Machine Learning in Google Ads: Tips, Tricks and Best Practices

You are a Google Ads Specialist? Well then, congratulations on your new role as… Call it: machine master, automation guru, smart ads manager... I hear you saying, “This is all bulls..t. Google is doing all of these”. You think, your job is mainly taken over by machines, PPC management is getting automated, you don’t really know how Google exactly targets your ads and you loose more control?

Well, you might be right after seeing how your role evolved over time. In the past, we Adwords managers were so busy with bid management, campaign optimization, analysis and reporting; we thought we are VIPs and nobody could ever replace us. And all of a sudden, boom! We are desperate now because we lose control over our day-to-day work. But, how can we learn to live with this new situation?

Well, first of all, the time has come to work smart and let the machines do all the boring work. We can now take our time to focus on what really counts.

Here are my personal tips and best practices on the new era of machine learning in Google Ads. Hope my advice will help you feel better after all :)

Understand User Behavior

Understanding your audience is more important than ever. Only you know what they want and only you can decide how to provide them with the relevant campaigns and ads. In order to be able to make the right decision, invest time in understanding the business, the products, and the service value. Conduct a proper keyword and competitor research, because this will help you to set up the right campaigns, ad formats and create a sustainable optimization plan.

Deep-dive into Data Analysis. Maybe in the past, you didn’t have the time to look into Google Analytics in-depth, but now use your valuable time to learn more about the tool. Because in a data-driven world, data says more than a thousand words: Data is understanding user behavior, but at the same time it helps us understand the machines’ behavior as well.

But when analyzing the data, don’t just focus on the numbers. Try to understand what the data is actually telling you, or maybe isn’t telling you? With the increasing complexity of user behavior, looking only at data always brings the risk of missing something. What, for example, if you are a car brand with users having a long-term buying cycle, online and offline interaction, different device touchpoints etc.? Here you will definitely miss some users if you solely rely on data-driven marketing. Marketing nowadays is more difficult than looking at the data only.

Invest Time In Strategic Planning

First of all, be clear on the goal and set realistic KPIs. Focus on the goal first and then create a strategy around that. If you are not clear about it, then how can you set up your account and campaign structure, or know which smart bidding or smart ad type to use?

Creativity is more important in Google ads than before, because we humans have the emotional intelligence skills, and robots don’t. This is something where we will always beat the machines. Ad Copy writing is not as easy as you might think, it needs a great ability to convince users to click on the short ad snippets. And here, your copywriting skills can come into play. For example, if you create Responsive Search Ads you are the one that gives the input, writes the headlines, decides on the descriptions to be used. AI & ML comes into place after that, so if you don’t give enough input here, the output won’t be as good as you think.

Look At Online Marketing Holistically

Understanding the Google Ads system, keeping up to date with all the changes is crucial. But don’t forget to look at the bigger picture as well. We Google Ads Specialists, as the name says, are specialized in our role, but we have forgotten to look at marketing and user behavior holistically for a very long time. Especially the combination of SEA and SEO; doing both at the same time definitely gives you an advantage in understanding search behavior and in creating the right strategy. You cannot imagine how much insight you can get if analyzing paid and organic data together.

Smart Work in Google Ads
Smart Work in Google Ads

Some Best Practice Examples

1. Asset Testing

If you think you cannot optimize smart ad campaigns, you are wrong. There is indeed a very important thing you can do and which also can have a big impact on the performance: Asset Optimization. Assets happen to be in UAC campaigns, local campaigns, smart display campaigns, responsive campaigns etc. From my own experience, I can say that continuous asset testing and optimization can have an uplift in your performance.

Analyzing the performance grouping is important: Are my assets performing “best” “good” or “low”? How does the conversion data for each asset look like?

You can basically test all assets. But start with the low performing ones. Take the best performing assets as an example and make some small changes in design, layout or wording and replace the low performing ones with the optimized assets. It’s important to give the system time for learning. The asset testing and optimization should start after the learning period which really depends on the data of your campaign. Don’t run your analysis too early!

uac campaign performance grouping
uac campaign performance grouping

Also, always check why the performance is labeled good or bad. Try to use your own performance criteria, because sometimes Google’s performance grouping doesn’t reflect the actual performance.

Pay attention and have a close eye on the performance, don’t trust the machines blindly.

