It’s here. There’s no doubt that we’re working in an age where machine-learned marketing works and works really well. I’ve recently introduced by company to automation through Google’s target ROAS technology for the PLA program I manage. While it does exist in somewhat of a blackbox, I can’t say it doesn’t work. The program, officially called tROAS from Google’s paid search engineers, allows advertisers like me to focus on higher level strategy rather than day-to-day bid management. I can say first-hand this has been a game changer and has also allowed me to focus more time on SEO for the company too.

How Does Machine Learning Work?

With PLAs, an advertiser submits a data feed of product information to Google every day. Google consumes the data in the feed to understand your products. Then, in Google Ads, advertisers set up campaigns based on their data with the ability to segment it by brand, product category, attributes, price, margin, etc. Advertisers can subdivide product partitions by whatever KPI makes the most sense for their business. Once campaigns are created and follow Google’s typical shopping campaign guidelines, the advertiser determines their target ROAS for the campaign to hit. With the target ROAS given to Google, automation takes control of the bids with its eye on the return target.

Now I’m probably making it sound more simple than it really is, but there’s not a lot more to it. As long as you have a realistic return target and a high-quality data feed, you’re ready to be successful with tROAS. I will advise that it takes time for Google Ads to really learn all of your data and gather enough conversions to make the program successful. In most cases, I would give the technology a minimum of 3 weeks to start to see a positive trend. During this 3-week period, be very cautious to make to many changes or you will change the integrity of the test.

Other Types of Automation

Target ROAS is not the only type of machine learning for Google Shopping campaigns. Advertisers may also take advantage of target CPA if it fits their business model. Similar to tROAS, an advertiser can use tCPA to give Google a target cost-per-acquisition to aim for. Over time, Google will tune bids to reach the intended CPA. And at the very least, advertisers should be taking advantage of eCPC. Enhanced CPC (eCPC) takes existing manually-input CPC bids and increases or decreases them by as much as 20% to try to convert a shopper. Like any automation Google has introduce, eCPC also shows positive results and is helping advertisers move away from manual bid management.

What Are You Waiting For?

I understand automation can be a scary topic. Frankly, it seems to really frighten the CEO of my company’s sister site. However, I will reiterate that it can work if you have the right data (make sure your product feeds are clean and you deliver a solid experience on your site), have a good campaign structure and give it time. Anyone experienced in product listing ads or paid search in general should know how to set up a solid campaign architecture.

Give it a try and give it time. I would be very surprises if it doesn’t improve your return on ad spend and revenues over time.