The Issues with Automation in PPC

 

In recent years, Google has been making an ever-greater push towards simplifying the management of campaigns within their Ads platform.  The logic here is simple… if campaigns can perform as efficiently but management can be less time consuming, more ad revenue should come Google’s way.  There are a number of ways they are trying to make campaign management easier.  Most are based around automation utilizing Google’s machine-learning algorithms.  In this article, we’re going to take a look at perhaps the most advanced automated feature within Google Ads, auto-bidding strategies.

Historically, advertisers would need to set the maximum bid they were willing to pay per click for keywords, products targeted in shopping ads, display ad impressions, etc.  Bids would then be adjusted on an ongoing basis dependent upon their performance relative to goals.  With the introduction of automated bidding, advertisers were given the ability to set KPI targets, enabling Google’s machine-learning to adjust bids in real time based on the multitude of data signals at its disposal, optimizing bidding and hopefully achieving those targets.

In the early days of bidding automation, advertisers who were hands-on in the account and actively adjusting bids often found much better results through the manual method than with a fully automated bidding strategy.  However, in the decade that has since passed, Google has continued to improve upon its AI’s capabilities.  Advertisers utilizing automated bidding have now come to see improved performance over even the most diligently optimized manually bid campaigns.  That said, there are still a number of situations when relying on automated bidding strategies falls flat and can even drive poorer performance overall.

 

Brand Keywords

Bidding on variations of your brand name is key to the overall success of any advertiser’s PPC program for a number of reasons, but the allocated percentage of your total budget should be kept to a minimum.  Given the high relevancy, advertisers typically see the highest quality scores on their brand keywords, which results in very low average clicks costs while maintaining a very high impression share.  The impression share is essentially the number of times your ads were shown as a percentage of the potential impressions you could have received.  Once advertisers start to see impression share scores in the 90-95% range or above for a keyword, increasing its bid typically results only in increased average cost per click, without any increase in volume.

What we have seen (with very few exceptions) is that with an automated bidding strategy running on brand keywords, Google’s AI will continue to increase bids in effort to drive impression share near 100%.  The reason being is that your brand keywords will almost always drive the highest performance when looking through the lens of last-click attribution, but it’s essentially failing to realize overall account goals and the lack of additional available traffic for your brand name. Through a manual approach of incrementally adjusting bids over time, advertisers are easily able to maintain a high impression share and sales on their brand keywords while minimizing the associated ad spend, thus maximizing return.

 

Lower Conversion Total Campaigns

Automated bidding can be very effective, and we’ve seen the best results for those campaigns with a conversion or revenue goal.  With proper management, advertisers will usually see marginal or, in some cases, significant improvements in overall cost per conversion or return on ad spend for a campaign.  If Google’s AI isn’t able to acquire enough data, however, performance can fluctuate wildly.  Though not explicitly stated, Google’s AI appears to weight recent performance significantly.  This puts advertisers with lower conversion totals in a difficult position, as the algorithm tends to get too aggressive with a slight improvement in recent performance.  On the opposite hand, it can become too conservative with even a slight drop in performance, resulting in a sharp decline in spend.

It's possible for lower conversion total accounts to see improved return over the course of a long period of time, but many cannot afford the swings in performance from month to month that it will take to get there.  In cases such as this, manual bidding is often still the best approach.  Having a deeper understanding of the program allows an individual to review performance in both the short term and long term, to evaluate to what degree bids should be adjusted.

 

Setting, Then Forgetting

As the name might imply, automated bidding can tempt account managers to check in less often. However, by switching all campaigns to automated bidding strategies and leaving Google’s AI to effectively take control, advertisers will almost certainly see an overall performance decline when coming from a well optimized, manual bidding approach.  Those managing automated campaigns effectively understand that with the relinquishing of control over bidding, comes an increase in the need to monitor and adjust other settings within a campaign.

Most campaigns set to a specific budget and target goal will either fall short of the allotted budget to hit the target KPI goal or be limited by the daily budget limit at that goal.  It is, in a way, a balancing act, and one that needs monitoring and adjusting on an ongoing basis to make sure that spend and return are optimized.  In addition, advertisers often see spikes in performance around sales or promotions, which Google’s algorithms are slow to respond to on the front end and slow to revert when conversion rates come down to normal on the back end.  By having a level of manual oversight, you can review past performance and input anticipated conversion rate changes, as well as the dates that change is expected.  Google will then factor in these settings, allowing for better performance both during and after the promotion.

Auto-bidding continues to be the most highly adopted automated feature in Google Ads, but others continue to roll out at an increasing rate.  Two examples that are of potentially significant impact are the various ad and keyword auto-apply options.  In brief, Google will take the liberty of building out new ad copy and keywords, based on what has worked well in the account as well as what users are searching for on Google.  Advertisers aren’t forced to incorporate recommended elements into their campaigns, but advertisers should make sure their account settings are appropriate or they can auto-apply if not dismissed manually.  For obvious reasons, we would never recommend any advertiser rely on Google or any platform to control ad copy messaging or targeting methods such as keywords.

Machine-learning capabilities in Google Ads and other platforms will continue to expand.  That much is certain.  Advertisers who remain active in their campaigns and take advantage of the appropriate automated features for them will come out on top.  Relying on the platforms entirely without manual oversight, however, will undoubtedly lead to a decline in performance.