Buy a pound of meat at the butcher's and you know the scale's been calibrated and certified by the appropriate New York State authority. If you can demonstrate you were short-weighed, that is, the butcher gave you only 3/4 of a pound when the scale said it was a full pound, you have recourse. And so it goes pretty much wherever there are buyers and sellers. But not when you buy advertising on the World Wide Web. Ascertaining what constitutes fair, at least in a way that is borderline objective, requires some diligence on your part.

Measurement has been the focal point for controversy in interactive/online advertising almost since the outset of this rapidly evolving medium. The first serious attempt to address measurement head-on was in 1995; ages ago in Internet years. The Coalition for Advertising Supported Information and Entertainment (C.A.S.I.E.), a consortium representing advertisers, ad agencies, the media, and content creators convened expressly to forge agreement on just what was meant by "hits," "visits," "impressions," "page views," and other terms coming into the vernacular.

The one-to-one, on demand nature of content delivered on the World Wide Web, where many different outcomes - or responses - are possible, contributes to the measurement problem. So, while definitions are now less ambiguous thanks to C.A.S.I.E., the greater issue of which measurement data are best applied in a given situation remains murky.

Sensing the need for accountability, and armed with a new vocabulary to boot, a number of enterprising companies jumped in with software-based tracking solutions for Web sites hosting third party advertising. Many of these first generation measurement companies had very short life spans in both Internet years and real years. While their solutions offered some semblance of accountability for the individual sites on which the software sat, there was no carryover from one site to another. In a perverse way, these well-intentioned efforts actually made the measurement problem worse as Internet advertising grew.

A rating point scheme - verified by some independent "higher authority" - would be a nice, tidy way to rationalize both buying and evaluation of on-line advertising. Unfortunately, the nature of on-demand, interactive media renders such a one-size-fits-all solution impractical. The reason is that advertising on the Web can have many legitimate rationales and objectives applied to it. Whereas on TV, flights of :10, :15, or :30 commercials are bought mainly to sharpen a brand's image (a diamond is forever), or provide air cover for a sales promotion (big rebates at your Chrysler-Plymouth dealer), the Web offers many valid types of "responses." On the Web, it is appropriate to link advertising's value to the behavior desired from the user. One such behavior is the click-through. When the advertiser's goal is to drive as much traffic as possible to its site, the number of click-throughs generated by a banner on the hosting site can be a valid behavior-based measure. Some Web sites will sell their avails on a click-through basis; most do not. Files served - and banner ads, no matter how engaging, are just files - is the more common measure. Just because a file is served offers no assurance that it was viewed. Much like TV commercials that air when some of the audience is in the kitchen or the bathroom, advertisers seeking click-throughs but paying on a files served basis, are probably paying too much.

On the other hand a branding campaign built around on-line ads, running over a large number of Web sites in a defined span of time, can be a very cost-effective way of reaching a target market. Interestingly, if click-through is not central to the goal, the advertiser doesn't even need to have a Web site! Here, where it's total impressions that matter, the market has moved toward a cost-per-thousand pricing model that mimics that used by magazines, and prices have settled into a range. The more stratified the audience, the higher the price. So buying - perhaps through an ad-serving network such as DoubleClick or 24/7 Media - has become easier, right?

Not exactly. While today the largest on-line advertising dollar commitments come from major consumer brands, the important action soon will be on sites catering to narrower interests. Business-to-business advertising is where the Web offers the largest competitive edge. Smaller companies can mount serious, Web-centric, branding or promotion-oriented efforts with cost efficiencies impossible to attain solely in conventional media. But, paradoxically, getting what you pay for can be far more difficult in this part of the online world.

Media Metrix and NetRatings, the two major players left in the Internet measurement business, use different panel recruiting methodologies, and neither provides adequate mechanisms to track workplace usage. Many people using the Web for business purposes do their surfing from behind a firewall - a corporate security blanket - which panel-based surveys and server side traffic logs are ill-equipped to integrate into their ratings systems. (America Online users, or multiple individuals who share a computer, are similarly under-represented by panels.)

What can you do today to get a fair deal? Compile data from several sources and see how it fits with the objectives you've set for your online advertising program. Ask specific questions regarding which verified traffic log data will be made available. A lot of sites can capture data with high granularity but won't necessarily be forthcoming about sharing it unless pressed to. Buy on a trial basis if possible. A good account executive will want to show you that your advertising is delivering. You have bargaining leverage; negotiate.

Shop around, but be consistent in your comparison methods. The goal in uncovering the value proposition isn't getting your advertising cheap so much as it is to have confidence that the advertising you are paying for fairly tracks the types of responses that are meaningful to you.