Ariel Arrieta [D-MTKG]

When we buy (sell) a CPA campaign, it is necessary to agree on which will be the conversion and how it is going to be measured.

The alternatives vary according to the type of campaign that you want to run:

1- Impulsive campaigns (paying post click)

If it is one of purchasing by impulse or with a low resilience such as given permission to receive an email, where the client (user) doesn’t require much analysis to go through the conversion and at the moment he clicks he goes through it, for example filling up a form. The media charges only if the conversion is made in the same session, if the client returns the following day and makes the conversion then, the media that contributed to the transaction will not charge because the deal was not closed before.

2- Campaigns with complex products (paying post view):

The transaction may take longer to go through than on the same session, for example any e-business transaction that implies an analysis on behalf of the client. On this case the media charges regardless how old the impression was and the post view time is negotiated according to the product (usually 30 days).

Even though in theory is done, the usual problem that arises are the discrepancies between the media figures (of a site or a network) and the ones from the client, since the client has to believe the figures of the adserver where it says that the sale was generated by you, and the problems appear when the advertiser has 100 real sales, the Networks are claiming fees for 150 (one 60, the other 50 and 40 the next) and it is almost impossible to determine the overlap.

In this case what ends up happening is some of the following:

* An advertiser with a Wide knowledge of the business dynamic assumes that he will probably have to pay fees to more than one media for the generation of each sale, since all of them contribute to it, but it isn’t impossible to know which the real contribution of each media/network was.

o Practical Solution: take on a higher fee cost and reduce the value that you are willing to share (a portion that covers the total potential risk)

* A more suspicious client, with less results history, believes that he is being cheated on and this discrepancy problem, wears out not only the campaign but also the relation with the advertiser.

o Practical solution for the advertiser: to make the one with less history or less volume like the “fake”  and to pay the commission to the one with a higher commitment.

There is also another model that in theory is more efficient, which to run a campaign by LTV or Life Time Value, where the goal is to compensate for the client training proportional to the vital value that the user will have, since there are many times that the more expensive media can generate more valuable clients, but if you take only the average to evaluate the campaigns, you may exclude them for not being so competitive, becoming this a fallacy.

The only drawback of this model is that it hasn’t be technically proved, but it is a matter of reliability more than technological, because the media is the one who has to rely on the client making its “cooking” transparent to generate an strategic, long term and efficient partnership.

Finally, the advice depends on each campaign and the type of relation that the Publisher-Advertiser has and the degree of understanding the business. The most simple step where to start a serious relation is to make a CPA campaign with a Post click model and that the advertiser knows that it is receiving more untracked conversions, y this variation can be adjusted increasing the CPA value on the same proportion to traction more inventory.