The Web suffers from not having standardized media metrics that are comparable to other media. It is also the toughest medium for an agency to reconcile for payment due to a lack of standards in metrics from the ad servers. Isn't it time for "the most measurable of all media" to stand up and be counted?
The process for simplifying this lies in the hands of the research and third party server companies. And they need to deliver soon, or the industry will find other solutions. Without standardization, the Fortune 500 and the Global 2500 will find it easier to wait until they know what the are getting for their money. Without the big, mainstream companies, there is no "ad model." Without the ad model, we are all out of business.
PRE AND POST
We need solutions in two areas, pre buy and post buy.
For pre-buy, we need GRPs (at least as broad as Nielsen TV demos, but hopefully deeper, more like MRI data. Then, we need pre-buy reach and frequency data. For all demographics. The data that Jupiter Media Metrix reports out now as reach is actually audience cume potential not the "opportunity to be exposed" provide by a purchased schedule. We want to know the impact of a campaign, not the potential if we were hard coded into every page of the site. Hardly a useful statistic.
We also need the ability to put this together with traditional media R/Fs to look at combined reach and frequencies on 1+ and 3+ bases, frequency distribution, etc. We need this for the US at first and then for other countries soon. Jupiter Media Metrix, Nielsen NetRatings and (hopefully) DoubleClick's NetScore should be able to do this. Of course, they will need to define universes for each demo they wish to report on-By country. Now, how difficult can that be? Once we have this, we will need them to break out user data by country. This will give us composition by demo. We apply this to the impression schedules that are proposed to us and then we'll have GRP measurements, whether they report them or not (impressions/universe x 100=# GRPs) with real GRP estimates of schedules. It is a short step from this to reach and frequency, either from the individual companies or from third party interfaces like Telmar and IMS. We can then use the Methringham formula or others to combine with traditional media reach and frequency.
The benefits to this will be enormous and immediate.
* Buyers will see what the relationship is of the schedule they are buying to the site as a whole. I have a sneaking feeling that we are under buying a lot of sites. Except with business development deals, where sites are probably overbought. Would be nice to know, wouldn't it?
* Planners and media management will quickly see that they might be allocating 4000 GRPs to television and 40 to Interactive annually, and adjust their budgets to something more realistic. Importantly, though, it will also help reveal an opportunity for advertisers to reinforce their message against the light TV viewers.
* Sites will have more information on how to sell a schedule that has the best chance optimizing their audience through reach/frequency analyses. This could result in bigger buys on their site.
* Advertisers will understand what the Web adds to their schedule. For the big advertisers, it will not add much, if any, reach. But, through demographic and geographic analysis, it may help to balance out impressions against key groups-those light viewers mentioned above.
It is possible that we will get this capability soon from the eTelmar IMC suite, if the research vendors agree to work with Telmar on this. We look forward to finding out.
We have also It has also failed to deliver on basic agreement of an impression and have any comparability between the counting differences between the ad servers.
Simply said, there are currently FIVE different ways that ad servers count impressions. So when an agency gets reports from the site reports, they rarely agree with the agency data from their third party server. We need one definition, without which we will not have comparability in counting from site to site. This was the primary reason why agencies went to third party serving. Sure, facilitating trafficking is nice, but any agency in its right mind would give this up if the sites could do it competently and quickly.
In addition to basic impression counting on an "apples to apples" basis, use of a third party server gives the most complete post-impression data available which, when combined with post-click data, provides the best CPW (Cost-Per-Whatever a client is trying to measure) data possible.
Back to the five different ways that ad servers count impressions. From an agency standpoint, the goal is OTS or opportunity to see, same as with other media.
* Some sites count the impression at page build time. This is fine for hard coded ads but not satisfactory for served ads.
* Some do it at the ad request for the ad image, whether or not a server responds.
* There is support for what is called the "302 Redirect". This is an HTML Web server call that involves the browser request for a page that contains a banner. The server then returns HTML instructions. The browser then requests the ad. The request is sent to the counting server that returns a "location redirect URL". The impression is counted when the "counting server" receives the request. The browser then makes final request to the "image server" for the actual image. This methodology is used by Engage (AdKnowledge) and Yahoo.
* Ad fully sent from the image server. This methodology is used by DoubleClick for both DFA and DFP and by AdForce.
* The ideal involves proof of a fully downloaded ad onto the browser. At this point, this methodology is only used by Solbright.
It is doubtful that technology deployment will permit the fully downloaded ad to the browser to be adopted widely at this point. Acceptance of either the 302 Redirect or the ad fully sent by the image server, the two methods used most widely are going to involve pain, cost and compromise. Many sites have been using their methodology for years. Their promises to VCs and other investors, internal metrics, income, billing systems, and general infrastructure are built on our definition, they will say. But, is it too much to ask that metrics be reevaluated every five years or so?
We know that it will take time to change systems. Still, a simple solution is available. Agree on an impression definition and agree to accept "the other guy's numbers" if they are closer to the definition. As we have said before, we believe that the ad fully sent from the image server is the best solution with today's technology. Will that result in some sites leaving "money on the table?" Most certainly. But, if discrepancies go down, they will get paid faster and get renewals more promptly. That has got to be worth a lot.
The industry needs to band together on this one. We need solutions and we need them now. Demand standards from your vendors. Lobby with your trade group. Whether you are in a formal group like the AAAA's or the IAB. Whether you are a part of FAST or part of the Aspen Group. But let's get this done.
- David L. Smith is President of Mediasmith, Inc., a San Francisco and New York based Integrated Interactive Media Agency and Consultant.