The effect of ad blocking on website traffic and quality

• Author: Joel Waldfogel,Johnny Ryan,Benjamin Shiller
• Published date: 01 March 2018
• DOI: http://doi.org/10.1111/1756-2171.12218
• Date: 01 March 2018

RAND Journal of Economics

Vol.49, No. 1, Spring 2018

pp. 43–63

The effect of ad blocking on website traffic

and quality

Benjamin Shiller∗

Joel Waldfogel∗∗

and

Johnny Ryan∗∗∗

Ad blockingsoftware allows Internet users to obtain information without generatingad revenue for

site owners, potentially undermining investmentsin content. We explore the impact of site-level ad

blocker usage on website quality, as inferred fromtraffic. We find that each additional percentage

point of site visitors blocking ads reduces its traffic by 0.67% over 35 months. Impacted sites

provide less content over time, providing corroboration for the mechanism. Effects on revenue

are compounded; ad blocking reducesvisits, and remaining visitors blocking ads do not generate

revenue. We conclude that ad blocking poses a threat to the ad-supported web.

1. Introduction

Intellectual property is, in general, protected by a combination of law and technology that

may together allow sellers to appropriate enough of consumers’ willingness to pay to finance

the creation of new products. The development of various digital technologies over the past few

decades has undermined sellers’ ability to appropriate consumers’ willingness to pay, raising the

possibility of a diminished supply of new information and entertainment products. For example,

new digital technologies have challenged traditional revenue models through direct piracy (e.g.,

Napster), diversion of attention to aggregators rather than the underlying information sources

(e.g., Google News), and pure-playadvertising platfor ms such as Craigslist, whichdivert revenue

∗Brandeis University; shiller@brandeis.edu.

∗∗University of Minnesota and NBER; jwaldfog@umn.edu.

∗∗∗PageFair; johnny@pagefair.com.

Wethank two anonymous referees and the Editor for very useful suggestions, and colleagues at Harvard Business School’s

Digital Initiative for lively discussions. We also would like to thank our discussants, Ana Gazmuri, Josh Gans, Bradley

Shapiro, and Ken Wilbur,for constr uctivesuggestions. Last, we would like to thank participants for helpful suggestions

at numerous seminars [Brandeis University (2016), Carlson School of Management (2016), The FCC (2016), Harvard

Business School (2016), MIT (2016), and NBER Productivity Seminar (2017)], and Conferences [IIOC (2017), NBER

Economics of Digitization (2017), Toulouses IDEI-TSE-IAST (2017), and ZEW ICT (2017)].

C2018, The RAND Corporation. 43

44 / THE RAND JOURNAL OF ECONOMICS

from newspaper classified ads (Rob and Waldfogel, 2006; Oberholzer-Gee and Strumpf, 2007;

Seamans and Zhu, 2014).1

Despite these and related challenges, much of the commercial information on the web is

provided under ad-financed business models. Although a growingnumber of newspaper sites have

moved behind paywalls (Chiou and Tucker, 2013), most other commercial information sources

rely on advertising. The finance of information production with ad revenue has come under

threat with the development of ad blocking technologies. Ad blockers, such as Adblock Plus,

are browser extensions that allow users to browse the web without having to see sites’ ads. As a

result, many site visitors obtain information without seeing ads, clicking through to advertisers’

sites, or otherwise generating revenue for site owners. The use of ad blockers has grown sharply

in recent years. By some estimates, roughly a quarter of site visitors used ad blockers by 2016, a

magnitude of ad blocking that may produce a serious threat to site revenue and, if revenue losses

undermine investment, a possible threat to consumers’ access to appealing content.2This is the

question this article seeks to explore.

Weare not the first to consider this problem. Anderson and Gans (2011) develop a theoretical

model of the impact of ad-avoidance technologieson content markets that is useful for guiding our

analysis. Bergemann and Bonatti (2011) and Johnson and Myatt (2006) also use similar models.

Adapting their setup to this context, users choose whether to incur the one-time cost of adopting

an ad blocker, balancing the cost of adoption against their distaste for ads. Ad-intolerant users are

more likely to adopt, with two consequences.3First, ad blocker usage by a site’s visitors reduces

the site’s revenue if at least some of those users would have visited the site in the absence of

ad blocking. The relevant mechanism, as in the traditional literature on the relationship between

intellectual property revenue appropriation and supply, is that reduced revenue may undermine

a site’s ability to invest, which could manifest itself as a diminished site that is less appealing to

potential visitors. Webusers then visit the degraded site less, reducing the site’s traffic. Second,in

the presence of ad blocking, a site’s remaining revenue-generating visitors are less ad-intolerant,

leading the site to run more ads, increasing the nuisance cost of visiting the site.

Studying these mechanisms empirically requires data on site trafficand site-specific measures

of ad blocker usage. This is challenging, but wehave obtained unique, proprietary, and site-specific

data on the share of site visitors using ad blockers at a few thousand sites, matched with Alexa

traffic data. These data allow us to explore the impact of ad blocker usage on site trafficbetween

2013 and 2016. We also explore the mechanism linking ad blocking with traffic using data, from

the Internet Archive, on the number of ads at these sites, as well as the amount of content (i.e.,

number of words), at the beginning and end of the sample period.

Aggregate ad blocking has grown rapidly between 2013 and 2016. During this period, we

find that sites with a high proportion of ad blocking visitors experience deterioration in their traffic

ranks, relative to sites with fewer ad blocking visitors. In our main analyses, we find—using a

conservative approach—that each additional percentage point of site visitors using ad blockers

raises (worsens) a site’s traffic rank by about 0.6% over a 35-month period. We have two ways

of dealing with possible concerns about unobserved heterogeneity. First, in contrast to our basic

result—worsening ranks at sites whose users block ads—those same sites did not experience

worsening ranks prior to the widespread use of ad blocking. Second, instrumental variables

based on the geographic proximity of sites’ users to the source of the dominant ad blocking

software (Germany) deliver consistent, albeit weaker, results. Data on site content supports our

interpretation; sites experiencing greater ad blocking decrease their word counts over time.

1Several historical examples exist of technology that allowsconsumption without compensation to rights holders,

including TiVo and Aereo.

2www.emarketer.com/Article/US-Ad-Blocking-Jump-by-Double-Digits-This-Year/1014111.

3Anderson and Gans (2011) also outline conditions for an equilibrium with the complete unravelling of the ad-

supported content market. Our exercise explorestwo of their major predictions. Our main question is whether the growing

adoption of ad blocking technology undermines content investment.

C

The RAND Corporation 2018.