• by David L. Smith , December 3, 2001

Last week, my esteemed colleague Jim Meskauskas published the second of a two-part treatise on destroying the impression as the basic unit of purchase in ClickZ.

While the article was interesting, we think he has it all wrong. We must continue to use the impression, as it is the basis for all metrics. Why should we? Because we can. The impression represents the finest granularity we can measure, so why not do so? All other measurements represent a variation of the impression with less control for the buyer over impact, cume, reach and frequency. Besides, all other measurements will revert back to the impression as a touch point anyway.

For example, let's take rating points. Whether you call them GRPs, TRPs or IRPs, they depend on impressions to develop the metric. The basic formula for a rating point is impressions/universe=rating points. What about reach and frequency then? Well, for those of you who have not examined this, reach is simply rating points netted down for duplication (uniques across sites). While a GRP represents gross ratings (a percentage of the target you are reaching), reach represents net rating points. Frequency is always related here to net vs. gross. In fact, RxF=GRP.

OK, so what about his other idea, that content pages in publications like Salon should be treated like magazine insertions, with a fixed position? This might please the advertiser looking for his ad, but probably not accomplish much else. By buying a unit larger than the impression, you lose control of whether your communication is effective. Someone will need to compute audience cume of the page (reach of this single vehicle), average frequency to see how many ads should run over the period of the sponsorship to prevent wearout, and you can be sure that at the end (and maybe at the start too), somebody will compute the expected and delivered number of impressions to get an CPM for the buy. As you see, we cannot escape the impression.

So, what should we be doing? We should be buying impressions - the smallest unit we can buy - and packaging it up the way that it will do us the most good to accomplish our goals.

For example, a buy on a single page will probably result in low reach and lots of frequency vs. a scatter buy on the same site. We do this for magazines since (a) we have no choice and (b) research shows that most readers read every page of a magazine at least once. That does not happen on the Web. The other issue with sellers giving buyers exclusive on a page we have written about before. If all of the good pages go to individual buyers exclusively, there will be no way for an advertiser who wants reach (consumer goods, etc.) to achieve the reach potential of the site. It would be like buying NBC on a scatter basis and getting no prime time.

We need sellers to provide a true rotation when we buy ROS. This means that the top pages they’ve sold in business development or strategic deals need to be made available, so we can really have the potential to reach all of the sites' visitors.

Once we get reach and frequency tools, we need to figure out how to maximize the cume of various sections of a site. In other words, how many impressions or GRP's does it take before we have reached the uniques that are affordable and we are just piling on additional frequency? We do this all the time in radio. Why not the Web? Research has shown that you need to achieve a frequency of 4-7 to achieve branding. How many impressions do we need on a site or a section of a site to achieve this frequency level? It would be nice to know.

Within a site, we need to make sure that we get enough of a schedule on the big audience pages to have a chance to get close to the cume potential of the site (what Jupiter/Media Metrix-JMM mistakenly calls reach on the Web. This is the best argument I have seen for day parts on the Web. Not the day parts we use in radio and TV (which are different, by the way), but a day part description that is unique to the Web. For example, different times of the day probably accumulate visitors in different ways. As such, you may need a different approach for each day part on a site relative to the percentage of the cume potential you buy in that day part to achieve the overall site cume potential. Another cut at day parting could be habits of the visitors on the site. For example, at the recent iMedia conference, the JMM people stated that much of the purchasing done by upscale males is done at work. So if you are deploying a shopping bot, or trying to sell something online, this is a day part you should be emphasizing.

So let's end the debate on impressions. We have the unit. It is granular. With it we have control. Real soon now, according to the IAB, we will even have a universal definition that all can accept. Then we need to get on with the other metrics that give us better data on how impressions accumulate so that we know how to deploy them.

David L. Smith is President of Mediasmith, Inc., the Integrated Solutions Media Agency based in San Francisco and New York.