By Alistair Barr
Retailers have long struggled to determine whether online ads
fuel sales in their bricks-and-mortar stores. Now, Google Inc. is
testing a way to solve that puzzle.
A pilot program launched by the Internet giant is helping about
six advertisers match the anonymous tracking cookies on users'
computers to in-store sales information collected by data providers
like Acxiom Corp. and DataLogix Holdings Inc., according to people
familiar with the test.
One participant in the program is the arts-and-crafts chain
Michaels Stores Inc., the people said. The other participants
couldn't be identified.
"We are running a number of tests to help clients use their own
sales data to measure how their search campaigns impact sales,"
said a Google spokesman.
Online advertising has grown into a $117 billion-a-year
business, and Google is the industry's leader, with ad revenue of
$50.5 billion last year.
The company's new pilot program involves AdWords, its biggest
advertising program, in which advertisers place links next to
Internet-search results. Google gets paid when users click on an ad
and visit the advertiser's website.
But while tracking clicks is an effective way to measure a Web
ad's impact on online sales, the system breaks down when shoppers
enter a physical store, where their behavior is harder to
monitor.
"Clicks are just the beginning, and everyone knows that there's
more value to tying ads to purchases," said Benny Arbel, chief
executive of myThings, an advertising-technology company that works
with Google but isn't involved in the pilot program. "If Google can
demonstrate that people did not just click on an ad but that they
actually bought something, that is the Holy Grail."
Google advertising executive Neal Mohan told a conference
earlier this year that big brands spend more on television and
other traditional media because they aren't sure how to measure the
efficacy of their digital ads.
"The world of impressions, clicks and conversions doesn't
necessarily apply to what they are looking to accomplish," said Mr.
Mohan, who isn't involved in the AdWords test. "That's the $200
billion nut that is sitting out there, basically ready to come into
the digital sphere."
In addition to Acxiom and DataLogix, the other data handlers
participating in Google's program are LiveRamp Inc. and the Epsilon
unit of Alliance Data Systems Inc., according to people familiar
with the test. Those companies have access to troves of information
about Americans' purchases, through loyalty cards and other
marketing programs.
Acxiom and Epsilon declined to comment on whether they are
involved in the Google test, but executives at those firms
confirmed that they are working with other companies on linking
online ads to in-store sales. DataLogix and LiveRamp didn't respond
to requests for comment.
Google rival Facebook Inc. has been tapping data on
physical-store sales since late 2012 to demonstrate the
effectiveness of advertising on its site. Facebook works with
DataLogix, which says it has information on more than $1 trillion
of consumer transactions, and recently started a similar
partnership with Epsilon.
The data providers compile names, emails and addresses of
millions of consumers, along with in-store purchase histories.
Facebook, meanwhile, knows the identity of its users, their email
addresses and what ads they have viewed online.
To avoid breaching personal privacy, Facebook and DataLogix
don't share their data directly, according to executives at the
data companies. Instead, the two companies use an algorithm to
scramble their data into random numbers. Because both companies use
the same algorithm, information about a specific consumer is
assigned the same number in both sets of data, and can be matched.
The process, known as "hashing," allows the companies to share
their data without passing along any information that might reveal
a person's identity, the executives said.
Facebook takes those matched, anonymous profiles and divides
them into two groups of users--one that is shown ads, and one that
isn't. The data companies can tell Facebook whether the group that
viewed the ads bought more of the products being advertised.
Sheryl Sandberg, Facebook's chief operating officer, told
analysts last May that Facebook ads for Bud Light increased sales
by 3.3% in one study, generating a sixfold return for
Anheuser-Busch InBev NV on the cost of the ad campaign. The beer
company didn't respond to requests for comment.
Google's new pilot program, dubbed In-Store Attribution
Transaction Reporting in AdWords, works a bit differently,
according to the people familiar with it.
When a user clicks on an ad, Google sends an anonymous "click
ID" to the advertiser. The advertiser likely has a cookie on the
user's computer, and matches that cookie to the click ID.
Days or weeks later, the user might buy a product in the
retailer's store. The data company and the retailer can take that
purchase and link it back to the user's cookie. Then they match the
cookie to Google's click ID. Ultimately, Google can tell
advertisers which ads generated in-store sales and how much they
generated, the people said.
The Google spokesman said the company designed the test so that
Google never knows the identity of the user.
Currently, AdWords advertisers get reports on the performance of
search ads they've bought. There are columns for things like
click-through rates, cost per click and post-click information such
as whether people bought something on the website they were sent
to. The pilot program adds a new column that shows in-store sales
spurred by the ads, the people familiar with the situation
said.
Subscribe to WSJ: http://online.wsj.com?mod=djnwires