Google’s Machine Learning — Your New Advertising Crush

Illustration of a brain inside of a head, surrounded by various icons

In this post I want to take a look under the hood at something special and shiny that’s appeared within Google Ads: Machine Learning.

I recently went hunting for some internal decks on how Machine Learning for Google Ads works, but I discovered they read like stereo instructions — unless you know the guts of their ad system really well, it’s not going to make much sense. So instead, here's the basic version:

The Swash Code

Swash Labs has worked hard to 1) build a real system that measures as much about ad campaign performance as possible, and 2) maximize performance with targeting and data science. This is always married to a specific function — if we want to generate leads for a service or sales on a website, for example, we maximize around that goal using data and real measurement. As a result, our performance indicators — cost per click, cost per conversion, return on ad spend (ROAS), etc. — generally run ahead of industry standards.

Google Badge of Honor

But back to Google. Swash Labs is proud to be a badged Google Partner Agency. We earn this status by maintaining mandatory advertising spend levels, and we keep up-to-date with platform changes to ensure the most benefit to our clients. This is truly a team effort. Our Media team (Scott, Charlie, Stephanie) is constantly learning and maintaining certifications in multiple disciplines and programs offered within Google Ads, all while making sure our quality parameters stay within Google’s guidelines. And a large part of our quality rankings can be attributed to the copywriting work of Faith and Jessica, and images/graphics from Andi and Todd in the Art Department. We work hard to keep our clients’ ad accounts operating above benchmark standards. And we finish it off, team-style, by demonstrating excellent customer care in the way we manage Google Ads for our clients; we make sure that in a world of bad actors, we are wearing the whitest hats available when it comes to ethics in digital advertising.

Our Special Sneak Peek

It is because our clients trust us with their continued investments on the Google advertising platform that, earlier this year, Google invited us to participate in an early, advance training for their new suite of Machine Learning-based ad optimization tools. These tools are revolutionizing digital ad buying.

Machine Learning is Artificial Intelligence with a narrow job function and skill set. In this case, the AI's job is digital advertising efficiency. The result is a ruthless, merciless robot that sits on top of campaigns that were already performing well and turbo charges them.

Essentially, it takes the targeting we build inside of Google Ads, which account for about a thousand dimensions (geography, behavior, expressed preferences, search history, cookies, demographics, etc.), and then integrates 70 million behavioral signals from Google's entire data set.

The end result is that a campaign with prior excellent performance gets even better, or more specifically, gets more efficient by learning which ads lead to specific desired behavior (clicks, conversions, calls, etc.). Then, when given enough data, the campaign can choose an audience with a higher likelihood of conversion. A possible result of this increased efficiency is that something which used to cost X dollars to do now costs less, or that the expected return for a certain conversion behavior may increase. In short, if you know what you're doing, Machine Learning takes campaigns that already work and makes them work ridiculously well. And this is good for us... because we know what we’re doing.  

What Can Machine Learning Do For You?

I think Machine Learning could be impactful, if not transformative, for any business not currently making use of the technology. And we’re one of the most experienced, best-equipped shops for helping a business turn that corner.

Josh Berthume