Optmyzr Blog

PPC in 2022: Predicting the Unpredictable

Frederick Vallaeys

Co-Founder & CEO



If the last two years have taught me anything, it’s to expect the unexpected…and the unpredictable. While I don’t have a crystal ball, I do have some thoughts on where PPC is headed. Here’s what I think will matter in 2022.

The Three Truths

There are three ongoing trends that I expect will continue to shape our industry. These are the “three truths” of PPC advertising:

  1. Ad platforms will continue to automate.
  2. Advertisers will get less access to data.
  3. Control over targeting, bidding, and messaging will keep decreasing.

We’ll Shift to Managing Goals Rather than Details

As automation continues to improve, the need for granular controls will decline. Instead of managing PPC details inside the ad platforms, our roles will shift to managing higher-level goals at these systems’ periphery. Less managing search terms, more managing how we identify good leads and conversions.

Even Facebook recently added “Lead Conversions” to their platform. Advertisers are getting smart to the fact that leads are meaningless if they don’t convert into customers.

We’ll Rethink Old Assumptions

In the good ol’ days, if you were selling something on the internet, you could assume that there was basically an unlimited supply of the thing you were selling. You could go full steam ahead, offering big discounts and special promotions, and aggressively pursue campaign expansion. You knew the limiting factor was how many buyers you could drum up and not how much stuff you had in inventory. The supply-chain disruptions of 2021 changed that. You can no longer assume unlimited supplies. That means a modern PPC marketer needs to be more nimble than ever.

Here’s another assumption we’ve had to challenge. It used to be that, to get more traffic to an ad group, you had to add keywords or raise bids. That was true before RSAs, which are now the default and will soon replace ETAs. In our recent study of the RSA ad format, we found they can drive four times as many impressions as ETAs, simply because Google’s machine learning is able to boost the relevance of an ad by constructing it on the fly, based on a user’s query and circumstances.

Our decades-old best practice of using the metric “conversions per impression” to find the best ad in an A/B ad test was based on an assumption that is no longer true. To do A/B ad testing right, you now also need to consider that each ad has a unique impact on the ad group’s impressions and hence its ability to drive incremental conversions.

Here’s another false assumption: the higher the ROAS, the better. That was never really true, but advertisers are finally coming to realize this and are beginning to favor POAS (profit over ad spend) instead, since this metric better reflects actual business goals.

We’ll Have to Continue Grappling with New Questions

If you’re selling new cars, you’re faced with the question: do you actually need to advertise? You may already have a one-year backlog due to supply-chain disruptions. Why advertise just so that people can sign up on a waitlist? Is that really necessary? Or is tight supply your opportunity to reset the expectation that cars will be sold below MSRP into one that they will now command a premium over MSRP and you want to double down on ads while the margins are high?

If you sell fitness bikes, and orders get canceled in droves when patience with shipping delays runs thin, do you allow your systems that adjust order values to Google to continue as before? Do you mark these conversions as bad? Or was the fault actually not with marketing finding low-quality buyers? The answer may seem obvious, but whatever you decide has implications on what the machine learns and therefore does next, so tread carefully.

We’ll Need to Rely More on First-Party Data

Before ads became personalized, they were just annoying. Personalized ads are much more useful and relevant. That advertisers are willing to pay a premium for better targeted ads is what’s been floating the rapid development of the web in the last couple decades.

The most important method of personalization has been third-party cookies, which enable the system to draw a far more detailed demographic and psychographic picture of each individual user. But third-party cookies are now considered invasions of privacy and being eliminated. What happens when third-party cookies are gone?

If we can no longer do some of the personalization that made the internet useful, many websites are going to lose considerable revenue. Google did a study and concluded that in the absence of personalization, 62 percent of revenue for online newspapers would go away.

That is a real problem. As is so often the case, a major shift has unintended consequences. Privacy is a desirable goal that may well cause the web as a whole to decline in value. That could be catastrophic. But it’s a problem only if there’s no replacement for third-party-cookie technology. To prevent this, new ways to let advertisers continue to subsidize a free web need to be developed.

Several initiatives to protect user data, while still enabling interest-based advertising and remarketing without third-party cookies, are in development. The most prominent has been Google’s now-discontinued FLoC (Federated Learning of Cohorts).

While we can’t predict the exact solution, it is certain that it will involve machine learning, which draws inferences from masses of data. The more data a machine-learning system is trained on, the more accurate it will become at audience segmentation and personalization.

One possibility is that audience segmentation will be based on devices surveyed, not individual users. I tend to use my laptop for business and my iPhone for personal matters. Under the new system, it’s possible that I’d fall into different cohorts depending on my device and that could actually be really helpful to advertisers who could then target cohorts of work personas or personal personas.

All of us as individuals have many dimensions, which so far have been hard for advertisers to target. Our technical problem-solvers may be able to make a virtue of the necessity of restoring personalization by other means.

The problem is so urgent that we can expect any number of solutions to be developed in the coming year. As always, there will be a competitive shake-out of the various approaches, with one or two eventually emerging as the favorites. We’re well-advised to keep an eye on this trend in 2022. Cliché though it is, it’s also true that every crisis brings new opportunities.

And while waiting to see how things shake out, advertisers should work overtime to increase the first-party data they collect, so they can cut back their dependence on data from Google, Amazon, Microsoft, Facebook, and the other platforms.

Happy New Year

In a recent PPC Town Hall, Chris Moreno from Google said, “The only thing that we know is that we don’t know anything.” Basically, 2022 is going to be as full of surprises as the last two years were. We need to be ready to be nimble and able to respond to whatever may happen.

Happy New Year and good luck to us all!