Ad measurement sparks big debate

By Published on .

Around New York's Silicon Alley, Web ad measurement starts more arguments than impeachment. There's a good reason for this. When you buy print or broadcast ads, you know what you're getting and you can measure the medium's performance. You can't do this with Web banners.

It's part of the nature of the Internet. Some users leave pages before an ad appears and never see it. Others use text-only browsers or software designed to strip out ads.

Still other page requests come from "spiders" or "robot" programs that are testing links, collecting data or just checking connections and page load times.

On the other hand, some servers save or cache busy pages for delivery to hundreds of users, but the ad moves just once so the ad server counts it as one impression.

Opposing sides

The differences can be significant. Jim Nail, a senior analyst at Forrester Research, Cambridge, Mass., says third-party measuring services such as Matchlogic count 20% fewer ads than the numbers delivered on publishers' reports.

Both sides have firm positions. Publishers want to charge advertisers every time they serve a page. Advertisers want to pay only when an ad has been moved and cached on a user's machine.

The fur flies on mailing lists like Online Ads, run by Tenagra Corp. of Houston, or the Internet Advertising Discussion List, run by eyescream interactive, a Web ad agency in Portland, Ore. Here, advertisers compare publishers' demands to magazines that print blank pages and try to charge for what should have gone on them. Publishers say caching means advertisers are getting all sorts of bonus circulation they won't pay for.

Then there's a third school of thought, which argues that using a cost per thousand, or CPM, measurement is clueless. These advertisers prefer a "pay per action" model, in which publishers are paid only when users respond to an ad in some way, such as by clicking on the banner, filling out a form or actually buying something.

Microsoft's LinkExchange unit offers such a program, called ClickTrade, but even here there are arguments. Publishers are accused of inflating clicks with software.

Audits and third-party data

Another answer is to get an audit from I/Pro, ABC Interactive or BPA International. Ad servers and large sites are often audited, but cam-paigns seldom are, says Dick Bennett, ABC senior VP-audit services.

You can also get uniform campaign data by having a third-party server, such as Matchlogic, AdForce or DoubleClick, deliver the ads. The problem is that both solutions cost money, says Adam Boettinger of eyescream, and that reduces a campaign's efficiency.

The logical solution is to set some standards. That's what the FAST Forward group, which supports campaign audits, is now grappling with, under the leadership of Rich LeFurgy, chairman of the Internet Advertising Bureau in New York.

The need is urgent. Total Web ad spending in 1998 will probably come in at more than $2 billion, and it could double this year. Imagine spending $4 billion on something we can't measure.

In cases like this, my instinct is to follow the golden rule: He who has the gold makes the rules. Ultimately, it's the advertiser who must be satisfied.

Dana Blankenhorn is an Atlanta-based freelance journalist who specializes in e-commerce issues and is publisher of the Web site A-Clue.Com.

Most Popular
In this article: