Counting encounters on the Internet isn't as easy as 1-2-3
Quick, define the word "hit." Stumped? Try these: User. Visit. Log file. Session.
Chances are if you're on the World Wide Web, you've seen those terms before. But do you know what they mean?
For a medium with so much promise and opportunity, Web measurement is downright confusing.
There are a dozen or more companies offering counting, tracking, auditing and/or analytical services to marketers and media companies. They range from old names like Nielsen Media Research and Audit Bureau of Circulations to entrepreneurs like Internet Profiles Corp., NetCount and Interse Corp.
It's getting to the point where keeping the names straight is harder than understanding what they do.
"When I started in the publishing industry, there was ABC and BPA and that was all there was," sighed Susan Feigenbaum, a 10-year magazine veteran who's now chief researcher for Playboy's Web site. Now, "you're looking at [computer log] files and the abbreviations make absolutely no sense to you."
Said Jayne Spittler, senior VP-director of media research, Leo Burnett USA, Chicago: "I'm not only confused, I'm a little frustrated."
The Web holds enormous promise for marketers and media: the ability to communicate one-on-one with computer users (who, for the most part, are prime demographic targets); to present a broad range of information about a particular product or service; to customize a consumer's brand experience; and, eventually, to build a long-term relationship with a customer.
Everything a user does is recorded somewhere on a computer--each site accessed is a series of clicks or "hits" on a server and each bit of information downloaded is noted. For that reason, the Web, like no traditional medium, can offer precise logs of user activity: which sites a user visited, how long he spent, how he got there.
Web measurement, in theory, should be easy. In practice it's not. In what may be one of the greatest ironies of the interactive world, the sophisticated tools that monitor Web site activity cannot offer anywhere near the precise, accurate data that marketers think they do.
"A lot of the information [Web measurement tools] give is not only inaccurate, it's deceptive," said Emily Green, senior analyst with Forrester Research, Cambridge, Mass. The problem, she hastened to add, isn't that the tools don't work. "It's the limits of the network itself," she said.
When does an ad count?
Consider this scenario: Marketer A buys an ad, or in Internet terms, a "hot link" on a particular media site. The ad is positioned at the bottom of a page of text. For most computer users, seeing the ad means scrolling to the bottom of the page.
But what if the user doesn't scroll all the way down? What if he gets bored and decides to go elsewhere on the Web? Most measurement tools would count that page as having been viewed, and the ad as well.
Let's take it one step further. Suppose someone actually clicks on the hot link and jumps to the marketer's site. But before the marketer's home page fully loads into his computer, he changes his mind and stops the transfer. Again, for most Web measurement tools that's a successful link. Marketers have no way of knowing otherwise.
And unless a Web site registers users, there's no way to know the most valuable information: who those users are.
Due to privacy concerns (and the potential for password overload) few sites ask consumers to register. While some Web measurers such as Internet Profiles (I/Pro) have proposed a universal registration tool, implementation so far has been difficult.
But perhaps the biggest barrier to accurate Web site measurement is cacheing. Many large companies that offer Web access, as well as online services like America Online, store copies of popular home pages on their own servers, a practice called cacheing. That allows pages to be served up more quickly than if they were accessed directly each time.
The problem is that all the traffic to those popular sites is measured not at the site itself but at the computer cacheing the page.
"The AOL audience for us is huge, but I don't see the hits," said Eileen Kent, VP-new media for Playboy Enterprises' publishing group. "No piece of software is going to capture your traffic [from there] because it's not hitting your server."
It's a double-edged sword for Web sites, however, because, as Forrester's Ms. Green puts it: Cacheing "manages the traffic load on the 'net and lets popular sites stay popular" by allowing them to load quickly.
The solution is for the companies that cache to open up their server logs to Web measurement companies. It's an answer that appears easy on the surface but is in actuality a statistical nightmare.
Industry pushes for definitions
Before those kinds of problems can be solved, however, the industry must agree on how to define terms associated with Web measurement. A user or a visit can mean different things to different measuring systems, making comparisons between sites practically impossible.
One word that has been much maligned is "hits."
Hits indicates the number of graphical or text files a server downloads. A home page could contain five to 10 files or more, meaning one person calling up a page on the Web could generate that many hits.
"Hits are not the true measure of advertiser activity. They are, unfortunately, the only thing that seems to be out there today," said Bruce Judson, general manager of Time Inc. New Media and chairman of the MPA task force on Internet advertising measurement.
Despite the industry backlash, most Web tracking tools still count hits, then use statistical and modeling tools to generate data on "visits" (separate accesses by one user to a Web site) and "sessions" (how much time the user spends within the site).
