Online Advertising for Customer Retention Research by Pradeep K. Chintagunta, Jean-Pierre Dubé,
and Puneet Manchanda
Banner advertising helps companies retain customers by bringing
them back to a company's Web site faster and encouraging them
to spend more.
Total spending on Internet advertising now exceeds spending
on some traditional media, but despite the quick adoption of
online marketing by many firms, remarkably little is known about
the potential payoffs of such efforts.
Most online advertising exists in the form of banner ads, which
combine graphics, text, and a link to an advertiser's Web site.
Consumers access the advertiser's site by clicking on the banner
ad, which is referred to as "clickthrough." In the
early days of e-commerce, the fact that consumer behavior in
response to advertising could be measured instantaneously and
objectively by calculating the clickthrough rate was seen as
an exciting development. However, clickthrough rates typically
have been less than one percent of all exposures. In addition,
these rates only measure visits to a site, ignoring actual purchasing
behavior. Previous research has shown that few visits translate
into actual purchases.
University of Chicago Graduate School of Business professors
Pradeep K. Chintagunta, Jean-Pierre Dubé, and Puneet
Manchanda, together with doctoral student Khim Yong Goh suggest
that, as in traditional advertising, exposure to banner ads
may result in purchase behavior after a temporal gap.
"Most theories of advertising note that the effects of
advertising are not immediate," says Manchanda. "We
therefore wanted to link individual exposure to banner advertising
to individual behavior while allowing for a temporal gap. Our
approach expands upon the work of previous studies that have
only documented attitudinal changes as a function of exposure
to Internet advertising."
The authors report their findings in the new study, "The
Effects of Banner Advertising on Consumer Inter-purchase Times
and Expenditures in Digital Environments."
The authors find that banner ads are effective for bringing
existing customers back to a Web site sooner to make additional
purchases. Thus, for a given period of time, current customers
who were exposed to banner advertising are likely to spend more
money than customers who were not. The authors also suggest
that the industry-wide practice of judging banner ads by the
number of clicks they generate understates the effectiveness
of banner ad campaigns.
"If you just measure clicks, you are not capturing the
real effect of advertising," says Dubé. "What
you want to see is whether people are purchasing items."
Even though banner ads are typically regarded as doorways to
bring in new customers, the long-term viability of a Web site
depends on its ability to retain customers. Many industry studies
have shown that retaining current customers, relative to acquiring
new customers, is more profitable to a firm over the lifetime
of the customer. For online firms, the question then becomes
whether banner advertising can modify the behavior of repeat
customers as they become more experienced with the firm's Web
"Online advertising budgets have been shrinking since
the dot-com bust," says Dubé. "It's a known
fact that it's cheaper to market to a current customer than
a new customer. Our study shows you can use banner ads to stimulate
business in your repeat customer base."
Following the Cookie Trail
What distinguishes Internet advertising from traditional advertising
is the ability to match individual advertising exposure to individual
consumer behavior. If a consumer sees a television commercial
for a product, it is very difficult to then match up the commercial
with whether or not a consumer purchases the advertised product
in a store. The Internet allows researchers to put the whole
story together, from awareness building through actual purchasing,
going beyond what is possible in the traditional world.
The technology that makes this possible is based on small files
called "cookies" that are stored on an individual
consumer's computer once they visit a Web site. By keeping track
of cookies, firms have detailed data on when, where, and to
how many ads each individual cookie was exposed. This advertising
data can then be matched up with purchases made via that computer.
Advertisers can therefore return to their clients and let them
know which cookie actually resulted in a purchase, and how many
dollars that cookie generated.
Chintagunta, Dubé, Manchanda, and Goh use data from
an Internet-only firm that sold healthcare and beauty products
as well as nonprescription drugs. Their data spans a period
of three months during the third quarter of 2000. Most data
used for studying online environments feature browsing behavior
only. Their new dataset is unique because the authors are able
to measure individual stimulus (advertising) as well as response
(purchase visits and dollars spent).
The majority of the company's banner ads focused on brand-building,
and typically consisted of the name of the Web site and a line
describing the benefits of purchasing from the site. More than
eighty percent of the company's advertising activity during
this period was on portal and alliance Web sites such as Yahoo!,
America Online, Women.com, iVillage.com, Healthcentral.com,
and E*Trade. A given banner ad typically appeared on these sites
over several weeks.
To ensure that they only included repeat customers in their
analysis, the authors used data from customers that had made
at least two purchases from the site. The final sample consisted
of 2,192 cookies.
Industry measures such as clickthrough only account for direct
action, rather than measuring the awareness building that takes
place over an advertising campaign. The authors therefore allowed
for the possibility of customers seeing an ad, mulling it over,
and then returning to make a purchase after a period of time.
The behavior of repeat customers was measured using statistical
models that captured purchase timing (when to visit) and purchase
expenditure (how much to spend on a purchase visit). The authors
measured the effect of the following advertising variables:
the time between purchases and dollars spent; the amount of
advertising exposure; the time since consumers last saw an ad;
where they saw the ad; and the ad copy and graphics they saw
for each banner (the "creative").
The authors find that seeing the ads more frequently brought
customers back to shop sooner. The more recently consumers saw
an ad, the faster they came back to buy. Exposure to banner
ads at more Web sites also had a similar effect. However, exposure
to a higher number of ads with different creative treatments
delayed consumers' return to the Web site. For purchase expenditure,
the effect of advertising is small. Instead, the best predictor
of current expenditure tended to be the amount spent on the
last shopping occasion.
The authors also find differences among the shopping behavior
of repeat customers. Their data and statistical model indicate
that the Web site's customers can be classified into three different
segments. Customers in these segments are affected differently
by the frequency of banner advertising, how recently they saw
the banner ads, and the monetary value of their past purchases.
The largest segment consists of loyal but infrequent shoppers,
followed by a segment comprised of relatively frequent purchasers.
There is also a small segment of impulse shoppers who tend to
shop on the same day that they are exposed to a banner ad.
Overall, for most consumers in the sample, the authors find
evidence of a temporal lag between exposure and action. They
speculate that this gap exists because banner advertising acts
as a brand building tool and reminds consumers to visit a site.
Ingredients of a Good Campaign
The study provides insights into the nature of consumer response
to banner advertising. First, the authors find evidence of temporal
differences between exposure and behavior for a majority of
the consumers in the sample. This result implies that managers
can correctly evaluate the effectiveness of advertising campaigns
only if they account for this temporal gap.
Second, the time since last exposure and the number of creatives
has a much larger effect on purchase timing relative to the
number of exposures. Regarding advertising copy, exposing the
same consumer to several unrelated creatives may be less beneficial
than a single creative, assuming that all the creatives are
of the same quality.
"More companies are starting to realize that redundancy
is a good thing for their online advertising campaigns,"
says Dubé. "If you have a single, effective message,
it's easier for consumers to extract pertinent information from
your banner ad even as they are inundated with ads from other
In terms of the timing of advertising, it may be more useful
to expose consumers to a series of evenly spaced ads rather
than massed exposures. Given the strong same day purchase effect
resulting from advertising exposure, advertisers can potentially
use banner advertisements to smooth out sales and run special
Finally, there seem to strong positive benefits from ensuring
that customers are exposed to the same advertisement across
many Web sites.
"For the category of frequently purchased, nonseasonal
products in our study, steady and consistent advertising on
many different Web sites is the right managerial approach,"
In terms of the dollars spent on a visit, exposure to a variety
of creatives increases dollars spent while advertising across
many sites decreases dollars spent. However, in contrast, exposure
to more creatives delays repeat visits while exposure on more
sites brings consumers back sooner. Therefore, managers need
to optimize the number of creatives and advertising sites by
making the tradeoff between quicker visits and lower expenditure
per visit, or slower visits and higher expenditure per visit.
Pradeep K. Chintagunta is Robert Law Professor of Marketing
at the University of Chicago Graduate School of Business. Jean-Pierre
Dubé is Kilts Center Fellow and assistant professor of
marketing at the University of Chicago Graduate School of Business.
Puneet Manchanda is Kilts Center Fellow and assistant professor
of marketing at the University of Chicago Graduate School of