement. A solution like Google Ads can also act as a one-

stop shop solution for such marketers, as it can be used to buy inventory from (a) YouTube; (b)

third-party publishers across the web; and (c) Google Search, by far the most important source of

search inventory.

3. Do advertisers consider that they must have access to Google’s DSP service?

As explained above, Google’s DSP services are likely considered by most if not all marketers as a

“must have” technology to access valuable YouTube supply. In addition, using Google’s DSP

services comes with cross-technology benefits such as integration with other Google products (like

Google Analytics) and enhanced interoperability with AdX.

4. Apart from YouTube ad inventory, is access to other exclusive ad inventory sold

through the ad tech supply chain essential?

No, other than YouTube we are not aware of any other exclusive ad inventory to which access is

considered essential by marketers.

5. Does selling ad inventory through multiple DSPs create privacy or technical problems

for publishers?

We do not think that selling ad inventory through multiple DSPs creates privacy or technical

problems for publishers that justify making such inventory exclusively available through the

publisher’s own DSP (as is the case with YouTube). Google’s privacy-related arguments should be

viewed with a sceptical eye, especially if one considers that at the time it announced its decision to

cut third-party DSPs’ access to YouTube inventory there was no reference to privacy

considerations.

In addition, as the ACCC correctly notes, Google itself collects a large amount of

data on users visiting non-Google sites

– and of course on Google’s owned-and-operated

properties like YouTube. Moreover. Google’s own DSPs (DV360 and Google Ads) participate in

real-time bidding auctions to purchase inventory of third-party publishers, yet Google does not

seem to be concerned about any privacy considerations in this case.

Neal Mohan, “Focusing investments to improve buying on YouTube”, DoubleClick Advertiser Blog, 6 August

2015, available at

https://doubleclick-advertisers.googleblog.com/2015/08/focusing-investments-to-improve-

youtube-buying.html (stating that “[t]o continue improving the YouTube advertising experience for as many of

our clients as possible, we’ll be focusing our future development efforts on the formats and channels used by most

of our partners. To enable that, as of the end of the year, we’ll no longer support the small amount of YouTube

buying happening on the DoubleClick Ad Exchange.”)

Interim Report, page 127.

In any event, we note that the CMA dismissed Google’s privacy justification, noting that that

“Privacy Enhancing Technologies (PETs) have been proposed by Google itself to allow targeted

advertising without user profiling; similar solutions could be adopted for YouTube as well.”

6. How easily are advertisers able to purchase YouTube inventory directly, or through

YouTube partners? Is this a viable option for all advertisers? Are there advantages

purchasing from YouTube ad inventory via the ad tech supply chain, rather than

directly?

We consider marketers are best placed to respond to this question.

Google’s vertical integration across the ad tech supply chain (page 132)

7. How important is access to Google Ads demand to publishers?

Access to Google Ads demand, and more generally, access to Google AdX is very important for

publishers. Most publishers consider access to this demand as a “must have”.

As far as Daily Mail is co

ncerned, internal data

Across our i

ndirect ads business, we serve around

of all impressions to Google AdX which generates around of all reve

nue (see Figure 1

below). We should note that the percentage of impressions won by Google AdX increased

after the r

ollout of Unified Pricing Rules in September 2019.

This is because

Competition and Markets Authority, Online platforms and digital advertising, Market study final report, 1 July

2020, available at

https://assets.publishing.service.gov.uk/media/5fa557668fa8f5788db46efc/Final report Digital ALT TEXT.pd

f , (“CMA Final Report”), paragraph 5.265.

Within Google AdX, Google Ads is a very important demand source and accounts for around

of all AdX revenue and around of all AdX impressions (see Fi

gure 2 below).

Do publishe

rs consider that Google Ads demand is accessible through non-Google

SSPs?

No, they do not. Publishers (us included) consider that Google Ads demand is predominantly (if

not exclusively) accessible through Google’s own SSP, AdX. For several years, Google Ads

publisher ad server, DFP. In other words, AdX displays limited interoperability when interacting

with third-party ad server.

From a pure technical perspective, it is possible for a publisher using a third-party ad server to

contact AdX demand. In particular, it is technically possible to implement an ad tag which when

executed will query AdX whether it wants to sell a given impression. However, this comes with

severe limitations by reason of Google not participating in header bidding. Under this setup, it is

impossible to place AdX in a real-time competition with other SSPs.

It is recalled that under the waterfall system publishers prioritized non-Google SSPs according to

their estimated performance. This system was inefficient, as it did not allow such SSPs to compete

on the basis of their real-time demand (and thus maximize yield for the publisher). AdX, on the

other hand, was always able to compete on the basis of its real-time demand by reason of a DFP

feature called Dynamic Allocation. Publishers and rival ad tech vendors created header bidding as

a mechanism to mitigate Google’s advantage and enable non-Google SSPs to compete for the first

time on the basis of their real-time bids. Header bidding is widely regarded as a superior

monetization solution compared to the old waterfall system, yet Google has refused to participate

in it and instead launched its proprietary Open Bidding solution. To the best of our knowledge, all

major ad tech vendors support header bidding but for Google.

