Machine Learning (ML) may sound like a futuristic and complex concept to implement in our advertising campaigns; however, Facebook and Google already have some powerful ML techniques that you can implement with some small tweaks. The current economic landscape is making a lot of companies review and adjust their digital marketing efforts. If you have already reviewed and confirmed your campaign is being properly managed but you still need an extra push, ML could give you the edge your campaigns need. And truth be told, the sooner you start using these features, the better. You don’t want to be left behind by your competitors that do use the power of ML to their advantage.

Important note for marketers before starting

When using ML in our campaigns, we are leaving some of the old best practices behind. Historically, when managing digital advertising efforts, we want granularity and control. For example:

• Several ads per ad group in Google that we turn off manually depending on the result.
• Specific and small audiences in each Facebook ad set.

This structure allows marketers to understand what is working best and to get better quality scores in different advertising platforms. When using ML, we are making a tradeoff between control and the speed at which we can discover optimizations. Instead of several ads, we have one smart ad with different elements (images, headlines, etc) running different combinations until it finds the best performing one. Instead of several ad groups with different audiences, we have just a couple of rather “big” audiences as we let Facebook find what sub-audience performs the best. The trick with this new advertising framework is that it puts a lot of emphasis on choosing the right parameters to feed the ML beast. No amount of ML will get us good results if the audiences we are targeting are not a good fit or if the headlines we are using suck.

1. Target CPA in Google Ads

This feature will allow you to tell Google how much you want to pay for your conversions (e.g. an e-commerce purchase or a lead). Sometimes a conversion will cost more and sometimes will cost less but on average you will pay close to that target cost. This flexibility means you will not pay more if there is no need and that you will not lose a potential conversion because you were a few cents short. Truly fancy stuff.

Young Tom Hanks approves.

More about Target CPA.

2. Smart display campaigns in Google ads.

Now this very revolutionary stuff. You can create a campaign defining what conversions you want Google to acquire for you and they do the rest based on your conversion history and industry information. Add some headlines, copy and images you want to test in the ads and you have a very powerful campaign created in just a couple of seconds.

More about Smart display campaigns.

3. Campaign budget optimization in Facebook ads

Facebook has been pushing for advertisers to move to a simplified account structure. One reason behind this change is their Campaign budget optimization feature. You can set an overall daily or lifetime budget to a campaign that has several ad sets and let Facebook decide where to allocate your budget. Facebook will optimize based on the conversion history of the campaign to get the cheapest results in real-time.

More about Campaign budget optimization.

The future of digital advertising

Using ML for digital advertising purposes is already becoming the new standard while doing things manually is turning to be more and more inefficient by the day. The winners in this new landscape will be those who understand the power of ML and work hard at defining the best strategy to leverage it. Here at Meteor Affinity, we offer a free consultation process where we review your data and prepare a proposal full of recommendations and tips. If you then decide to proceed and implement it with us, great. If not, that’s great too! No strings attached. Contact us today and start using Machine Learning in your campaigns.