AI is all the rage in the digital world now, and paid media is no different. While there are certainly lots of ways you can use AI outside of the ad platforms to enhance your campaigns, there are also features built directly into the platforms themselves that you can leverage to infuse AI into your campaigns.
In this article, we’ll dive into the AI features available right now to help you personalize and optimize your ad campaigns broken out by audience targeting, ad creative tactics, and budget and bidding controls.
Contents
- Behavior and interest targeting
- Lookalike audiences
- Automated targeting (with and without guidance)
- Text and image asset creation
- Dynamic service
- Ad enhancements
- Predictive performance forecasting
- Smart Bidding strategies
- Dynamic budget allocation
AI features for audience targeting
It may be a little more behind the scenes, but many of the targeting options we have access to within the ad platforms leverage AI in one way or another.
1. Behavior and interest targeting
All of the targeting that is housed in the Behaviors and Interests sections of the major platforms are based on actions those users have taken around the web, both on platform and not. The AI portion comes into play when Google, Facebook, Snap, or another platform analyzes those user patterns and then groups them into the targeting options you see in the Ad Manager.
On Facebook, we’re able to target users who are interested in Advertising and Marketing simply by checking a box. That opts us into targeting users who have “expressed an interest in or like pages related to” Advertising and Marketing. All determined by Facebook’s AI systems, which we typically refer to in the ad world as machine learning.
🛑 Are your search ads driving the best performance? Find out with a free, instant audit >> Google Ads Performance Grader
2. Lookalike audiences
For an even more automated approach, i.e., using even more AI/machine learning in targeting, advertisers can leverage a few tools that don’t even require behavior and interest selection.
Lookalike audiences on Facebook have been around for a long time and are still one of the better-performing targeting options available on the platform, in my opinion. For this audience, advertisers need to create a seed audience of desirable users, whether it’s customers, pipeline, or some other high-value group, then Facebook will analyze that user group for patterns and create an audience that “looks like” them. Hence, lookalike audiences.
Many other platforms have their own version of lookalike audiences, but they all work in about the same way.
An example of a Snapchat lookalike audience set up.
👀 See how your Facebook ads could improve with our free Facebook Ads Grader >>
3. Automated targeting (with and without guidance)
For the most AI-driven targeting, I want to use both Google and Facebook as examples.
In recent years, both have released new targeting types that work in a more novel way: Advertisers can suggest targeting that the platforms take into account, but they’re able to target anyone the algorithms believe is potentially going to drive performance.
For Facebook, this is called Advantage+ audience.
When you opt in, you’re able to give suggestions for targeting, leveraging all of the regular tools you have at your disposal, including saved audiences, demographics, behaviors, interests, etc. As a note, Facebook will still honor all location and language targeting you set. This is not dynamic.
Typically, the platform will start off by targeting your suggested users, but then expand beyond that list once it finds good patterns and sees results.
Google’s offering is slightly different in that it doesn’t explicitly say it prioritizes the targeting you provide first while learning, but it’s known that it will go after remarketing audiences to see early performance.
This targeting option is also a full campaign type: Performance Max or PMax. These campaigns run ads across all Google-owned properties and the Google Display Network, giving them the widest reach of all the campaign types on Google.
With Performance Max campaigns you’re given two tools to influence targeting.
First is search themes. These do not operate quite like keywords, but instead let advertisers indicate search queries you believe your customers may be looking for. These are optional, just like all other suggested targeting in this section, and are additive to any of the machine learning Google uses to find your target audience.
The second option is audience signals. Rather than search queries, this section lets you use audience segments to influence PMax targeting. For this option, you can use your own audience signals from remarketing lists or data from Google like Demographics and Interests and detailed demographics targeting.
AI tactics for ad creative
When it comes to the ad creatives themselves, there are a few ways the ad channels use AI to enhance the user experience.
4. Text and image asset creation
The simplest version of AI in ad creation happens across just about every platform now.
Whenever you begin creating a new ad in their editors, just about every platform will have some version of what you see above in Google Ads. The platforms will suggest text variations you can use in each of the different fields, usually based on the website you provided and other assets already in your account.
Both Facebook and Google now take AI ad creation one step further and have tools that will generate imagery for your ads as well. For Facebook, this is done through www.meta.ai, and on Google, this can be found in the asset creation portion of the interface under “Generate Images.”
With just a simple prompt, these platforms will create images for you to use in your ads, and you can tweak and adjust them to your heart’s content before you launch.
An example of an AI-generated image from the Meta platform.