Google’s refusal to participate in header bidding means it is very hard for publishers using a third-

party ad server to place AdX in a real-time competition with other SSPs; AdX will be stuck in the

waterfall, while all other SSPs will be called to participate in the header bidding auction. This can

severely disrupt programmatic revenue for publishers. There seems to be a technical workaround,

whereby the publisher first completes the ad selection process within the third-party ad server and

then contacts AdX, giving it the chance to outbid the winning bid from the third-party ad server.

However, as the CMA found in its Online platforms and digital advertising market study, this

technically complex workaround is highly inefficient, since it requires the publisher to run two

sequential auctions, which impacts latency and increases costs for the publisher.

In addition,

under this setup AdX potentially obtains a last look advantage over everyone else.

AdX’s limited interoperability with third-party ad servers, coupled with AdX’s almost exclusive

access to Google Ads demand, is one of the key reasons why publishers are locked into DFP and

do not switch to rival ad servers. As the CMA found,

“The effect of channelling most of Google’s demand through AdX and linking AdX to

Google’s publisher ad server is to increase the barriers publishers face in switching from

Google to a different ad server, reducing competition in ad serving.”

These two ties (Google Ads-AdX and AdX-DFP) have raised artificial barriers to entry and reduced

competition in publisher ad serving.

CMA Final Report, Appendix M, paragraph 436. The ACCC is of the same view. See Interim Report, page

134.

CMA Final Report, Appendix M, paragraph 436.

CMA, Final Report, paragraph 5.279.

13. Are there any impediments or disadvantages to using a third-party publisher ad

server, due to the way that Google’s SSP interacts with it?

Please see our responses to the previous question.

14. Why might an SSP decide not to participate in header bidding? Do any other SSPs

refrain from participating in header bidding auctions (or similar auctions)?

We are not aware of any legitimate reason why an SSP would not participate in header bidding. As

explained above, header bidding is a superior monetization solution, while it is also regarded as a

transparent and equitable auction, in that no vendor is afforded preferential treatment. This explains

why, whenever it is technically possible, header bidding will be preferred over the inefficient

waterfall system. To the best of our knowledge, Google is the only vendor not supporting header

bidding. This comes as little surprise, considering that Google viewed header bidding as a threat

which it had to quash, going as far as entering into an unlawful agreement with Facebook, as

explained in the recently filed Texas lawsuit.

In any event, Google’s justifications for not participating in header bidding are likely to be

pretextual, and have all been dismissed by the CMA within the context of the latter’s market study

into Online platforms and digital advertising.

In particular:

- It is not for Google to decide on the appropriate trade-off between latency and monetization.

This is a decision for the publisher to make. In any event, publishers can minimize any

effect on latency through a series of measures (e.g., limiting the number of header bidding

partners and imposing strict time-outs).

- A lack of transparency in header bidding should similarly be a concern for publishers, not

Google. In any event, publishers (us included) consider header bidding far more transparent

when compared to Google’s black box Open Bidding solution. Google’s argument that a

header bidding partner may pay a price lower than its bid without the publisher realizing it

is technically wrong; publishers can compare the price a header bidding partner pays with

the price that it bid (e.g., by using Google’s Data Transfer file which includes header

bidding bids) to detect any suspicious activity.

- Finally, Google’s concerns over not being able to protect its buyers’ data are largely

overblown. None of the DSPs or media agencies contacted by the CMA was particularly

concerned about the risk of disclosure of their bid data in header bidding.

After all, if this

were indeed a concern of buyers, then header bidding would most likely never have gained

traction as a method of purchasing inventory programmatically.

Complaint of the State of Texas et al. v. Google LLC, Case No 4:20-cv-00957-SDJ, redacted version available

https://www.texasattorneygeneral.gov/sites/default/files/images/admin/2020/Press/20201216 1%20Complai

nt%20(Redacted).pdf, paragraph 153 et seq.

CMA Final Report, Appendix M, paragraphs 438-444.

Id., paragraph 443.

Proposal 3 – Rules to manage conflicts of interest and self-preferencing in the supply of ad

tech services page 146

The ACCC is considering whether rules should be introduced that would aim to prevent and

manage the competition and other issues that can arise from vertical integration. In

particular such rules could aim to prevent anti-competitive self-preferencing and manage

conflicts of interest. The high-level obligations which could be covered by these rules include:

• requirements to put measures in place to manage conflicts of interest, such as

preventing the sharing of information between ad tech services, or obligations to act

in the best interest of publisher or advertiser customers

• requirements to provide equal access to ad tech services (i.e. level playing field

obligations to prevent self-preferencing), and

• requirements to increase the transparency of the operation of the supply chain.

15. Do you consider that such rules are necessary to promote competition in the supply

of ad tech services?

Yes, we think rules such as those contemplated by the ACCC are necessary to unlock competition

in the ad tech sector.

