Clickfraud and CPA v CPM/CPC

By September 27, 2006 One Comment

Business Week Cover click fraud

Click fraud just doesn’t seem to want to go away.  Despite Google’s $90m settlement and Yahoo!’s uncapped but similar settlement both of earlier this year the issue continues to rumble on.  The Oct 2 issue of BusinessWeek leads with it and there is a lot of commentary in the blogosphere – e.g. Esther Dyson here and.  Eric Frenchman here critiquing the BusinessWeek article.

As Esther says this is an issue of accountability.  One of the big benefits of CPC (cost per click) when it started replacing CPM (cost per thousand eyeballs) was that the click was supposed to evidence genuine interest.  Unfortunately it also opened up the possibility of fraud.  The value per eyeball is so low that fraud is a waste of time, but as the price of popular keywords like pension and mortgage regularly head north of £10 there is enough cash to fund a bit of organised crime, as the BusinessWeek article describes.

I have long thought that moving to CPA (cost per acquisition) pricing was the answer to this.  That way the network only gets paid if a customer has parted with her cash and has been truly acquired.  No way to fake that, so no fraud. 

Over the last couple of years CPA has become more prevalent but I have been getting a sense recently that the rise in CPA has flattened out.  That was confirmed for me in two conversations at Adtech in London this morning (great show with a great buzz – the sector is definitely feeling good).

The thing is that as far as I can tell just about all advertisers (of any scale at least) are indifferent as to whether they buy CPM, CPC or CPA – they all track the downstream metrics so it doesn’t matter.  For example a CPM rate of £5 per thousand where one visitor per thousand clicks through and makes a purchase is equivalent to a CPA of £5.  Because the terms can be calculated from each other they are interchangeable and the industry has standardised on CPM to make it’s life simple.

Advertisers can make use of blind networks from Google, Yahoo! and others without worrying about click fraud provided they know what a customer is worth to them and monitor conversion rates from clicks to purchases.  So long as their CPA (cost per acquisition) is less than they think a customer is worth then they shouldn’t worry too much about how much click fraud there is.  Advertisers who adopt this model will drift towards networks with less click fraud anyway, as they will, all things being equal, be more efficient and give a lower CPM/CPC/CPA.

The whole beauty of web advertising is that is measurable and trackable.  So long as advertisers take advantage of that they should be OK and click fraud should in the long run disappear.  The measurability and trackability is also what makes the distinction between CPM/CPC and CPA moot, at least for the larger campaigns.  At the smaller end where the volume isn’t enough that the law of averages kicks in then maybe we will continue to see a trend to CPA.