🤖 Use AI in your campaigns the right way. Get the guide >> 9 Powerful Ways to Use AI in Google & Facebook Ads
5. Dynamic service
Once your ads are created, depending on how many assets you included, many ad platforms also leverage machine learning to dynamically serve ads with the combination of assets they think will perform the best.
With responsive search ads on Google, you’re given up to 15 headlines and four descriptions. Depending on how you’ve leveraged pins, Google can then create almost any combination of those assets to fill the three available headlines and two available descriptions for each ad.
Facebook has a similar option with its ad creatives as well.
Each time an ad auction takes place, the platforms use machine learning to dynamically generate an ad from your assets that it thinks will perform best for that given user, and is focused on achieving the stated goals of your campaign.
While this use of AI is entirely behind the scenes and out of your control, aside from the assets you provide, it’s a highly important and impactful use of AI in the ad platforms.
6. Ad enhancements
This last option is really just for Facebook at this point but is still important to point out. When you create ads on Facebook, you can opt into a number of what are called Enhancements.
These are all options that Facebook can either adjust your current creative or add to it with additional media to enhance the user experience.
There are eight different options for enhancements, including music, site links, text overlays, visual touch-ups, 3D animation, text improvements, image animation, and brightness adjustments.
Some of these you simply opt into, like the brightness adjustment, but others you can specify how you want the enhancement to look, even going so far as to choose the text and colors as well as the music. So even though they’re “automatic” enhancements, you do have some level of control over a few of the options.
AI features for bidding and budget controls
The last portion of AI within the ad channels I want to address will be around all of the algorithmic inputs for bidding and budget controls. But before we get there, we need to address the performance prediction abilities within the platforms.
7. Predictive performance forecasting
No matter which ad channel you use, their backend platforms are always trying to determine what it thinks your performance will be based on factors from other campaigns. Those could be campaigns within your own account or from other businesses in a similar sector. That’s why the platforms are almost always providing some estimated performance metrics within their interfaces, even if you’ve never advertised before.
But that doesn’t mean you can’t have an impact here. If anything, it’s the opposite.
The more information you give the systems, the better they can perform. And I’m not talking about just giving the platforms more budget.
Although Google claims they don’t need any conversion volume to optimize your campaigns for performance, the truth is they will always do better when you have a solid foundation of conversions to go on. The more data, the better.
The same is true for Facebook. The Learning Phase is a time period where the platform is still learning how to get the best performance from your ads, and conversion (or result) performance is a big portion of that.
So, although the platforms are very smart and always looking to forecast performance in your campaigns, the best results will always come when you’re using actions that have lots of volume and good data associated with them.
8. Smart Bidding strategies
As a result of the platforms having predictive performance models, they’re able to adjust your bids in real-time as you enter the auctions for different users.
Google Ads has quite a few automated bidding strategies that can optimize toward just about any performance outcome you’re looking for.
Facebook also has a couple of different options depending on if you’re looking for volume or profitability.
No matter which you choose, just as I mentioned before, make sure you’re providing the platforms with as much data as possible to help them see patterns and predict where to best serve your ads for optimal results.
9. Dynamic budget allocation
The last way we can use AI in the platforms is to let them determine where our budget gets allocated. I left this one for last as it’s the extension of basically all the other factors combined.
Based on the conversion performance we see in a given campaign and the forecast performance, the platforms have some settings to choose where our budget should be spent.
For Google Ads, this can be used via a Shared Budget. With Shared Budgets, you can combined multiple campaigns into a single budget, then Google will decide where to allocate the spend based on predicted performance. Rather than having $20/day allocated to each of five campaigns, you could have $100/day allocated across those campaigns, then let Google decide where those funds are best spend.
Facebook’s answer to this is called Advantage+ Campaign Budget. The same logic applies, just across the ad sets within a campaign. Rather than setting individual budgets, you can add a budget at the campaign level, then Facebook will dynamically adjust the spend across ad sets based on what’s been performing well and what’s expected to help reach your campaign goals.
Take advantage of AI to give your ads a boost
While there are always going to be much sexier versions of AI being talked about in external tools that may or may not cost money, don’t lose sight of the fact that there’s plenty of AI functionality baked directly into the ad platforms themselves. Whether you realize it or not, those systems have likely been helping you achieve better results for a while now. Hopefully, highlighting how AI plays a role in the interface will help you see it and take better advantage of it for personalizing your ads in the future.
For help automating your ads even more (AKA not running them yourself), reach out for a demo.