As a news publisher which depends on and embodies freedom of expression, we are reluctant, as a

matter of principle, to support state regulation. However, the ad tech supply chain has been

monopolized by Google, which has leveraged its market power to systematically favour its various

ad tech solutions. Features such as Dynamic Allocation, the dynamic revenue share, Open Bidding

and Unified Pricing Rules are examples of Google engaging in leveraging practices and distorting

competition to the detriment of publishers, advertisers, and ultimately consumers. The Texas

lawsuit asserts that Google has gone as far as entering into an agreement with Facebook to restrict

competition.

Against this background, we do not see any potential for the market to correct itself absent bold

regulatory intervention. While we support one-off interventions to tackle the source of the problem,

that is Google’s market power (e.g., separation measures), we realize that the ACCC might wish to

propose establishing ex ante rules which would act as a complement to ex post competition rules.

Ex ante regulation can be particularly useful in ensuring that any solution will be future-proof and

will not be rendered obsolete if e.g., Google decides to move part of the ad selection logic of its

publisher ad server to the browser as part of the Privacy Sandbox. Google constantly mutates its

conduct, hence there is a need for a flexible ex ante regime.

At the same time, the ACCC should not hesitate to initiate antitrust proceedings against Google

under its existing enforcement powers. The Texas allegations are very serious, as they suggest that

Google colluded with Facebook to restrict competition, and should be investigated by the ACCC

as well.

16. Do you consider whether the regulatory burden imposed by such a regime would be

justified by the potential benefits?

Yes. Establishing a regime to promote competition in the ad tech sector (with a focus on preventing

leveraging practices and conflicts of interests) will entail certain compliance costs for affected

companies like Google. However, we consider that any such regulatory burden would be justified

by the regulation’s potential to unlock competition. The current state of competition in the ad tech

supply chain is clearly unsatisfactory, and Google has over the years distorted competition to the

detriment of publishers, advertisers, and consumers. We do not see any other alternative to remedy

this situation (save perhaps for even bolder measures, like separation remedies).

17. If you consider such a regime should be implemented, what matters do you think such

rules cover and what would be the best way for such rules to be implemented?

We consider that such a regime should address all the issues listed by the ACCC in its Interim

Report. In particular, it should include rules to (a) prevent and manage the conflict of interests; (b)

ensure equal access to ad tech services (i.e. level playing field obligations to prevent self-

preferencing); and (c) increase transparency.

We think inspiration for such a regime can be drawn from the UK’s ongoing work to develop a

pro-competitive regime for firms with Strategic Market Status (SMS) comprising codes of conduct

and pro-competitive interventions. This represents a unique opportunity for the ACCC to

coordinate with the UK and introduce similar rules with respect to ad tech. A code of conduct for

ad tech services could be designed around some high-level objectives, similar to those

contemplated by the CMA, namely “fair trading”, “open choices”, and “trust and transparency”:

- Fair trading – address auction manipulation and demand/supply access restrictions. Ensure

equal access across all demand sources.

- Open choices – address self- preferencing concerns and ensure reasonable integrations are

made available across rival services/ad servers. Mandate interoperability across Google

Ads and GAM, as well as other Google products such as Google Analytics.

- Trust and transparency – reduce information asymmetries around fees, data, bids/pricing

and compliance with industry standards. Ensure Google does not utilise rival bid

information to favour its own demand.

We are sceptical as to whether such rules would be developed and implemented by the industry (in

the form of e.g., a voluntary code of conduct), as Google has clearly no incentive to participate in

constructive negotiations (or worse, Google has the incentive to ensure that the obligations in any

code of conduct will be weak and hard to enforce). We think the ACCC should step in and take the

lead in promulgating a mandatory code of conduct, but in close cooperation with industry

stakeholders. The ACCC should also be vested with the power to investigate complaints regarding

breaches of the code of conduct and impose remedies (e.g., under the threat of financial penalties),

including interim relief in urgent cases.

18. Do you consider that the provisions of the CCA are currently sufficient to address

competition issues arising from vertical integration in the ad tech chain?

The ACCC identifies alleged anti-competitive conduct by Google on pages 10, 26, 122 and 123 of

the Interim Report. That conduct, if proved, is in breach of the CCA. The CCA makes it illegal for

a firm with substantial market power to damage the competitive process by preventing or deterring

rivals (or potential rivals) from competing in the market. Misuse of market power through the

exploitation of vertical integration, in the ways described, is in breach of the CCA.

In our view the provisions of the CCA are therefore broad enough to catch the conduct that has

been alleged against Google. In fact, we are not aware of any conduct that has been alleged against

Google, in respect of its misuse of market power through the exploitation of vertical integration,

that is not caught by the existing provisions of the CCA. Accordingly, for the ACCC to take action

against such conduct does not, in our view, require amendments to the CCA.

Rather, in our experience, what is required is adequate resources to be allocated to the

identification, investigation and prosecution of such conduct. Our primary recommendation,

therefore, is not that time and effort be devoted to the design and implementation of amendments

to the CCA but, rather, that the ACCC should commit, whether with or without a corresponding

commitment from the Federal Government, to allocate sufficient resources to the identification,

investigation and prosecution of such conduct under the CCA.

The ACCC is also aware of other measures that have been proposed in Australia and

elsewhere to address the issues arising from Google’s vertical integration. In particular

requiring Google to structurally separate so that it is no longer integrated across the supply

chain. The ACCC invites views on whether such measures would be an effective and

proportionate response to the issues identified in this chapter.

