In this week’s instalment I have one sharp new tip, and a few reminders of other key ways to use Smart Bidding without getting into hot water.
As PPC account managers, many of us want some control. Yet under typical conditions, Smart Bidding in Google Ads gives us zero levers to pull. Boo.
Well, look over here as I peel back the vinyl covering (it looks like a BBQ cover, but it has the Page Zero Media logo on it)… look, secret levers!
1. Cap CPC’s, even under Smart Bidding
An examination of a Smart Bidding campaign might show that under the automated bidding, Google handed you, on average, much higher CPC’s.
This strategy isn’t easily accessible under Settings. It only becomes available once you set up what’s called a Portfolio Bidding Strategy. All that it means is that it’s a Bid Strategy setting you store in the Shared Library.
Figure 1: You’ll need to dig into Tools & Settings > Bid Strategies, and create a Portfolio Bid Strategy, to tap into the additional control mechanism (capping CPC’s) which may improve Smart Bidding performance in some cases.
The controls are lurking and hard to find.
Figure 2: You’ll need to scroll to the bottom of your Portfolio Bid Strategy status page (or during a Portfolio Bid Strategy setup). And then click the little blue link, “Advanced Settings.” Then this screen becomes available to you.
2. Go narrow with geography
Especially as times change with the pandemic economy, perhaps you have international campaigns spanning several countries that no longer make sense. In theory, Smart Bidding will eventually twig to the fact that your prospects from Belarus or Brazil will no longer be buying at the same rate. Why wait around for the system to ponder the nature of the current universe? Take control. Remove all but your most reliable country markets from the locations you’re targeting. If times change, you can always revert to the more international focus.
3. Be intentional with match types
Using less traditional broad match will give the system less license to play crazy tricks with queries, either testing too many idiotic ideas or cannibalizing sessions away from other campaigns they should remain associated with. Continue to negative out non-converting or clearly irrelevant queries to improve efficiency. That being said, you might be better off with the full range of match types, presuming the keywords are highly relevant and you’ve taken other refining steps such as putting upper and lower limits on CPC’s.
4. Put Smart Bidding on a leash by keeping its share of traffic throttled via a Campaign Experiment: forever.
There’s no rule that says you have to run all of your traffic through a Smart Bidding strategy. Google’s fervent hope is that you test and find that a Smart Bidding strategy “wins” over some version of human management, so you turn on Smart Bidding and never question it again. Keep it honest. Run it 50-50 or even 30-70 in the split of a Campaign Experiment. Forever. Or at minimum, one full year. Why is Google in such a hurry to herd you, the human analyst, off your own campaigns? P.S. You can keep managing “your” side (the Control).
5. Take charge of ad copy
Using your own testing criteria, keep the ads that work best, and phase out the ads that persistently fail. And ignore advice about how to best make Smart Bidding happy. If a greater number of ad versions seems to work for your business in this environment, then use more. If this isn’t true, pause or delete ads that never seem to come around.
Responsive Ads, in theory, offer an incredible testing ground because of the high number of available permutations. If everything falls into place and you’ve been really diligent with your research and creative messaging, Responsive Ads combined with Smart Bidding should show the exact best combination to each and every quirky and different user (as I intimated in Part 4). Note also that traditional advertising tech forced us to target our marketing to aggregates, whereas in the future, AI and clever setup methodologies will allow high-volume advertisers (especially in consumer goods, content, and other things involving variety and long-tail preferences) to show ads tailored to beautiful snowflakes, a view of the consumer I alluded to in Part 12.
But three things may prevent good outcomes:
- First, you might not have enough differentiation in your ad assets, so one version beating another is purely due to randomness.
- Second, with upwards of 10,000 permutations possible just from one of these multivariate ad units, you might not be coming close to the volume you need to wring out enough behavioral feedback to arrive at the “right” answer(s).
- Finally, the system might well have a bias towards letting itself learn for a protracted period of time. You need to make a profit now, not train Google’s bots.
For these reasons, if you’re most comfortable running just a couple of regular ads (ETA) that have been proven winners, then you do you! Do, however, ensure that you return to inject new ideas and variations into the ad testing process now and then. The time-honored maxim “Garbage In, Garbage Out” may apply to your creative effort, no matter how adroitly you or a bot can yank on various targeting levers.
Read Part 26: A PPC Bidding Hack That Could Save You Money on So-So Queries