Jan 11, 2017
The Value of Behavioral Targeting
This study seeks to provide an initial understanding of the effect of behaviorally targeted advertising on
advertising rates and revenues. A survey of twelve ad networks was conducted to obtain quarterly data on
pricing (CPM data), conversion rates, and revenues across various types of ad segments (run of network
advertising and behavioral advertising). The survey results reveal three key findings: (1) Advertising rates are
significantly higher for behaviorally targeted ads. The average CPM for behaviorally targeted advertising is just
over twice the average CPM for run-of-network advertising. On average across participating networks, the price
of behaviorally targeted advertising in 2009 was 2.68 times the price of run of network advertising. (2)
Advertising using behavioral targeting is more successful than standard run of network advertising, creating
greater utility for consumers from more relevant advertisements and clear appeal for advertisers from increased
ad conversion. (3) Finally, a majority of network advertising revenue is spent acquiring inventory from
publishers, making behavioral targeting an important source of revenue for online content and services
providers as well as third party ad networks.
This study was sponsored by the Network Advertising Initiative (NAI). The NAI is a coalition of more than 40 leading online marketing companies committed to building consumer awareness and reinforcing responsible business and data management practices and standards. For a description of the NAI and a list of its members, see http://www.networkadvertising.org/index.asp (last visited Mar. 3, 2010).
1. Howard Beales has been an Associate Professor at the School of Business at George Washington University since 1988. He formerly was the Director of the Bureau of Consumer Protection at the Federal Trade Commission.
P a g e | 1
In recent years, popular Internet content and services (such as online news, blogs, e-mail
and social networking services) have been funded increasingly by advertising rather than through
charges to consumers. Many providers of Internet content and services (publishers) depend on
advertising revenue to develop and present their offerings to consumers. Web publishers range
from large scale media websites and portals that sell advertising directly to potential advertisers to
so-called long tail sites with smaller and more specialized audiences.
If advertising is to remain the primary means of financing Internet content, then advertising
rates will be a critical determinant of the kind and quality of Internet content available. Unless
publishers can effectively capture some of the value they create for viewers, they will not be able to
provide as much content, or content of the same quality as viewers have come to expect. The
fundamentals of online advertising markets and behavioral targeting are described in more detail
in Appendix A.
One strategy widely used to increase the value of advertising is behavioral targeting. Using
information about online behavior, including sites visited and interest in particular types of
content, behavioral targeting seeks to serve advertisements that particular groups of consumers
are more likely to find interesting. If advertising better matches consumer interests, consumers are
more likely to respond to the message, and advertisers will be willing to pay more for ads delivered
to such an audience.
Behavioral targeting is used in different ways. Large publishers with diverse content
offerings can use behavioral targeting across their various sites to offer their users more targeted
ads. Additionally, third party firms can specialize in parts of this process or can encompass all of it,
offering targeting across a broad range of publisher content. For example, data exchanges specialize
in data collection and analytics that they sell to advertisers. More comprehensive third party
advertising networks (ad networks) can handle both the collection, analytics, and servicing of the
P a g e | 2
ads. This study focuses on transactions involving the final advertiser, because that is the market in
which the value of advertising is determined.
Because behavioral targeting makes use of predictive data derived from users online
behavior, the practice has raised privacy concerns. To date, however, there has been no hard data
about the effectiveness of third party behavioral targeting or its importance to content providers.
This study seeks to fill that gap, providing data about advertising prices and revenues for a sample
of advertising networks.
The study surveyed members of the NAI seeking data for different types of targeting. We
obtained data from 12 ad networks, including nine of the top 15 ad networks by total unique
visitors according to comScore's December 2009 rankings.2
Our survey questions were designed to
obtain quarterly data on pricing (CPM data), conversion rates, and revenues across the various
types of ad segments (run of network advertising, behavioral targeting, and retargeting). Table 1
below summarizes the results of the survey. The data is presented in aggregate to protect
2. Press Release, comScore Releases December 2009 Ranking of Top Ad Networks, available at http://www.comscore.com/layout/set/popup/layout/set/popup/Press_Events/Press_Releases/2010/1/comScore_Releases_December_2009_Ranking_of_Top_Ad_Networks (last visited Mar. 1, 2010).
P a g e | 3
TABLE 1 SUMMARY OF KEY SURVEY RESULTS
FULL YEAR 2009
AVERAGE CPM (WEIGHTED BY BT REVENUE) Run of Network $1.94 $1.98 $1.89 $2.06 $1.98 BT $4.09 $4.22 $4.07 $4.11 $4.12 Retargeting $3.00 $3.12 $3.13 $3.02 $3.07 BT Avg. Relative Price Over RON Ads (X Greater) 2.77 2.71 2.79 2.46 2.68 Retargeting Avg. Relative Price Over RON Ads (X Greater) 1.98 1.84 2.11 1.59 1.88
AVERAGE CONVERSION RATE Run of Network 2.1% 3.6% 2.2% 3.1% 2.8% BT 5.5% 8.8% 6.4% 6.6% 6.8%
REVENUES Total Ad Revenue ($ Millions) $708 $780 $795 $1,040 $3,323 Percentage Attributable to BT (Aggregated Across Firms) 16.2% 17.2% 18.3% 19.4% 17.9% Avg. % of Display Ad Revenue Used for Inventory Costs 54.7% 56.9% 53.0% 53.6% 54.6% Avg. % of Display Ad Revenue Used for Data Costs 8.5% 8.8% 9.1% 9.4% 8.9%
The results lead to 3 major conclusions. First, advertising rates are significantly higher for
behaviorally targeted (BT) ads. The average CPM for BT advertising is just over twice the average
CPM for run of network (RON) advertising. On average across participating networks, the price of
BT advertising in 2009 was 2.68 times the price of run of network advertising. Second, advertising
using BT is more successful than standard run of network advertising, creating greater utility for
consumers and clear appeal for advertisers. Conversion rates for BT advertising are more than
twice the rate for RON advertising. Third, a majority of network advertising revenue is spent
acquiring advertising inventory from Web content and services providers, making BT an important
source of revenue for publishers as well as ad networks.
The rest of the report is organized as follows. The next section outlines the methodology of
our survey. Section III presents the results in greater detail. Finally Section IV discusses the key
conclusions from this work.
P a g e | 4
II. METHODOLOGY OF STUDY
Over the course of two months, we spoke with representatives from a number of ad
networks, all of whom were members of the Network Advertising Initiative (NAI). These
networks graciously provided crucial background information on the role of BT. They also provided
guidance in crafting a survey that was likely to be effective in obtaining data to assess the value of
BT to networks and publishers. The survey had to be general enough to encompass the variety of
business models seen in the industry, but specific enough to assess the primary research question:
what is the effect of BT on advertising revenues and rates?
Twelve NAI member ad networks provided data in response to the survey.3 Nine of the
twelve participants appeared in the top 15 total unique visitors list according to comScore's
December 2009 rankings.4 These nine firms averaged 159 million unique visitors in December of
2009 and reached an average of 78 percent of the total U.S. online population.5 The remaining three
ad networks are not as large, but provide some representation for the smaller networks in the
marketplace. Study participants had total ad revenues of over $3 billion in 2009, accounting for
approximately 40 percent of total Internet display advertising revenue.6
The survey questions requested several key pieces of data that are essential to determining
the value of behavioral targeting. The data were then compiled, and are presented in an aggregated
form to protect participants confidentiality. The survey requested the following:
Thus, the sample provides
a reasonably accurate view of the overall marketplace for behavioral targeting.
3. All current NAI members were eligible to participate in the survey. 4. Id. 5. Id.; Navigant Econom