Yes, we are of the view that separation measures (be it in the form of functional separation or even

full structural separation) would be an effective and proportionate response to the issues identified

by the ACCC. Separation remedies seem the only means available to tackle the root of the problem,

that is Google’s market power across the ad tech supply chain. We think that some form of

separation measures could also complement any ex ante regulatory regime and bolster its

effectiveness. For instance, ordering Google to functionally separate and erect strict Chinese Walls

between its DFP and AdX business would go a long way towards preventing the sharing of

information across Google’s businesses and ensuring that Google does not use DFP to favour AdX

and vice versa.

Chapter 5 – Pricing, fees and margins in ad tech

Undisclosed fees (page 159)

The ACCC is seeking stakeholder views on the extent to which ad tech providers are able to

charge undisclosed fees, as well as the extent to which this impacts on competition and

potential benefits from improved transparency

We agree with the ACCC’s preliminary view that a lack of fee transparency can create competition

issues if it limits advertisers’ and publishers’ ability to easily compare the performance, price and

efficiency of different ad tech vendors and make an informed decision on which vendor to use.

As noted in our response to the Issues Paper, greater access to information regarding fees may

allow publishers and advertisers to better agree on the most efficient routes for their ad spend

ensuring that more of the advertiser’s budget is spent on working media rather than the “ad tech

tax”.

While we realize that in some cases improved transparency may not be sufficient to spur

competition (e.g., because publishers and advertisers are locked into particular ad tech providers

so that they cannot switch to rivals) we firmly believe, as a matter of principle, that both publishers

and advertisers should have full visibility into the fees charged alongside the ad tech supply chain.

As the ACCC observes, publishers generally do not know what advertisers pay and conversely

marketers do not know what publishers receive for selling their ad inventory.

In itself this creates

the potential for ad tech vendors to charge undisclosed fees.

As regards Google Ads in particular, we welcome Google’s efforts to increase transparency by

providing certain point-in-time figures about its take rate – even though we are surprised it took so

many years for Google to finally disclose its fees with respect to Google Ads. We agree with the

ACCC that Google’s decision to make available these point-in-time figures is only a partial

response to the fact that Google Ads’ margin is not disclosed, not least because such figures are

limited in time and may vary considerably across inventory type or publisher. In addition, there is

nothing precluding Google Ads (and potentially other ad tech vendors) from extracting larger

margins in the future. We think it is a matter of principle and trust in the programmatic supply chain

that appropriate measures should be put in place to ensure that no hidden fees may be charged in

the future. To this end, all programmatic transactions should be receipted so advertisers and

publishers can track the flow of money. This could be achieved through the imposition of a common

transaction ID, as discussed below. At the very least Google should be obliged to provide its

publisher and advertiser customers, upon their request, with information on the price paid by the

advertiser and the remuneration paid to the publisher.

Chapter 6 – Transparency of the price, operation and performance of ad tech services

Stakeholder concerns with opacity of auction mechanics and results (pages 169-170)

The ACCC is seeking submissions from stakeholders on the following questions to help it

assess whether advertisers and publishers receive sufficient information to make informed

choices about the services and providers they will use. Specifically, we are seeking responses

to the following questions:

Interim Report, page 158.

Interim Report, page 155.

Note that a similar obligation is envisaged in the Digital Markets Act proposal of the European Commission.

See Article 5(g) of the Proposal for a Regulation Of The European Parliament And Of The Council on

contestable and fair markets in the digital sector (Digital Markets Act), COM/2020/842 final, according to

which a platform designated as a “gatekeeper” shall “provide advertisers and publishers to which it supplies

advertising services, upon their request, with information concerning the price paid by the advertiser and

publisher, as well as the amount or remuneration paid to the publisher, for the publishing of a given ad and

for each of the relevant advertising services provided by the gatekeeper.”

1. What information do you need about auctions used by an ad tech provider to assess

and compare their services to others in the supply chain?

(a) Why do you need this information and how do/would you use it?

(b) Do you receive this information?

There are a number of considerations that a publisher must take into account when selling its

inventory. Gaining accurate information allows us to proactively make decisions to better optimise

our access to demand.

First, we must consider the payment risk of any given exchange and their buyers (namely the risk

they will not pay us). To do this we must understand: the payment terms between the buyers and

the exchange, as well as our terms with the exchange; the level of insurance across both the

exchange and our own insurers; as well as any clauses for sequential liability and revenue claw-

back or withheld revenue. We also need to know what fees are charged and whether there are fees

on both the buy and sell side for an exchange. Exchanges are typically not forthcoming with this

information, often because they have poor terms with buyers that shift a lot of the risk to the

publisher.

We also need to analyse granular data either from a user interface reporting tool, or via event-level

files. We look at win rate, render rate, error rates and bid rate metrics as well as cross examining

by dimensions such as country, ad slot, browser, device, page type. The goal of this analysis is to

optimise any floors or blocks that need to be in place, as well as finding any issues.

