Advertisers buy people, not ads, so measurable impressions count more than ad impressions.

The debate continues over ad server discrepancies and whether ad deliveries or requests are useful indicators of audience reach. The justifications for miserable response rates grow more creative, but still tell us little. Albert Einstein described this state of affairs when he told us that insanity is doing the same thing over and over again and expecting different results.

Before we hop on the bus to the nearest psychiatric facility, let's look at the problem differently and treat the causes, not the symptoms.

1: Ads are not people, and advertisers don't buy ads. It's useful to know how many opportunities the "average" viewer had to be exposed to an ad. Far more relevant is identifying a site's visitors and measuring how ad exposures over time affect behavior.

Audience segmentation and targeting techniques -- including, but not limited to, behavioral targeting -- enable publishers to identify audience members as shoppers, enthusiasts and so on, based on site navigation and other behaviors and characteristics. But you have to close the loop by integrating ad views and interactions into your visitor profiles to get the full picture. When you tie ad data to visitor profiles, you can measure the impact of ad frequency on target audiences, and determine total reach more accurately than through your ad server or targeting tool alone.

2: Economics and accountability make uneasy allies. Some imaginative publishers effectively balance their economic realities with the demands of advertisers for accountability. It's not easy to tell an advertiser that the best way to meet a campaign goal is to reduce impressions, but sometimes it's the right thing to do. Buyers are seeing success with audience targeting, and they are increasingly willing to pay higher CPMs to reach a greater percentage of their target at realistic frequencies.

We have participated in dozens of campaigns that produced outstanding results by targeting audience segments built from navigation-based behavior, and from combinations of behavior, demographics and other data in the publisher's possession.

One factor present in every success is risk. It takes a bold publisher to tell its most committed advertisers that they can improve on results that are already good by adding audience filters that subtract from the target population. The publisher assumes at least three risks:

• Loss of immediate revenue by limiting audience size and impressions
• Fewer total responses, especially if also imposing a lower frequency cap
• Loss of future revenue if response falls.

In essence, the site is betting that targeting more precisely to audience interests will overcome their tendency to ignore ads. The publisher can't win this battle for attention alone. The agency helps beat the odds by providing great creative tailored to a precisely targeted audience.

Buyers demand an end to the practice of blindly delivering 90 percent of campaign impressions to the 10 percent of the audience who generate most of the site's inventory. Success, coupled with shortages of key inventory, is spurring them to pay more for the opportunity to get meaningful, measurable reach.

3: Doing well by doing good. The price a publisher pays for selling fewer, well-targeted impressions has to be made up somewhere. We think the day is coming when market pricing will be based on availability and delivery of identifiable audience. Here are two recent examples of direct reponse and branding efforts:

Case No.1: Online selling keeps getting better. A marketer of fine art and furnishings for new homes discovered a trove of new home buyers visiting a site featuring content on finding, buying and furnishing a new home. The publisher's behavior-based audience segment consistently outperformed other vehicles. After six successful months, the marketer challenged its media partner to do even better by excluding households with less than $100,000 annual income.

Negotiations concluded with agreement to decrease audience size by 20 percent, drop the frequency cap from six to five, and increase CPM by 50 percent. The publisher got a rich 12-month contract and the marketer realized increases in number of sales, average sale size and repeat sales. The publisher resold audience impressions that the marketer had let go. Today the site is ahead in total revenue, advertiser satisfaction and audience loyalty.

Case No. 2: New campaign for an old brand. To put a new face on an old brand, a consumer packaged goods advertiser sought sites that reached aging baby boomers who were likely to remember the brand fondly from childhood. The goal was to inspire this audience to become fans all over again and introduce the products to younger family members.

The campaign ran broadly, with dismal results. The target demo encompassed such a wide variety of interests, incomes and lifestyles, that ads wound up confusing more people than they informed.

The agency next recommended an audience targeting strategy by which they would find people within the target demo likely to share the specific interests of the target. These included family travel, organized youth sports, and content about musical artists popular with both older and younger crowds.

Targeting older boomers by interest brought the publishers significantly higher CPMs. The payoff came in third-party verification that purchase intent soared among targets with younger children, grandchildren, nieces and nephews.

When you know your identifiable audience, it's no longer the ad impression that counts, but the measurable impression you make on the target. This is exactly what audience segmentation, targeting and measurement allow us to do. Take advantage now, before this inventory sells out, too!

Bennett Zucker is Executive Director of Customer Success for TACODA Systems in New York City.