• by David L. Smith , May 27, 2008

At a recent conference, there was renewed conversation about the clickthrough. The MRC is very far along on standardization of this metric, yet agencies feel that it is dead from a relevance standpoint. They feel the way this hallowed metric is being used harms the campaigns it is purporting to measure. How can this be true? Read on.

Many agencies are seeing that there is a negative correlation between clickthrough and conversion. This smacks of an "everything we know is wrong" moment. Clickthrough is supposed to be an indicator of success, not the opposite. A logic check says that it is not so. The more enticing a banner ad or piece of keyword/search text is, the greater the number of people who click on it. But a lot of these people might be clicking because of the enticement, not because they are actually interested in the product or service being sold. Voila! The result is bailout and low conversion. If it is boring, fewer people click on the ad. But those that are searching for the solution may click no matter how interesting the copy or graphic is, given that this is the answer they have been seeking. This produces higher conversion than the more interesting ad.

We've written before about the need to measure viewthrough in addition to clickthrough in measuring effectiveness of a banner ad. Viewthroughs generally have higher conversion rates, as consumers have been branded in some form -- they come to the site from the URL line, not through a random (and often then bailed-out) click.

But we have a problem. Almost all performance networks optimize on clickthrough, as they cannot see the viewthrough or the back-end conversion. Google gives clickthrough a high priority in its search ranking algorithm, as do most other search engines. They cannot see the end performance, either.

There is a solution that should not be that hard to implement. Let me give some examples. Many banner ads served on networks are served by third-party ad servers (3PAS) on behalf of an agency or client. The third-party ad server reporting for the agency includes back-end performance for clickthroughs and viewthroughs. It also includes conversion and, assuming that the agency inputs cost data, a CPW metric (cost-per-whatever an advertiser is measuring). Why can't the third-party ad server companies (the two biggest of which are owned by Google and Microsoft) create a capability for a contoured reporting function that a network, site or search engine could log on to that only showed their own performance in conversion and CPW? The resulting information could then be used for optimization.

I recognize that there are issues with this. At any point in the campaign, the actions from a clickthrough are apparent yet the future actions of a viewthrough are not. And sites and networks might not optimize on all of the same factors as an agency or advertiser might. But still, this would represent a far better outcome than optimizing on clickthroughs. Also, there is bound to be normal paranoia on the part of advertisers and agencies relative to letting the "fox into the henhouse." But technology could take care of that.

I'm certain that you can find some other things wrong with this proposal. But I am frankly looking forward to the day when an exec from a performance network walks into my office and tells me that they are prepared to optimize on the same things that we do. That's the network which will truly differentiate itself from the other 300 they are competing with.

Death to the clickthrough! Long live conversion and performance!