We need to understand exchange-level auction mechanisms such as first- or second-price auctions,

auction flooring, exchange-level bid throttling, as well as brand safety or traffic monitoring tools

that are being used. In the case of Google, this means we need to understand all functionality within

the ad server and across their vertically integrated buy side tools. If we do not have sufficient insight

on this, it can be difficult for us to understand any causes for reduction in spend levels. We also

need to understand the exchange’s source of demand, whether they are directly integrated with

DSPs and demand sources, or whether they are reselling across other exchanges. This is often found

out via their ads.txt files. Knowing how many intermediaries are present in any supply chain is

important, as long supply chains introduce more complexity, reduce the end payout to publishers

and can increase the time taken for issue resolution.

We also measure response times, page load times, errors, and ad quality from our exchange

partners. This can be done via in-house tools or via third-party technologies. It is important for us

to be aware of any issues with site load speed or user experience caused by an exchange.

2. What information do you require, and what do you receive, on the following:

(a) the factors which are used by an auction algorithms to select the winning bidder?

(b) the factors used by a bidding algorithm to determine a bid price?

Typically, publishers do not have much insight into the algorithms used by the buyer to determine

the bid price, or by the exchange to determine which bid they will submit to our ad server for the

final ad selection process.

Exchanges will give us, upon request, examples of the RTB bid request that they send to buyers.

This will contain information related to the ad slot that is up for auction. We understand that

exchanges will often simply pick the highest bid returned, subject to any publisher rules setup

within the exchange, such as block lists or buyer prioritization.

Bidding algorithms will use a combination of the bid request information, along with third-party

data tied to the user ID and wider buyer campaign inputs to generate the bid value. The buyer may

also be running brand safety tools and audience targeting that may exclude the bid request from

being bid on. Understanding how these brand safety tools work is an ongoing struggle for

publishers.

Post auction information available to publishers can vary a lot depending on exchange. This

depends on what reporting is available from the exchange. Publishers will often ask for bid data to

understand not just the winning bids, but also any losing bids. Publishers also require both gross

and net price for impressions on a per impression basis. This way publisher can better verify that

the correct revenue share is being applied.

3. Are there differences in the auction information provided by ad tech providers? If so

please explain these differences?

There are differences between the information provided from the ad tech provider to the publisher.

Typically, publishers only receive ad auction information from the SSPs (we get no information

from other ad tech providers in the supply chain with whom we have no relationship). We require

information on both the won and lost bids, as well as gross and net prices. The below list represents

the most common metrics and dimensions that we utilise.:

- Metrics: Impressions, Revenue, Clicks, Bids, Matched Bids, Viewability, Competition rate,

Error rate.

- Dimensions: Advertiser, Buyer, DSP, Bid Outcome, Auction Type, Country, Ad Slot,

URL, Operating System, Device Type, User ID, Time, Deal ID, Ad Size, Price Floor Rule.

The main differences between providers are whether they will give event level data, or if not, what

dimensions are available in their reporting tool. Most of the above dimensions and metrics are

initially captured by the SSP, but often the method in which the SSP stores the data and ultimately

makes the data available, can reduce some of the data granularity. Some dimensions such as User

ID or exact time are only available via event level data.

Transparency over the pricing of ad tech services (pages 171, 174)

4. Do publishers currently receive sufficient information from SSPs to verify the

accuracy of the fees charged?

Publishers can of course only have insight on the fees charged by the SSP that they work with

directly. The fee charged is usually a revenue share, and the percentage will be stated in the

8. How does a lack of information about fees or take rates impact the ability of

advertisers and publishers to make informed choices about how they will use

services in the ad tech supply chain?

Without sufficient information on the fees or take rates charged alongside the ad tech supply

chain we are unable to make informed and intelligent business decisions about which operators

to use and identify the most cost-effective path to demand.

An interesting and perhaps overlooked example relates to the average revenue share feature

which is enabled by default for AdX, according to which AdX may adjust its revenue share on

an impression-by-impression basis, as long as it achieves on average the contracted revenue

share over a certain period of time.

In its help centre, Google warns publishers that disabling this feature reduces AdX yield, which

may give the impression that this feature is beneficial for publishers.

This may be true in a

case where there is no other SSP that can solicit a bid exceeding the publisher’s floor. In this

case, if AdX retains its full revenue share, there is a risk that it will not exceed the floor and

the impression will be thus left unsold.

Reducing its revenue share in a given impression thus

helps avoid leaving impressions unsold.

But the answer can be very different if there is

another SSP that can sell the impression charging a lower revenue share than AdX.

In this

case, AdX can adjust its revenue share so that it can win the rival SSP.

This may very well

harm the publisher as in a later impression AdX can charge a much larger revenue share.

while normally the publisher would have sold one impression through a rival SSP (with a low

revenue share) and one through AdX (with its higher revenue share), it ends up selling both

impressions through AdX (with its higher revenue share). The effect is that AdX wins more

impressions, resulting in larger fees for the publisher in the aggregate (in other words,

publishers are precluded from benefiting from the lower fees charged by other SSPs). It is

practically very hard for us to precisely calculate to which extent this may happen across our

inventory

(Google does not provide gross bids on its log files, hence it is not possible to

calculate Google’s revenue share on a per-impression basis), and whether the dynamic revenue

share feature’s net impact on our revenue is thus positive or not.

