I'm sorry I'm a little behind on my posts, but I do have to work sometime. Anyway, I got quite a number of emails from friends and co-workers on what I thought of Google testing a cost per action deal structure. Funny thing is that I caught wind of it a while back and wrote a post called Google Testing CPA Network - Yahoo? Anyway, here are my thoughts on CPA and while I'm not as cynical as that original post, I'm not that far off from where I was back in June of last year.
- Read the fine print, this is for Google's content targeting network, not for Google search results or search results on partner sites. If you are like most advertisers, Google's content network is either turned off or a small percentage of your spend.
- A CPA deal will go a long way to solving a click fraud problem, but I didn't read anything in their details about visibility as to which publishers will be running the CPA ads. That's always been my biggest issue with content and until they tell you where the ads are running, that lack of visibility is still a problem.
- You'll need to implement Google's conversion tracking codes which of course work really well, maybe too well. As outlined in their own help on conversion codes, Google places a 30 day cookie on the user's PC which will be used to connect a conversion with a click. This 30 day window opens up a large latency and double counting issue with your other ad buys (yes, I just linked to a post from 2005).
- The double counting becomes an issue because Google won't have any visibility (unless you let them) into your other ad buys and could potentially count a conversion that you've allocated elsewhere.
- I usually solve that problem by insisting on using a 3rd party for tracking, but didn't find any options for that path other than turning all tracking over to Google.
- Finally, I found no mention of impression based conversion tracking versus click based. Again, Google's PPC works well because it starts with a click, but how will they count a conversion that's generated from a view through impression and should you be paying Google for that? What happens if you are also running ads via a different network or are already on that site due to your own wheeling and dealing?
I tried looking around for clarification on the above, but couldn't locate it based on my usual blog readings. So in the end, would I test CPA? Yes, but only for clients that have shown successful content campaigns already. Do I think this is some change the game or advertising altering panacea that delivers great value for advertisers while rescuing Google from click fraud issues? No.
I've run a TON of CPA deals over the years totaling in the multi-million dollars. What I've found is when publishers are desperate for ad dollars, an advertiser can make a killing, but otherwise a CPA deal is just one of many options available to you. Sometimes it is better for you, sometimes it is better for the publisher, and sometimes nobody wins and you get outbid for the inventory; heck, there is always the possibility you don't convert enough and the publisher drops you altogether. However, if you aren't already translating your CPM or CPC deals into a CPA than you really don't understand performance marketing and a straight CPA deal won't be able to cover up your marketing short comings.
PardonMyFrench,
Eric
*****By the way, my friend Kevin Lee from Did-it wrote a post for ClickZ which mirrors my post almost to the letter. Good to know great minds think a like.
Comments
You can follow this conversation by subscribing to the comment feed for this post.