There has been a lot of talk within the industry recently around the viability, truth, life and actual meaning of the cost per acquisition – or CPA - metric.

While for us, the metric sounds great, it means displaying impressions until the user converts for Ad Networks, but it also means blanket coverage (within reason) from the point of view of “we know converters are in our network so one of these impressions must allow them to convert.”

Speak to many within the industry and they'll tell you how well their CPA deal is working - always hitting targets, always with the same network. The CPA metric is still good for some campaigns, but the reliance on it has to change. We’ve been educated into looking to the bottom right of our Excel sheets and in turn, this is what many clients tend to monitor most closely.

A lot of display advertisers will want to move towards this model, letting the Ad Network throw away impressions in favour of conversions, firing cookies across the web and retargeting until your cookie pool is more akin to a dried up riverbed.

The CPA is built on a fallacy. It is built on the rule that the last click or impression is the one that wins (i.e. the last click before that conversion gains the authority to claim that conversion) This means that the Ad Networks then need to charge us as an agency, which is a cost that is then passed on to the client.

Run a standard exposure-to-conversion report in DFA and you'll see the last click probably owes a lot to a PPC impression/cllick, a display impression/click or even a term searched for through Google. It's certainly something we’ve seen within clients here at MEC, with the instigation of natural search tracking post-view display conversions disappear.

The reason for this is that people were seeing an impression and searching through Google, which is then attributed within DFA as a natural search conversion. However, it should be noted that the search would not have happened without the display impression. Of course, the impression itself will not get the credit, as the client will focus only on that figure that sits at the bottom right of the Excel sheet.

We can learn much more about all of this in the way we approach the measurement of our media to find the true value of display, PPC and SEO, and how we begin moving clients towards ‘attribution models’. This, of course, is a term that has been thrown around a great deal of late, with many out there claiming to be attribution platforms - when in fact they offer little more than a few more people to help manipulate spreadsheets from DFA. But this is something that we need to go beyond, taking into account raw log files from ad servers, analysing the data and applying the relevant insight.

All of this involves the careful consideration of the user’s online journey, which sites are feeding conversion rates, the propensity of those sites/users in conversion. We should also look at latency models which are able to give us an indication of the length and number of exposures before conversion occurs. Currently our CPA metric takes none of this into account - yet we place reliance in it and use it to bash the BARB or RAJAR panel over the head with accountability. Our main benefit is the "potential" to offer more accountability than any offline channels - we are only part way towards this goal presently.

There are a couple of platforms that allow us to do this from the start and are ready to use. But first, we need more people asking the question - What does my CPA tell me and what else is there to know?

Infectious Media recently addressed this topic as did a number of industry experts in a piece for Exchange Wire recently. One constant between these two articles seems to be that the CPA metric is far from viable if we are to progress to a more holistic view.