See https://support.google.com/admanager/answer/7031785?hl=en.

Assume AdX charges a 20% revenue share and a publisher has set a price floor of 0.90. Assume AdX returns a

gross bid of 1.00, that is a net bid of 0.80. The impression will be left unsold.

Assume in the above example AdX adjusts its revenue share to 8% for that impression. Its net bid is 0.92 (above

the floor), hence it sells the impression.

Assume in the above example that there is a rival SSP charging a revenue share of 9% and which also returns a

gross bid of 1.00, that is a net bid of 0.91 (above the floor). In a normal situation (without the average revenue

share feature) the rival SSP should win, not AdX.

Assume in the above example AdX adjusts its revenue share to 5%, so that it returns a net bid of 0.95 to win the

impression.

Assume that there is a second impression whereby AdX faces no competition and where the publisher has set a

low floor (say 0.10). Assume AdX returns a gross bid of 1.00. AdX applies a revenue share of say 35% and thus

returns a net bid of 0.65.

Transparency over the performance of demand-

side services and digital display

advertising (pages 176, 179 and 182)

The ACCC is seeking feedback from advertisers and publishers regarding their

satisfaction with the service provided by verification and attribution providers, including

on the following issues:

Are you satisfied with the services provided by verification and attribution

providers? If not, what are you not satisfied with regarding their service?

Whilst marketers are best placed to answer this question with regards to attribution services,

we do have some experience with verification services.

Verification providers offer services including fraudulent traffic detection, viewability, site

context and brand safety measurement. The main ad verification vendors are Moat, Integral

Ad Science, DoubleVerify and ComScore. Google also offers verification services which are

built into their DSP. We utilise verification services to understand how our inventory performs

for our buyers.

There are a number of issues with the services and products offered by verification providers.

We have concerns with the speed in which verification services can classify our inventory. As

a news publisher the lifecycle of any given article on our site is usually less than 72 hours, this

means that we have a short window of time to monetise our content whilst it remains in the

news cycle. The problem is that will take the verification providers up to 48 hours to classify

new content, so any advertiser using verification technology to target specific categories will

often be opted out of bidding on our inventory as our content will be “unclassified” within the

verification provider’s system. It is often the case that “unclassified” content will be treated

the same as “unsafe” content.

The opt-out process for unsafe or unclassified content can occur both before and after the

impression is won. The opt-out process before the impression is won is achieved by simply

not bidding on the request. The opt-out process after the impression is won is achieved by

rendering an ad overlay on top of the marketers creative to “hide” the messaging from the

user.

In some cases, the marketer will not pay us even if their verification service opts out

after winning the impression. These ad overlays can waste our ad inventory, cost us revenue

and give a poor user experience.

We also find that verification vendors will scan the entire page, not taking into account any

nuance around page content. Daily Mail pages are long and contain many links to other Daily

Mail articles. These links will contain the article titles from the landing page of the link. Some

verification vendors will scan the entire page for negative keywords including the hundreds of

other content links on our page. This results in many false positive classifications and hampers

our monetisation, as we may be unfairly classified as unsafe.

See https://twitter.com/aripap/status/1236455200845684736?s=20%C2%A0 for an illustrative example.

The verification providers have admitted to us that there are issues with both the speed in

which they scan our content, and also the nuance used to avoid scanning content which is not

relevant to the body of the article. Firstly, verificat

ion vendor’s customers are primarily

marketers, and so it is difficult to engage in conversation with the verification vendors to help

us better understand how their technology works. Secondly, their solution to this problem often

includes working more closely with us via a paid relationship, where they can better and faster

classify our content. They ask us to purchase their services, which we would otherwise not

need if they had classified us correctly in the first place, and of course, it is not possible for a

publisher to engage in a paid relationship with every verification vendor.

Additionally, across our direct sales efforts, we find that verification vendors have difficulty

in measuring viewability for non-standard ad formats, and so custom work is often needed for

marketers to properly measure viewability across our site. Our direct deals often involve

meeting specific goals as measured by the marketer. We find that even if both parties use the

same verification vendor, their measurements via the publisher integration will differ from the

measurement via the marketer’s integration. This is likely due to technical differences in the

way the verification code is loaded, but it is frustrating that the same verification company

measuring the same content will have discrepancies between the publisher and the marketer.

Publishers often do not know what verification vendor

is being utilised to measure their

content, or the extent to which this vendor is blocking our content, o

r the extent of the

commercial impact.

10.

Do you consider that the metrics you received from your verification and

attribution provider are accurate?

Please see our response to the previous question.

11. Would you be able to switch measurement and verification

providers if you

wanted to? What are the largest obstacles to you switching, if any?

Switching costs with respect to verification vendors are not high. However, switching does

require technical implementation, along with the necessary contractual negotiations, and so

the decision to change providers is not trivial. The primary reason a publisher would switch

providers is to better align their own measurement with the measurement of the marketers that

are buying ads on their site. We try and work with the most popular verification vendors to

ensure best possible alignment with the buy side.

