Deeper Understanding of CPA Bidding: Best Practices, a Case Study and Breaking It Off

As part of our ongoing series on automation and time-saving, we felt that talking about AdWords’ Cost-Per-Acquisition bidding was a must.  Aside from the Display Campaign Optimizer, which is a whole post in and of itself that will happen at some point in the future (though sadly not in this series),  CPA bidding is the most automated you can get within all of the different AdWords bidding methods.  Bethany Bey covered the basics of CPA bidding just over a year ago, and amazingly (considering the speed with which Google makes changes) almost all of her info still applies.  If you’re a little bit fuzzy on the basics of CPA bidding, check that post out.

In the year since that post was written, we’ve learned more about this option and want to present even more tips about how to best use it.  I’m going to run down some further best practices, walk through a case study of when CPA bidding was right for me, and finish up with what to do when you fall out of love with CPA bidding.  Prepare for an emotional roller coaster with Google’s CPA bidding.

First off, the difference between max CPA and target CPA bidding.  There’s some debate about which is right for you, but I would always advise to use target CPA bidding.  If you use that option, it helps pin your campaigns to a more specific target, but it also includes a max CPA on the backend that we just can’t see.  You get to tell Google what you’re shooting for, and as a part of that target they’ll automatically generate a max CPA for you.  While we’ll never be able to see that figure, it’s inherent to the thought process of the target that you establish.  In this case, I think that it is best to cede some control to Google.

Second, recognize that average position may now become a secondary metric for you to track.  Amanda talked yesterday in this series about optimizing toward your best average position, but with CPA bidding your bids are going to be adjusted to their sweet spot automatically.  Once the Google calculations have enough data to know where your CPC needs to be to meet your target CPA, they’re going to adjust your bids accordingly.  You won’t want to track average position as you may have in the past.  You’ll need to watch your CPA and the overall volume that you’re driving.  We’ll get to this later, but CPA bids can really drive down the amount of traffic in your campaign or ad group.

Third, unlike the DCO, ad rotations matter when you’re on CPA bidding.  My campaigns are always set to rotate (for 90 days now! Yippee!), so the ads rotate evenly (clearly).  This can affect your CPA, as different ads have different conversion rates.  Be aware that whichever option you choose can change performance.  I wish that I had experience with optimizing for clicks or conversions, but I’m an ad testing man, through and through.  If you’ve ever used a different option, please let us know and what you’d recommend.

Next, if you’ve ever talked to a Google rep about an automated campaign, they’ll usually tell you that it needs more time and you should let it run a little bit longer.  While this can be true to a point, it’s important to act when things aren’t going well in your automated campaigns.  Just like with bid changes, you want to give CPA bidding some time to run before deciding whether or not your changes have had the desired effect.  Be aware that sometimes waiting is not the answer.

I’ve had CPA campaigns show a big dip in impressions, been told to wait it out (as I had modified the budget recently {yes, only the budget}), and then seem the same dip continue onward.  Not only did I lose out on two weeks of performance as I received assurances from my Google rep that everything would be back to normal soon, I looked like a lazy idiot because I hadn’t done anything to fix the loss of impressions.  My client was mad and I lost conversions because I gave it more time.  If something feels wrong with your CPA bidding campaign, it probably is.  Act; don’t wait.  Give more time to bigger changes (such as a new target CPA), but smaller changes (like daily budgets) should have a much tighter leash.  If it’s been a week and the only thing you’ve done is modify how much you’re spending in a day – do something.  Don’t let automation give you an excuse to sit on your hands.

Once CPA bidding has been running for a while in your campaign(s), it still requires routine maintenance.  Watch your keywords/ad groups and pause any that aren’t converting.  They’ll eventually be bid down by the CPA algorithm, but what’s the point in waiting for that to happen?  If you see something actionable, act.  I’ve also run CPA bidding on the Display Network with great results, and the top maintenance item there is routine site exclusions.  New sites are popping up all the time, and if you’re not careful a new potential placement could waste a lot of money across all of your ad groups.  Even though the algorithm is testing wisely on an ad group basis, when you see that it’s not converting in any of your ad groups you should exclude that from the entire campaign.

