Ever since Google rolled out the Optimization Score, all accounts are full of notifications and recommended changes that may improve your account score. These recommendations range from keyword suggestions, targeting changes, ad suggestions, and automated smart bidding strategies. Should you apply or dismiss these recommended changes?

In an account, you might see a score like 77.5% and several ways you can improve this score. If we applied all the recommended changes, we could have a 100% score. Although it is a good idea to review each and every suggestion and determine if it makes sense for your campaign goals and account strategy. One of the top recommendations in most accounts is around Smart Bidding Strategies.

Smart Bidding Strategies

How does Smart Bidding work? Google’s system evaluates your Search campaigns and uses machine learning to make recommendations for your account. The system will evaluate settings, historical performance, hour of day, location, and real-time auction information. These recommendations are designed layer on advanced machine learning to your campaigns to help you achieve an outcome. This means you are turning keyword level bidding and letting Google manage it on your behalf.

It is possible that these bid strategies and Google’s algorithms can use account signals to make changes that can have a positive impact on your campaign’s performance. However, we have also had situations where the automated bid strategy was not a good match for our campaign and had a negative impact. Just read the article about When Target CPA Isn’t Hitting Your Target for one possible outcome.

These smart bid strategies can be used to help increase your visibility, clicks, or conversions. They can also be used to maintain a specific cost-per-conversions (CPA) or improve your return on ad spend (ROAS). Target CPA bid strategies may work great for lead generation but may have a negative impact on ROAS. There may be cases where a CPA may be higher for a particular term, but the revenue generated is significantly above the ROAS goal. One of my colleagues wrote an excellent article about the difference between CPA and ROAS that still holds true today.

Notifications

Inside every account, you will see Google’s notifications that can range from highlighting disapproved ads, billing issues, keyword conflicts, as well as other smart bidding strategies. In one campaign it suggests switching several campaigns to Maximize Conversions or Target CPA bid strategies. You will have to look deeper to determine if these strategies make sense for these particular campaigns.

If you click the View button, it will take you to another screen where it gives additional information about the Target CPA strategy and how it will raise your optimization score by +9.9%. According to Google, this score is determined by real-time statistics, existing settings, and performance history. It would be tempting to click Apply All and make these changes, but is this a good idea?

In another account, the system is recommending we switch to Target ROAS strategy. However, the interesting thing to note about this account is that they are lead generation and so revenue is not being recorded. So, this is a bid strategy that would be worth dismissing. This is exactly the reason you need to carefully consider each recommendation before applying them to your campaigns.

Recommendations

Should you make these recommended changes to your account or not? One thing to consider is the machine will never understand user intentions. Google doesn’t know the difference between someone looking for a Charger Car or a car charger for your phone. Just consider how many irrelevant terms show up in search query reports. This is where we come in and are able to analyze the searcher intention and exclude irrelevant terms. Google also does not understand the purpose of your campaign it just looks at data and suggests recommendations.

These bid strategies can be applied to lower volume campaigns, but the machine learning algorithms are going to be more successful with more data. One Google rep said that bid strategies do not work as well with campaigns with wild conversions fluctuations. According to Google support information, they suggest in order to maximize results using Target CPA you need 30 conversions in the past 30 days. Target ROAS bid strategies require 50 conversions in 30 days.

Another thing to think about is your attribution settings in your account. Many accounts are still set to Last Click attribution which gives 100% credit to the last ad click. It may be a good idea to switch to a Multi-touch point attribution model instead. This will give partial credit to other campaigns that are part of the journey. One option might be Position-Based Attribution that gives partial credit to the first and last touch-point and splits the remaining credit to anything in between. You will also want to make sure all your conversion actions are important before letting an automatic bid strategy use that data to make decisions in your account.

Target CPA and Branded Campaigns

In one account, one of the campaigns bucketed into the Target CPA recommendation is the Branded campaign. In this account the CTR is high, CPCs are low, and the revenue is high. The purpose of the Branded campaign is to protect these keywords from competitors, control the ad message, and dominate the Search Engine Results Page. Therefore, using a bid strategy could limit the number of impressions for our branded terms. Do we really want to limit the number of times our branded terms can appear in auctions? In these campaigns, I prefer to use enhanced manual bidding (eCPC).

What if we decided to switch this campaign over to the Target CPA strategy? The system recommends we set the Target CPA bids at $16 based on the account average. One of the branded keywords has a higher CPA at $35, but the ROAS is 1476%. It is possible the system could limit this keyword due to the higher CPA even though it has high revenue. Other options would be to go with Target ROAS bid strategy, but again this would focus around revenue and could limit how many times we appear in the auction.

Another campaign that was recommended for this strategy is our Non-Brand mobile campaign. This campaign had 15 conversions in the last 30 days. According to Google, they recommend you have 30 conversions in the past 30 days in order to maximize the results. Another thing is Google is only able to use the information obtained from Google ads.

 

Assisting Revenue Campaigns

In Analytics, we can see that this campaign is assisting revenue to other channels or campaigns. The products sold by this client are expensive and require some planning. Plus, order these items is complex and it would be more difficult to do it through your mobile device. The low bounce rate and high pages-per-session suggest that these searchers are relevant. Another thing we know is that these Non-Brand campaigns serve more as an assisting role to other campaigns or channels. So, if we tested a new bid strategy, we would want to monitor any negative impacts on other channels.

Run an Experiment

Completely turning the reigns over to Google to bid on these keywords based on the CPA would be risky and could impact the assisted revenue. One way to test these goals is to run a 50/50 experiment to see if the system can do a better job. In an experiment, you would copy the original campaign and rung a test with 50% of your total budget. Once you activate the experiment, it is important to avoid making changes in these campaigns while the experiments are active.

In one account, we are running a Target CPA experiment. The experiment CPA is $84.20 compared to our original campaign at $82.56. We can see the average CPC is higher in the experiment, but the conversion rate is also higher. Running an experiment is one way to test out a bid strategy without risking your entire budget. Plus, if the strategy does not work you can just stop the experiment without having to reverse all the bids.

In these experiments, if you do not see the blue asterisk, this means that the test has not reached statistical significance. It needs additional time to run. Keep in mind, it is a good idea to set the experiment end date further into the future. You can always end an experiment early, but if the test concludes early you will have to start over again.

Closing Thoughts

Run some experiments with these automatic bid strategies to determine if they will work well for your campaigns. Also, if you decide to go with an automated bid strategy, it is a good idea to continue to monitor it regularly. In the past, I have had automatic bid strategies that performed well initially and then suddenly had wild performance changes. Also, for another perspective, read When Target CPA Isn’t Hitting Your Target by Laura Lowery.