I have talked in the past about the importance of continuous testing in your paid search campaigns. (Missed it? Read the post here). Testing is one of the best ways to stay on top of the ever-changing industry we work in. In an effort to heed my own advice, I conducted some automated bidding experiments to find out if Google’s automated bidding could really perform up to snuff. I tested these across a few accounts and thus different industries, and the results have changed the way I think about automated bidding.

Target CPA vs. Target ROAS

Before you begin, you must first determine the bidding type that fits your business. The type of automated strategy you use should depend mostly on your KPI.

Target CPA is an AdWords Smart Bidding strategy that sets bids to help get as many conversions as possible at the target cost-per-acquisition (CPA) you set. It uses advanced machine learning to automatically optimize bids and offers auction-time bidding capabilities that tailor bids for each and every auction.

When to use: If your goal is to get conversions, especially those with ambiguous value, such as leads, or form fill outs, then target CPA is the perfect test to find if you can get cheaper conversions than your manual bidding strategy.

Target ROAS is an AdWords Smart Bidding strategy that lets you bid based on a target return on ad spend (ROAS). This helps you get more conversion value or revenue at the target return-on-ad-spend (ROAS) you set. Your bids are automatically optimized at auction-time, allowing you to tailor bids for each auction.

When to use: If you’re an e-commerce business and aiming for a specific return on investment, then target ROAS would likely be the best test to run.

Target CPA Test

When looking at the aggregated data, the important thing to consider is that the overall compilation of data does not speak to how implementing these automated strategies will perform. There are multiple variables that will impact the overarching performance. Another thing to consider, is that for each one of these campaigns a different target CPA was used. Generally, the recommendation provided by Google was utilized. I can tell the suspense is mounting, so without further ado, let’s look at the hard data.

Original Campaign Data


Experimental Campaign Data


Yeah, I get it, it’s a lot to look at. To simplify everything, I broke this data down to the percent change from the original to the experiment.


After mulling the results, we can really understand how Target CPA might work to get a desired CPA. The experimental campaign on average moved ads down by .3 positions. In doing so it was able to cut the CPC by over 40%, while maintaining a very similar conversion rate. This allowed campaigns to lower the average CPA.

Target ROAS Test

Before testing it is important to have accurate e-commerce data imported into AdWords. Without this data, AdWords is unable to understand how to optimize in order to generate a return.

Original Campaign Data


Experimental Campaign Data


The results of this experiment make a great case for trying out Target ROAS automated bidding. Take a look at the overall changes to performance during this experimental run.


How neat is that? We found that using this strategy not only made our spending more efficient, but it increased the conversion rate by 85%, while gaining 38% more conversions!


The Verdict

Though this data aggregation came from a smaller set of data, it tells a solid story about why testing is important and how effective automated bidding can be. With individual campaigns I have seen some perform with immense success, while others have performed at the same or lower than the original. As the above data shows, there is definitely a case for testing new bidding strategies in your AdWords campaigns. 

Need a hand with PPC? We’ve been at it since the beginning. Check out our PPC capabilities.