You can only sell a finite number of a particular product in any given period. Once you get past a certain point, you may be able to increase your sales by a small amount but only if you increase your CPA by a very large amount.

Leaving aside all the great things you can do with AdWords, such as improving your ad text and keywords etc., if you have an ad group that does not change, then you will eventually find a CPA which gives you the most profit even if it doesn't give you the best ROI.

This sweet spot has been made a lot easier to find with the invention of AdWords Experiments.

Here is a sample screen shot :-

To keep it simple, lets look at the results above just in relation to 'Google Search'.

What I find really interesting about the results (and I have seen this more than a few times now) is that even though the average position for the experiment is higher (as in lower down the page), the CTR is actually higher!  This will most likely be to do with broad match getting more 'broad' with the higher CPA.

You can also see that the higher CPA ad group has a lower conversion rate and definitely a lower profit without even having to ask what the margin is on this product..


  • The campaigns above were all using CPA bidding, and all keyword match types.
  • For the purposes of this post I have totally ignored the fact that there will be some extra sales generated indirectly form your AdWords campaign that will never be attributed to AdWords by AdWords conversion tracking (you may want to take this into account).
  • AdWords experiments are not usually this clear cut.
  • For some reason, when experimenting with different CPAs for ad groups using conversion optimizer, Google don't give you those nifty little 'statistically significant' indicators that you get when experimenting with keyword bids etc. But if this was my business, I would be definitely be willing to lower my CPA to less than 84 for sure!

See here for more information on AdWords experiments.