Mic 300x257

You know how online advertising works. A web user sees your perfectly targeted ad… the creative sings… the viewer’s pupils dilate… their pulse quickens… they feel a small tingle in the pit of their belly.  They click with the intensity of someone making their last click on earth.

The landing page renders… pipe organs resonate… sparks fly from the keyboard… cherubs circle.  The user clicks “buy now” and enters all required information with the focus and purpose of someone filling out their marriage license.

Conversion complete, end of story.  Add up all the money you spent distributing your ads, divide by the number of times this series of events takes place, and you’ve got your CPA, right?

That’s obviously pure fantasy… or maybe romance novel, but it’s the way media buys are often measured.  The good news is that savvy media buyers are starting to catch on, and that the old CPA model is starting to evolve.  The problem is that the term CPA is now being used for all types of other attribution-based pricing models, and this bait and switch is posing a risk to forward evolution.

Beware the Playahs

The primary pit fall is that providers (ad networks, targeting platforms, etc) are using the term CPA, but are actually attributing credit to themselves in arbitrary ways for activity that’s not a conversion in the way the customer probably intended.  In reality, there is a ton of benefit delivered from a campaign aside from the last click, we all know that.  But calling these types of pricing models CPA and not explaining the actual calculations up front is causing confusion.

Advertisers hear that they can run their campaign on a CPA that works for their ROI objectives, but they end up getting taken by surprise when they realize that the provider attributes conversion credit for things they didn’t anticipate.

For example, a view through conversion (when someone sees an ad, doesn’t click, but converts within a specified time period) is an awesome metric for determining the effectiveness of a campaign.  However, when they are credited the same way as click conversions, the advertiser is double counting conversions, and the provider often games the system by delivering cheap, low impact impressions just to get as many views as possible in the hopes that enough of those people will convert later.  That’s clearly not the intent of bestowing credit to view through conversions, which should otherwise be a very smart move by the advertiser.

Tips for avoiding the CPA conundrum:

    • Demand transparency, so you know where your ads ran, and where they were clicked.  This will help to dissuade providers from buying up cheap low impact impressions in the hope of racking up view through credit.  If the ratio of clicks to view through conversions is too low, that’s a sign that something is up.
    • Don’t select a CPA pricing model, and instead align incentives with you provider.  Start with a month or two on a dynamic CPM model where pricing is totally transparent, and then move to a CPA or revenue share-oriented model if it makes sense.
    • Develop your own attribution model.  We’re seeing our most-savvy advertisers develop a model where they divide up the value of a conversion and attribute a portion to the first impression delivered to that user, a portion to the assist (view through,) and a portion to the last click that leads to a conversion.  The right way to divide this credit depends on your own internal metrics and the lift you see from view throughs on your overall site-wide conversion rate.

At the end of the day, moving to a CPA where attributed revenue is incorporated into the calculation is a great thing, and will lead to more intelligent and efficient marketing spend.  That said, if advertisers are burned by bait and switch pricing models, they’ll retreat from fair-attribution, and we’ll end up back with the last click as the end all be all performance metric.  That would be an unfortunate loss for media buyers who would otherwise leverage these insights for more effective campaigns, not to mention its stifling effect on innovation.  Who wants to develop new high impact marketing channels that make potential customers feel like characters in a romance novel if they don’t get fair credit for the foreplay when advertisers measure success?