12. Are advertisers able to independently verify the performance of ads served on

YouTube?

We consider that marketers are best placed to answer this question. Our general understanding

is that advertisers have no ability to independently verify the performance of ads on YouTube

inventory, since Google does not allow third-party measurement pixels on YouTube, allegedly

for privacy reasons.

Marketers may use third-party measurement providers which however

rely on the information Google makes available through its Ads Data Hub product. By

definition, this is not independent measurement.

13.

Can third party verification and attribution providers access sufficient data

through the Google Data Ads Hub to independently verify the performance of

ads served on YouTube? If not, what data do verification and attribution

providers require access to in order to perform this function?

We consider third-party verification and attribution providers to be best placed to answer this

question.

14.

Does providing third party verification providers with access to raw data, or

allowing them to place verification tags (or pixels) on ads, create privacy

concerns?

We consider marketers are best placed to answer this question.

15. Are advertisers currently able to conduct effective and independent attribution

of their ad campaigns?

Whilst marketers are best placed to answer this question, from our perspective it seems that

marketers are currently able to me

asure attribution across environments where identity is

available. We see marketers shift spend to environments and channels where they can track

attribution against user IDs. This independent attribution is primarily made possible via third-

party cookies on web, and via IDFA or GAID across apps. Without third-party cookies, this

independent measurement across the open web will not be possible.

16. Will upcoming changes Google is making to the data it shares and Google Chrome

affect advertisers’ ability to conduct multi-touch attribution? If so, what will this

impact be?

Whilst marketers are best placed to answer this question, from our perspective we have great

concerns over the ability for marketers to properly conduct attribution modelling across

Google Chrome after Google deprecates third-party cookies. As described in our response to

the Questions focused on Chapter 2 of the interim report, we believe advertisers may shift

spend away from the open web and into the walled gardens when they are no longer able to

measure attribution after the phase out of third-party cookies.

17. Will access to the data via the Google Ads Data Hub allow advertisers to conduct

full and independent attribution of Google’s DSP services?

We consider marketers are best placed to answer this question.

See Allison Schiff, “It’s Official: YouTube No Longer Accepts Third-Party Pixels”, AdExchanger, 15 January

2021, available at

https://www.adexchanger.com/platforms/its-official-youtube-no-longer-accepts-third-party-

pixels/.

18.

Does the use of user IDs and cookies in providing attribution services create

privacy concerns?

The use of user IDs for attribution purposes may create privacy concerns, as it can be utilized

to track users across online properties. However, these privacy concerns may in fact be smaller

than what one might initially think:

- Using user IDs for attribution purposes is far less intrusive from a privacy perspective

compared to using user IDs for ad targeting, which involves creating user profiles

based on the browsing habits of individuals. Conversion measurement and attribution

on the other hand only aim at finding out whether an ad campaign was successful or

not.

- User IDs are typically pseudonymous identifiers and are not linked to any personally

identifiable information like name or email address. They can also be programmed to

reset periodically, e.g., every 90 days (corresponding to the attribution window

marketers use). Google, on the other hand, links user IDs with (persistent) personally

identifiable information, namely Google user accounts.

19. Do stakeholders consider there are any other issues with the ability to conduct

attribution of ad tech services?

We do not have anything specific to add at this point.

Proposal 4 – Implementation of a voluntary industry standard to enable full, independent

verification of DSP services (pages 182-183)

To enable advertisers to assess DSP services fully and independently and encourage

competition, industry should develop a standard that allows full and independent

verification of DSP services.

This standard should set out minimum requirements for this, along with the categories

of data necessary to enable third parties to provide full and independent viewability,

fraud and brand safety verification services.

The ACCC considers that this should initially be left to industry to develop and

implement, but that other options could be considered if this was not successful.

The ACCC is seeking stakeholder feedback on proposal 4. In particular the ACCC is

interested in advertiser, ad agency and measurement and verification providers’ views

on the following questions:

20. Do you have to access the data you need to conduct verification of Google’s ad

tech services? If not, what data do you require that is not available?

We consider marketers are best placed to respond to this question.

21. How does the ability to verify the performance of Google’s DSP services compare

to other DSPs?

We consider marketers are best placed to respond to this question.

22. What measures would be most effective to ensure that all DSP services can be

fully and independently verified?

We consider marketers are best placed to respond to this question.

23. What are the risks to user privacy from third parties providing full verification

services? Could such measures promoting this be implemented in a way that

would protect the privacy of consumers?

We consider marketers and/or verification vendors are best placed to respond to this question.

Proposal 5 – Implementation of a common transaction ID (page 183)

Industry should implement a common system whereby each transaction in the ad tech

supply chain is identified with a single identifier which allows a single transaction to be

traced through the entire supply chain. This should be done in a way that protects the

privacy of consumers.

We are seeking stakeholder feedback on whether a common transaction ID would be

effective to address potential transparency issues, and whether it is possible to implement

a common transaction ID in a way that protects user privacy. In particular:

24. Would a common transaction ID assist in making pricing and auctions more

transparent?

Please see our response to Question 27 below.