Not that there isn't a drive toward standards, or at least comparability. Last week, the advertising industry's Coalition for Advertising Supported Information & Entertainment and the Advertising Research Foundation released a set of guiding principles for interactive media measurement.
The working paper, as it is known, calls for third-party measurement and/or auditing, the measurement of actual users and the number of times they access a medium, and total medium measurement.
"We're making a call for a single understanding of what is involved," said ARF President Mike Naples. "We don't want every home page to have a different measurement approach."
This summer, the Magazine Publishers Association issued its own guidelines, calling for much the same information.
Mr. Naples and others agree the Web measurement industry is far from a consensus, let alone a standard.
"It is at a very early stage," Mr. Naples said. "We are collectively trying to decide what makes sense."
When will there be a standard?
When will there be a standard? Possibly never.
"I was hoping you'd tell me," said Playboy's Ms. Feigenbaum when asked her opinion. Sorry, Charlie, but media research history says it isn't going to be so. Just as broadcasting has its Nielsens and Arbitrons and the print media have their ABCs and Simmonses, the Web will have more than one source of tracking information.
"Part of the confusion is that people think there will be a one-size-fits-all solution that will solve everybody's problems," said Burnett's Ms. Spittler.
That doesn't mean there won't be a dominant information provider, however, and many think I/Pro, because of its early start and its alliance with Nielsen Media Research, may be it.
I/Pro has about 75 clients for its various services, including Playboy, Yahoo Corp., Netscape Communications Corp. and Microsoft Corp. And Nielsen makes no bones about its desire to be a dominant provider of interactive media measurement services, just as it today dominates the TV measurement business.
"I wouldn't be disappointed if that happened," said David Harkness, senior VP-planning and development at Nielsen Media Research.
Tools of the trade
The measurement market seems to be lining up on one of two sides: tools for counting users or tools for auditing the counts that others provide. Some companies are providing both services, which could ultimately be a conflict of interest.
"The trouble with this whole industry is . . . people are mixing and matching," said Richard Tahta, president of Market Arts WebTrack, New York, developer of an auditing system the Audit Bureau of Circulations will use to verify Web traffic reports starting this month.
I/Pro, San Francisco, provides both counting and auditing services, as will NetCount, a Los Angeles-based spinoff of Web site developer Digital Planet that plans to unveil its Web tracking system later this month. Nielsen Media Research will co-market I/Pro's I/Audit and I/Count services.
Pursuing a separate course are companies that market software for Web sites to install on their own servers. Unlike I/Pro and NetCount, which require site owners to give up data from their server logs, these software products, including Sunnyvale, Calif.-based Interse Corp.'s market focus, let the site do its own tracking.
The advantage to this method is that sites get to keep much of their data confidential. But under the recent CASIE and MPA guidelines, they will still be subject to a third-party audit.
Even at this early stage, there are those who are critical of the direction Web measurement is taking. Much of that criticism centers on the role Nielsen wants to play.
"For Nielsen, the whole world looks like a TV set," scoffed Doug McFarland, a former Arbitron executive now directing marketing for FreeMark Communications, Cambridge, Mass.
"I cringe when I hear the word `audience,' " said Donna Hoffman, associate professor of management at Vanderbilt University and co-director of the university's Project 2000 Internet study. "To have the Web user described as a member of an audience is a direction we don't want to go." (Ms. Hoffman, it should be noted, is working with Nielsen and industry consortium CommerceNet on an upcoming Internet demographic study.)
Nielsen's Mr. Harkness said being a giant in a fledgling industry is "nothing but a giant positive."
And with all the hubbub in the advertising industry, it's easy to forget that Web measurement is about more than setting rates for online advertising. True, Web sites, anxious to make money, need to show media buyers the value of their ad placement.
But counting users and visits is only the first step. Knowing how many people access their site and where they come from, marketers can begin to do sophisticated mapping and targeting, providing content more appropriate for each individual user.
"If we're concerned with Web tracking at the moment only to set ad rates, it's basically because the other possibilities are at the earliest stages of development," said ARF's Mr. Naples.
Figuring out exactly who is on the Web is proving to be one of the most elusive tasks in the Web measurement industry.
"Right now many of us are working in the twilight zone on some of these issues," said Craig Gugel, exec VP-new media and interactive research at Bates USA, New York, and chairman of ARF's interactive media research committee.
Until systems like I/Pro, I/Code or NetCount's planned demographic analysis tool prove their viability, marketers and media will remain in the dark about one of the most promising aspects of interactive media.
Copyright October 1995 Crain Communications Inc.