Mike Rhodes is big on Google’s Artificial Intelligence (AI). You know this if you’ve listened to interviews he’s done recently. I detailed the best bits from his recent Perpetual Traffic podcast.
And if your ravenous hunger for Rhodes wisdom persists after reading this, then you should check him out on the Superfast Business podcast. It's also excellent.
Rhodes is the guy that convinced me it was time to hand over the keys to Google Ads because their Machine Learning and AI were so good, and that manually increasing bids could no longer compete.
In the past it was prudent to roll your own, so to speak, because Google was still getting its shit together with the AI. Anybody signing up to allow Google to take over accounts was basically donating a healthy percentage of their budget over to Google for their R&D purposes.
Mike and his team refused to do this. They dug in. They didn’t listen to the sweet siren voices of their reps telling them how easy it was, and how much better the results would be.
But, in the last year or so, he changed his tune. The threshold has been crossed. Critical mass has been achieved. This is not easy to admit for a proud man with so much hard-earned knowledge about how to manually squeeze the most out of a Google Ads campaign.
Rhodes is really smart, though, and knows when it’s time to change up the approach for the betterment of client results.
Google Cracked the Code
Google has cracked the code. Give them a Target CPA and let the AI do the bidding. (Notice I didn’t say do the bidding of the AI, we haven’t hit SkyNet level yet).
With Responsive Ads, simply give Google more headlines, descriptions, images, and logos to test and let them find the most powerful combinations.
Interestingly, Ralph Burns and Molly Pittman have announced similar results with Facebook’s AI and are moving to the Campaign Budget Optimization model to take advantage of how smart the algorithm has become.
I started testing Target CPA in Google Ads in November 2017 at my day job, and switched over all my campaigns a few months later after test results showed it working much better than my manual stuff.
I haven’t looked back. It saves me a ton of time and the results are better.
Responsive Ads Are Helping Google Intern-Written Ads Perform Better Than Mine
Recently, I let my Google Ads rep talk me into having "the Google team" write ads for me.
With recent team member departures, particularly content writers, I was desperate to get more ads into my account. Especially ads that took advantage of character expansion (I’ll take as much SERP real estate as they’ll give me for my ad dollar).
How about another headline and description line? Yes please. Now, the expanded ads don't get served up every time, but you might as well have them in there for placements that do.
These ads are also responsive, meaning you can plug in up to 15 different headlines and four description lines, and multiple images and logos. The machine will mix and match to find the best combinations.
It’s a fairly new technology, and it lacks reporting. I can’t drill down and find data on each specific asset combination at this point, which would help guide further asset creation, but I’m sure it’s coming.
Anyway, the Google team gave me ads for 19 search campaign adgroups. These ads averaged five headlines and three descriptions.
After running for a month, nine of the 19 ads are the best performers in the adgroup, five are middle of the pack, and five are losers. Not a bad result.
Honestly, I didn’t expect much. I figured they had AI cranking out the first draft with interns cleaning them up and passing them over. Yet, nearly half are performing better than mine. I attribute a lot of this to the responsive ads, and expect to have even better ads when a staff members, with more knowledge of my company’s offerings and customers, write them.
I definitely need to have these ads in all my campaigns.
Tweaking Target CPA Depending on Daily Budget and Impression Share
I have a strict monthly budget at the day job. Earlier this year a bunch of my campaigns were "limited by budget." My rep suggested, if I couldn’t increase my budget, then I should lower my Target CPA for those campaigns.
He reasoned that it was better for Google’s AI to go find people more likely to convert at a cheaper cost/conversion (smaller pool), than to try and find a conversion with a higher Target CPA, which opened it up to more potential people, thus loosening the targeting and reducing the conversion rate.
To help figure out which campaigns were good candidates for a lower target CPA (beyond being limited by budget), I added the "impression share lost by rank" and "impression share lost by budget" columns. Any campaign losing significant impression share because of budget were prime candidates for lowered Target CPA.
It worked. I circled back after a week, and lowered them even further if I was still limited by budget. I continued getting more conversions and a cheaper price with this exercise.
However, my budget increased in subsequent months, and I discovered I wasn’t pegging my daily budgets in some accounts.
So, recently, I bumped up the Target CPA in hopes of expanding my pool of people to show my ads to. It will be interesting to see how this affects my numbers. Hopefully, I’ll get many more leads at minimal increase in cost per.