Targeted information and limited attention

• Author: Shuo Liu,Andreas Hefti
• Published date: 01 June 2020
• Date: 01 June 2020
• DOI: http://doi.org/10.1111/1756-2171.12319

RAND Journal of Economics

Vol.51, No. 2, Summer 2020

pp. 402–420

Targeted information and limited attention

Andreas Hefti∗

and

Shuo Liu∗∗

We examine the implications of limited consumer attention for the targeting decisions of com-

peting firms. Limited attention alters the strategic role of information provision as firms may be-

come incentivized to behave as mass advertisers, despite perfect targeting abilities. We analyze

the consequences of limited attention for targeting, strategic pricing, market shares, attention

competition between firms, and the value of marketing data to firms. Accounting for limited at-

tention in an otherwise standard targeting framework can explain several recent key issues from

the advertising industry, such as consumer-side information overload or the increased usage of

ad blocking tools.

1. Introduction

Modern information systems, above all the Internet, allow sellers to gather an enormous

amount of data about their potential customers. In principle, such data allow firms to target their

ads toward precisely identified subsamples of consumers. Advertising research leaves little doubt

that targeted information has become a major source of advertising revenue. Large businesses

such as Axciom, IRI, and Nielsen earn their money by selling consumer data to individual com-

panies. Sponsored search advertising, allowing firms to advertise to consumers who indicate an

interest through their web search queries, has become “the largest source of revenues for search

engines” (Ghose and Yang, 2009). According to the annual report by the Interactive Advertis-

ing Bureau (IAB), search advertising alone already has a steady 40% of total digital ad revenue,

which was about $ 6.76 billion in 2006 and nearly tripled to $ 18.81 billion in 2014. Simi-

∗University of Zurich, Zurich University of Applied Science; andreas.hefti@econ.uzh.ch.

∗∗Peking University; shuo.liu@gsm.pku.edu.cn.

The authors are grateful to Jean-Michel Benkert, Alessandro Bonatti, Andrea Canidio, Fabrizio Germano, Julia Grünseis,

Igor Letina, Armin Schmutzler, Aleksei Smirnov, Philipp Strack, and seminar participants at the University of Zurich,

Swiss IO Day 2015 (Bern), the EEA Annual Congress 2016 (Geneva), the EARIE Annual Conference 2016 (Lisbon),and

the XXXI Jornadas de Economía Industrial (Palma) for discussions and useful suggestions. They thank John Christian

and Danielle Adams-Hausheer for proof-reading. They are also very thankful for the many helpful suggestions and

critical remarks by two anonymous referees and the journal editor, Mark Armstrong. Liu acknowledges the financial

support from the Key Laboratory of Mathematical Economics and Quantitative Finance(Peking University), the Ministry

of Education of China. Hefti thanks Ines Brunner for her support.

402 © 2020, The RAND Corporation.

HEFTI AND LIU / 403

larly, sponsored advertising has grown annually by around 30% from $ 1.12 billion in 2011 to

$ 1.88 billion in 2013, and is expected to grow further.1

Given the rich datasets about a firm’s potential customers and the broad array of information-

sharing technologies, including email, SMS, tweeting, and social networks, it may come as a

surprise that several recent press releases have substantially challenged whether real-world tar-

geting behavior really is beneficial for firms and consumers.2If the data about consumers is so

comprehensive and detailed, and tailored advertising is better and cheaper as ever before,3why

do so many consumers persistently complain about being overwhelmedby ads that seem to be of

little relevance to them, whereas firms are apparently targeting their messages far less accurately

than they could?4

In this article, we show that such puzzling observations can be explained by means of a

stylized competition model of targeted information when the standard presumption of unlimited

consumer attention is relaxed. Attention, as a central concept of cognition, has long been investi-

gated by psychology and neuroscience. Beginning with Miller (1956), there isextensive evidence

that people have limited attention, meaning that they fail to recognize all the information they

are exposed to once a certain threshold is reached. Given the superabundance of information

stimuli generated by modern advertising, limited attention, therefore, seems a particularly rele-

vant concept in a competition context. Indeed, the dominant and persistent preoccupation of the

advertising industry that a seller’s messages could simply be overlooked by the consumers they

were intended for implies clearly that limited consumer attention is a critical factor in markets

featuring targeted information.5

Main results. We study the effectsof limited consumer attention in a targeting context by

adopting a standard duopoly Hotelling framework with ex ante uninformed consumers. Besides

competing in prices, both firms need to choose which consumers to target with their messages,

where targeting is a costly activity. Consumers, in turn, always choose their best perceived prod-

uct. The standard (implicit) assumption of unlimited attention declares that consumers are al-

ways able to register and memorize any piece of information they are exposed to. By contrast,

attention-constrained consumers may perceive only a subset of all the information available to

them. Applied to our duopoly framework, the notion of limited attention means that consumers

will perceive at most one of the firms when making their final purchase decisions.

Our main results, Theorems 1 and 2, show that consumer inattention has a profound impact

on the equilibrium behavior of the firms. In particular, with attention-constrained consumers

and sufficiently low information costs, firms strategically choose to behave as if they were mass

advertisers, flooding the entire market with their messages, without considering consumer pref-

erences in detail, and despite an infinitely precise targeting technology. This is antipodal to the

standard prediction resulting from unlimited attention, where competition disciplines both firms

to send their messages exclusively to their prime consumers (i.e., the consumers to whom the

respective firm offers the most preferred option), even in presence of an arbitrarily small in-

formation cost. The driving force behind this discrepancy is that information targeted at prime

consumers works as an efficient shield against wastefulbusiness-stealing if and only if consumers

pay attention to every piece of information they see. In other words, limited attention annihilates

1Data available online on www.iab.com/insights and www.emarketer.com. See Yao and Mela (2011) for similar

facts on the importance of sponsored search in advertising, and Evans (2009) for a general survey on advertising data.

2Examples include the articles “Does sponsored content work for advertisers?,” Wall Street Journal, March 23,

2014 and “Does targeting work?,” The Ad Contrarian, Feb01, 2012, and the IAB report “Online consumers view and

usage of ad blocking technologies,” Sep 2014. Farahat (2013) showsthat previous studies on targeted advertising may

have largely overestimatedits effectiveness due to not accounting for selection bias.

3See, for example, “How online advertisers read your mind,”The Economist, Sep 2014.

4See, for example, a practitioner’s report posted on the website AdEspresso (available on www.adespresso.com/

blog).

5See, for example, “Advertising and technology,” The Economist, Sep 2014, Special Report.

C

The RAND Corporation 2020.