The last best practice I want to touch on is that not all conversions are made equal.  The biggest theoretical limitation to CPA bidding (in my humble opinion) is that you can’t bid on many-per-conversions or conversion values.  The algorithm sees a conversion as a conversion, which simply isn’t the case a lot of the time.  If your conversion values can vary wildly, consider splitting out highly profitable keywords out by themselves and bid accordingly.  I have an account where some keywords have a higher propensity to lead to a sale, so we’ve made them an entire campaign with entirely different target CPAs.  This could be done on an ad group level as well.  For some people it might not be that simple, though, so CPA bidding just might not be an option for you.

Is all of this worth it?  I recognize the possible pratfalls of CPA bidding and I know that some people are detractors, but it can really work wonders.  Let me show you performance over a few months in one of my accounts:

Month Cost Conv. Cost / conv. Conv. rate





















If you’d be so kind, could you please take a guess as to what happened in December?  That’s right, CPA bidding was enabled.  Despite a temporary decrease in spend, conversion rate skyrocketed (that’s probably not a true “skyrocketing” but that kind of gain still warms my heart’s cockles) and cost-per-conversion plummeted.  Of course, seasonality plays into this as well and there are always mitigating factors, but these are promising results.  Once Google finds areas that work, they really highlight them.  As long as those areas continue to work, your CPA bidding can just get more and more momentum.

Another quick thought – sometimes areas don’t work in CPA bidding, and they can be bid down to nothing.  If one of your primary ad groups has a bad week and conversions stop coming in, bids can be reduced over and over again down to the point where you’re no longer even showing up on those keywords.  If that happens to you and CPA is still kicking butt elsewhere, consider breaking out the poor performers into their own campaign for manual management.  It does reduce your level of automation, but it also picks your lead volume back up.

So back to those numbers.  They’re great, right?  Can’t you just picture me and CPA-bidding (whose anthropomorphized hands I picture as having big, white, fluffy Mickey Mouse gloves for some reason) skipping off into the sunset as the narrator says “And they lived happily ever after”?  THAT’S NOT WHAT HAPPENED, HOMBRE.  CPA bidding isn’t perfect, and if you would, please allow me to insert another set of data.


Cost Avg. position Conv. Cost / conv.

Conv. rate
























































Late April/early May saw ok performance, nothing like the heydays of December and January, but it was respectable and a pretty ok volume.  But then at the end of June, there were a few days in a row without conversions that just drove bids lower and lower to the point where the biggest campaign in my account was spending $50/day and not getting any conversions.  It was the low volume thing that happens with slow ad groups, only it was my ENTIRE CAMPAIGN that was slow.  CPA bidding had run its course in this account and so it had to be taken out behind the barn and Old-Yellered.  I’ve had a lot of great experiences with CPA bidding, but this was one instance where I had to act.  I didn’t talk to my Google rep, but I’m guessing they would have told me to wait a bit longer.  And then a bit longer.  This was a clear case of me not being able to wait any more, so I pulled the trigger.

If you decide that CPA bidding is no longer for your account, how should you handle it?  Once you change the settings back to manual control, your bids revert back to what they were when you enabled CPA bidding in the first place.  That could be months or even years ago, and think how much has changed in the interim.

What I think is best is to look at recent performance in terms of both average CPC and average position and decide where you want each keyword to be.  You can compare what CPA bidding did to your bids by looking at average CPC next to the old max CPC that you had originally.  That should give you at least an idea of where Google thought those keywords should be.  The final step I like to perform is looking at a similar campaign in your account (if one exists) and seeing how much headroom there is between bids and average CPC just to get a sense of your account’s competitive space.  That way, you can add a little bit of room above the average CPC that you want to pay when you set new bids for your keywords.

It might take a little time, but you can recover from a breakup with CPA bidding just like I did.  Or you could fall madly in love with them and have little weird people/CPA babies and live together forever.  Hopefully this post will give you a bit more clarity on the wonders of CPA bidding.  Please let us know any thoughts you may have below.