25. What risks does a common transaction ID pose to user privacy?

Please see our response to Question 27 below.

26. How could a common transaction ID be implemented in a way which mitigates

any risks to consumers’ privacy?

Please see our response to Question 27 below.

27. How should such a recommendation be implemented?

Implementing a common transaction ID would help improve transparency in the programmatic

supply chain. The recent PwC/ISBA report into programmatic supply chain transparency in

the UK found that publishers only receive 51% of advertising spend, while 15% of advertiser

spend could not be attributed (the so-called “unknown delta”).

The design of a common transaction ID should be such that each in

dividual impression

generated on a publisher site has a unique random identifier generated by the publisher at the

point of the bid request. This ID should allow for both the tracking of the ad throughout the

supply chain, and the ability for the advertiser to recognize duplicate bid requests for a single

impression.

Whilst it is possible for a publisher to generate an ID in this manner, it will require full industry

collaboration to design a standardised mechanism which would allow for the passing of this

ID into each ad tech vendor. It will also likely require some third party to link this information

together from each ad tech vendor. Each ad tech vendor

would need to agree to make this

information available via their reporting systems to an elected third party. Some solutions such

as Fenestra have been attempting to measure the supply chain ‘ad tech tax’, however this is a

difficult task that would be made considerably more accurate and effective if

a common

transaction ID were to be implemented. In our experience some ad tech vendors have not been

forthcoming in making their data available to assist with these efforts, however, there is little

pressure to do so given the difficulties in supply chain measurement. If a common transaction

ID was available, then the pressure would grow for ad tech vendors – both independent ad tech

and Google – to provide this data.

A common transaction ID would also allow the advertiser to identify multiple bid requests for

a single impression, and thus assist with buyers’

supply path optimization and reduce the

wasted computational costs associated with bidding on the same impression multiple times –

a problem compounded by Google’s decision to launch Open Bidding. Whilst there have been

developments in supply chain

transparency through sellers.json, which allows buyers to

recognize which ad tech systems a given impression has passed through, it can be difficult for

a buyer to recognize that multiple bid requests may relate to the same impression (which is

why The Trade Desk only buys via one integration i.e., Prebid or TAM or Open Bidding, to

reduce costs). If a buyer could recognize this better, they could reduce their bidding costs.

Publishers can already pass custom IDs into some independent ad tech systems at the point of

the bid request (for example audience/data segments), but Google does not accept custom

information from the publisher in the same manner, and so Google would need to develop their

products to allow for this.

We do not believe there is a privacy concern with generating a common transaction ID, as this

ID in itself would not be able to identify a user.

Proposal 6 –

Implementation of a common user ID to allow tracking of attribution

activity in a way which protects consumers’ privacy (page 184)

Introduction of a secure common user ID, which ad tech providers would be required to

assign to any data used for attribution purposes. Such a measure could be developed and

implemented by industry

The ACCC is seeking stakeholder feedback on this proposal. In particular, the ACCC is

interested in views on the following questions:

28. Would a common user ID be an effective way to improve transparency in the ad

tech supply chain?

Please see our response to Question 30 below.

29. Could this proposal be implemented practically and is it justified?

Please see our response to Question 30 below.

30. Could this proposal be implemented in a way which protects consumers’ privacy?

If so, how?

A common user ID would assist greatly with the tracking and attribution of ad spend. It would

help democratize the ad ecosystem with a shared common identifier available to all

participants that adhere to a common set of rules for its usage.

A common ID could be linked using third-party

cookies today, however, with Chrome’s

upcoming Privacy Sandbox browser changes

, the industry would need another method to

retrieve a common ID. It seems technically possible that this functionality could be built into

the browser to allow the passing of a common ID, or managed via some other first-party cookie

process.

Much like Google has developed an opt-out system for tracking its logged in users across the

web,

a similar system could be used for independent ad tech to track users for attribution

purposes. With collaboration across Google and the open web, a common transaction ID could

be generated that is not passed via third-party cookies and that provides a mechanism for users

to manage their consent preferences, and agree to a standardized set of terms of use for the

common ID. Any market participant that wished

to utilise the common ID would need to

adhere to these terms.

This mechanism would give comparable privacy protections to the current methods utilised by

Google for tracking its users. It will arguably pose a lesser privacy concern to Google’s current

usage of logged in user data, which involves linking multiple data sets containing personal

information from its various consumer products.

By making available a common ID along with the necessary consent frameworks, users would

benefit from being exposed to fewer ads (as the value of each ad would be greater), having

continuing access to free premium content, as well as having the ability to

control their

preferences. A common standardized ID for attribution purposes

would also reduce the

incentive for more surreptitious forms of tracking such as user fingerprinting whereby various

device attributes are utilised without the user’s permission t

o generate persistent forms of

tracking.

Ad verification problems for publishers (page 187)

The ACCC is seeking stakeholder feedback on any issues publishers face in dealing with

measurement and verification providers, and potential solutions. In particular:

31. What challenges do publishers face in their inventory being blocked due to brand

safety issues?

Please see our response to Question 33 below.

See https://adssettings.google.co.uk/authenticated.