Ref. Ares(2018)2540105 - 16/05/2018
Ref. Ares(2018)2751838 - 28/05/2018
Initial Response to
Google's Third Set of Commitment Proposals
By Adam Raff and Shivaun Raff
Co-founders of
Foundem and
SearchNeutrality.org
February 5, 2014
Introduction
Google’s third set of proposals suffer from the same fatal flaw that rendered its previous two
proposals1 considerably more harmful than helpful.
The main difference between Google’s second set of proposals and the third is the improved
visibility of Rival Links; what Commissioner Almunia has called “comparable display”. But, as was
pointed out in detail during both previous market tests, the visibility of Rival Links is entirely
irrelevant as long as they remain Paid Rival Links. This is because Google, not its rivals, would be the
main beneficiary of any profits derived from these links. Moreover, in a further devastating blow to
consumers, the
pay-for-placement elements of these proposals remove the last vestige of merit- or
relevance-based placement for commercial verticals.
There is overwhelming empirical evidence and a virtually unanimous consensus among market
participants and consumer groups that the introduction of Paid Rival Links would inflict additional
grave and irreparable harm to competition, innovation, and consumer choice. It is deeply troubling
that Commissioner Almunia seems determined to disregard this evidence and consensus.
Paid Rival Links are a New Form of Abuse – Not a Remedy
Google stands accused of manipulating its search results to systematically favour its own services
and demote or exclude those of its rivals, thereby diverting vast quantities of traffic and revenues
away from competitors and to its own growing stable of competing services. Given Google’s
overwhelming dominance of search and search advertising, these practices have a devastating
impact on competition, consumer choice, and innovation wherever Google deploys them. These are
not hypothetical risks: during the period of the Commission’s investigation, Google has already all
but eradicated the product-price-comparison market in Europe and the US, and it is now embarked
on a similar destruction of travel search and financial price comparison.
Google’s third set of proposals suffer from exactly the same fundamental flaw as its previous
proposals. Remarkably, they do not offer one word directed at ending or curtailing its abusive
search-manipulation practices. Instead, Google proposes to continue these practices entirely
unabated. The one so-called concession that Google offers is the addition of rival links. But, because
rivals are forced to pay for these links, they are nothing more than a confidence trick. Under the
guise of an attempt to partially dilute the anti-competitive impact of Google’s systematic self-
preferencing, Google is actually trying to get away with introducing an entirely new and
extraordinarily lucrative form of abuse; one that, due to the economics of over-subscribed auctions,
would force Google’s rivals to hand Google the majority of any profit generated from these links.
It is important to understand three fundamental points about Google’s proposals. First, although
Google’s proposals describe two varieties of rival links, the vast majority of these links would be
Paid Rival Links2. Second, there is no objective justification for making competitors pay for any of these
1 See
http://www.foundem.co.uk/Foundem_Analysis_Google_Proposals.pdf and
http://www.foundem.co.uk/Foundem_Comments_Google_Revised_Proposals.pdf
2 It is likely that the few domains that may start off with Free Rival Links (such as local search and review sites)
would soon be transitioned to Paid Rival Links, particularly given the very low bar that Google has set itself to
1
links. Third, the costs to competitors of these Paid Rival Links would be substantial, not nominal; as is
the case with Google’s existing AdWords system, bidders will be compelled to bid away most of their
anticipated profits—in the process, transferring most of their profit to Google.
To understand these three points is to understand that Google’s proposals would do nothing to
remedy Google’s abusive search manipulation practices—instead, the proposals would entrench,
extend, and escalate these practices.
Why Paid Rival Links would Hand the Majority of Rivals’ Profits to Google
In order to assess the importance and efficacy of a particular channel to market, it is essential to
consider both the profitability and the revenue of that channel. This is particularly important in the
case of Paid Search (and Paid Rival Links), where the economics of an over-subscribed auction
naturally drives prices up to the point where the majority of any profit is handed to Google.