My tip: Use as many assets as possible, ideally all texts, videos and images. Also try to be diverse in content, design, format, etc. We have experienced that text and videos are performing best, while small images are usually the worst performing ones. But you should try to build your own experience and best practice so that you can plan accordingly.

best performing assets in smart campaigns
best performing assets in smart campaigns

Machine learning combined with asset testing leads to very good results. Here is an example of conversions over time with the same budget:

uac download performance with asset testing
uac download performance with asset testing

2. Extend Local Strategy

Already mentioned before, performance depends on too many factors, so try to think out of the PPC box. If your work within the smart ad campaign is limited, think about ways how you can combine it with other optimizations which are closely related to the campaign you are running.

Let me give you an example: For one of our clients, driving in-store visits was crucial. The newly announced ad format, the ‘Google local campaign’, was a perfect fit for this. We created a Google local campaign, where we basically just set the budget, added text, and creative assets - and that was basically it. Apart from the asset testing and optimization, there was not much to do on our side. We had to find a way to extend our local strategy beyond Google local campaigns.

google local campaign format
google local campaign format

With local campaigns enabled we knew that more people will be engaged with Google maps or the local packs in the SERP in order to get company or location information. Users wouldn’t necessarily go to the website, as our main goal was to get them in-store. So optimizing the clients’ Google My Business entry was an important step to improve user experience. What we did with the GMB optimization:

  • We checked and corrected all NAP’s (Name, Address, Phone) and other information like company info, category naming, etc.
  • Added new and relevant business photos
  • Tracked the website URL by adding UTM parameters
  • Optimized the account frequently with new features; for example, posts
  • Answered all the reviews

GMB optimization helped the listings to appear more on the local pack and on Google Maps, which made the engagements going up. GMB optimization also had a positive impact on our Local Campaign performance. The overall local views and engagements increased and we were able to bring very important foot traffic in stores, paid and organically. Both worked towards the same goal: increasing store visits!

google store visit performance
google store visit performance

3. Experimenting

Evaluate Your Smart Bidding Performance With Experiments in Google Ads. Experiments allow you to directly compare performance to your previous bid strategy over the examined period of time.

Experiments can show you clear results sometimes but you may not always see a clear winner. If this is the case, keep on testing or consider to adjust other variables like budget, creatives, other settings, etc. You may also consider whether the chosen smart bidding strategy is the right one for your business. For example, if you are a B2B company with very low online conversions, testing maximize conversions might be the wrong strategy. So always look at the data requirements first before setting up your campaigns bidding strategy. Usually what counts: The more data you have, the faster and more accurate the experiment shows results.

smart bidding experiment
smart bidding experiment

Most Common Pitfalls of Machine Learning in Google Ads

“My traffic dropped since I use smart bidding.”

Bidding to maximize conversions or for a better CPA may cause a drop in your traffic due to higher CPCs. So before you start, don’t forget to keep your initial goal in mind. Be clear on the goals: do you want traffic or do you want conversions at a lower cost?


“Google Ads Management has become so easy, everyone can do it.”

Yes, with all the smart bidding, smart ads, keyword matching options, recommendations tab and so on, Google aims to reach small and medium-sized companies to help them manage their own accounts. But less control means it’s getting more difficult in understanding the system as you have to show more effort in understanding and analyzing the data, creative thinking, etc.


“Smart campaign didn’t bring any uplift in my performance.”

Smart campaigns are not magic and not a solution for all the problems. If your budget or data is too small, or, for example, your website is too bad, GMB not optimized, shopping feed not structured, etc., smart campaign won’t help you either. First fix your problems, evaluate whether the campaign is the right fit for you or not. If not, just leave it and go ahead with the non-smart campaigns.

Key Takeaways

Machine Learning in Google Ads has changed our day-to-day work as PPC Specialists. We are no longer system administrators who are setting up campaigns, managing bids, spend hours on reporting, etc. We became real marketers now! Our role as Google Ads Managers has evolved; in my opinion, in a positive way. As long as you apply the following, you will enjoy your work more than before and your life as a ‘Digital Marketer’ will be more important:

  • Strategy is key. Invest time in analyzing and planning.
  • Focus on your creative skills. Emotional Intelligence is stronger than Artificial Intelligence.
  • Get out of your PPC box and look at marketing holistically. Expand your horizon!
Halide Ebcinoğlu
10 Sep 2019

Not Happy With Your Online Performance?

Just Get In Touch With Us!

Be our next client


Our team is here to quick audit your ppc campaigns, organic traffic performance or website usability.

Just Get In Touch With Us!