A hypothetical but reasonably representative example might help to illustrate this crucial point. A
company that devotes considerable resources to its Paid Search advertising campaigns might
generate twice as much traffic and revenue from Paid Search as it does from Natural Search—
perhaps generating £20M revenue from Paid Search compared to £10M from Natural Search. The
economics of Paid Search, however, means that the advertising costs of generating that £20M might
be something in the region of £18M, whereas the costs of generating the £10M from Natural Search
would be considerably smaller, even negligible. Thus, even a company with twice as much Paid
Search revenue as Natural Search revenue would typically earn the vast majority of its profits from
its smaller, but vastly more profitable, free traffic—in this case, somewhere in excess of 80%.
Note that, in the above scenario, the main beneficiary of the company’s considerable Paid Search
efforts would be Google, which would pocket around £18M of the company’s £20M of revenue.
These harsh economic realities of Paid Search (for everyone but Google) hold true for the smallest to
the largest of Google’s advertisers:
“MoneySupermarket hire the brightest and best digital marketers who spend in the region of
£40M a year with Google. Every day, the MoneySupermarket team bid on more than 8 million
different terms through complex and sophisticated bid management systems. As such, every day
we adjust and flex our spend based on efficiency and profitability. Paid search is a difficult
marketing tool for businesses to make profitable because of its auction-based methodology; it
contributes very little to our overall profitability.”
MoneySupermarket CEO Peter Plumb, Insurance Times
, 1 February 2013 [emphasis added]
If the Commission allows Google to all but eliminate free Natural Search traffic for all businesses but
Google’s own, as Google proposes, the harsh economic realities of Paid Search become apocalyptic.
Were the Commission to adopt Google’s proposals it would be falling for a breathtakingly audacious
confidence trick—one that would leave Google in sole possession of the efficient, low-overhead
business model that has characterised and fuelled the Internet revolution.
qualify a domain for this transition. Certainly, Google would have a considerable financial and anti-competitive
incentive to make this transition wherever possible.
2
By becoming the main beneficiary of its competitors’ services, as well as the sole beneficiary of its
own, Google’s monopoly would become even more entrenched and unassailable. Any competitors
that survive the transition to such a radically altered and unfavourable marketplace would become
little more than advertising arbitrage services for Google.
If adopted, Google’s proposals would effectively hand Google a five-year mandate to extend its
monopolisation of horizontal search into a monopolisation of Internet commerce. It is easy to
understand why Google is doggedly pursuing a settlement based on these proposals, but it is
inexplicable that the Commission would even entertain it.
An End to Merit- or Relevance-Based Placement
The proposed selection and ranking mechanism for Paid Rival Links is based primarily on payment
not relevance. Paid Rival Links are therefore a significant and immensely damaging departure from
relevance- or merit-based organic search results.
Under Google’s proposals, the selection and ranking of Paid Rival Links would be based on the
keywords that advertisers choose to bid on, the amount advertisers are willing to pay, and the
predicted efficacy of the related ad text. Thus, there is no significant “relevance-based” component
to Google’s decision about which Paid Rival Links to display and in what order. Under Google’s first
proposals, which only allowed rivals to bid against entire domains, Google would have had some
work to do to calculate who from the bidding pool might be relevant to each specific query and how
to combine that information with who was willing to pay Google the most. But under the proposed
more granular AdWords-based system, the advertisers (in this case, Google’s competitors) would
assume most of this burden. Each advertiser would have to individually bid on every specific
combination of keywords for which their service is relevant, leaving Google free to focus almost
exclusively on maximising its revenues.
Not surprisingly, Google’s mechanism for selecting and ranking AdWords or Paid Rival Links has little
to do with quality or relevance and a lot to do with maximising Google’s ad revenues. As Google
explains, Paid Rival Links are ranked according to the following simple formula:
Bid multiplied by Predicted Click-Through-Rate (pCTR)
This formula tells Google how much money it can expect to make by displaying each particular ad.
By Google’s own admission, it selects the three most likely to earn Google the most money. Not the
three most relevant and not the three highest quality.
Sadly, there is nothing new in Google attempting to disguise its anti-competitive pursuit of profit as
the pursuit of quality.
3