How Should We Expect Google Ads Targeting To Change In 2021?

Table of Contents

Last month, Google accidentally teased a possible future for targeting on its ads platform. The story goes like this: when an advertiser went to add a keyword, they saw a new option, “Smart Matching.” They tweeted about it. Google quickly released a statement, saying it was a test meant for only a few advertisers. And the paid search community was a little confused. 

Google tests things. We get that. But what they test sends signals that shine light on how search advertising might evolve in the coming years. This is where the PPC community can get a little conspiratorial, I’ll admit. We read Google blog posts and press releases hoping to unearth some hidden, cryptic signal of how and when the platform will change next. 

In this case, how exactly the new keyword type might work is still a mystery. Whether it will even exist is also up in the air. 

That said, I’ve got my red string, and I’m ready to parse through the trends hidden in the steady clip of Google Ads updates we’ve seen. I’ll dive into the powerful forces driving changes to keywords and ads targeting. I’ll discuss why they’ve changed so drastically, and how our old friend, FLoC, is just the next link in this chain.

Overall, I want to paint a picture of what these changes might mean for our strategies in the future -- a future which could rely increasingly less on keywords. Get your tinfoil hats.

A Brief History Of Keywords

Before I break out the red string and sidewalk chalk, let’s zoom out. Keywords have been changing for a while, but here’s a quick timeline of the major updates we’ve seen over the last couple of years.

Looking at that, I’d wager we’ve all had something like smart matching in the back of our minds. The trendline is clear. Google has pushed a “smarter” brand of PPC. It’s continued to advocate for automated bid strategies. And it’s expanded keyword match definitions while reducing the data we as advertisers have access to. 

We can’t really deny that we’re on a path to broad, machine-learned keywords.

But why are we seeing this? And what can we do to be ready for it?


Before I get on my soapbox, there is some benefit to these automated bids, wider keyword targets, and other new features. They make for a better algorithm. 

As I’ve talked about in earlier rants, these changes give Google more data, because machine learning requires a lot of it. In turn, the Google Ads algorithms can identify more trends, make better predictions, and, in theory, help get you more bang for your buck. 

What we’re seeing with each new update and each new feature are the benefits of better algorithms and predictive models. And we owe that to the data Google has been able to collect from advertisers, to the less-than “exact” keyword matching many of us have rallied against. 

So when we talk about these changes, it’s important to acknowledge that they are, as every Google blog post will say, intended to create a better product. 

But it’s also important to acknowledge the massive incentive Google has to maintain a steady flow of data. This lets it create more products and test new features, which keep it competitive and lower the barrier to entry. It also lets it advertise a “better,” “smarter,” and “more targeted” platform, which again increases advertiser adoption. 

Google has every reason to keep expanding its targeting to feed the algorithms and the cash registers. And we have every reason to believe they’ll keep doing that. The only question is what that will look like for users and advertisers.

Data Privacy

There’s another trend driving these keyword changes: privacy.

PPC marketers have been spoiled by the keyword. It’s a neat, little packet of high-intent granularity that so many other channels just do not have. Nowhere else can you target the highest-converting users based on how they’re talking and thinking right before they want to buy. 

Keywords are remarkable tools in-part because they’re so specific. But as ad targeting and tracking is under the knife in the waning days of the pre-FLoC world, that specificity is their biggest problem.

The truth is that keyword data is, in many ways, personally identifiable. If we don’t aggregate this data, it’s possible to discover a user’s location, device type, and much more after a single search. If you’re an advertiser, that’s great. If you care about privacy, that’s a problem.

Recent platform updates have made things more anonymous. We’ve cut back on search term visibility, and we’ve limited advertisers’ abilities to stitch together keyword and user data. As we get ready for major changes to the cookie, we should expect these updates to continue. To-boot, with mounting legal and social pressure, we should also expect Google Ads targeting to adapt further to this more private, more anonymized online world.

An Obligatory Section About FLoC

The Federated Learning of Cohorts (FLoC) system Google has proposed is the next iteration of online tracking and ad targeting. I’m not going to get into a deep, technical overview of FLoC. And I definitely won’t weigh in on the myriad debates raging about it. I will, however, use it to demonstrate how we should be thinking about targeting going forward.

But first, a little context. 

Currently, we can target users across a number of axes. There’s the keyword, sure. Layered on that, there’s also in-market, behavioral, and demographic audience data. These are collected via third-party cookies that track browsing behavior across websites -- cookies like the snippet we all know and tolerate. 

Google uses that data to group users by interests, to understand their demographics, and to decide what kind of products they might be looking for right now. 

With FLoC, that changes. Google has already announced it will not replace behavioral audiences. Instead, it will group users based on browsing behavior without a cookie. Browsers with FLoC enabled will collect information locally. They will use that data to bin similar users into a “cohort,” which it will share with websites and advertisers. 

Advertisers will be able to use their own Google Analytics audiences and other customer lists from a CRM. But behavioral audiences will look a lot different in this new ecosystem. I won’t mince words. This targeting will be worse. We, as advertisers, will likely see a dip in performance at first. Still, there are things we can do to get ahead of this change and minimize potential disruption.

The Role Of Audiences In PPC

Taken together, in-platform automation and broader privacy regulations require a targeting tool that offers reliable data to feed the machine without sacrificing individual privacy. That tool is audiences.

Audiences are broader, aggregated pools of similar users. By definition, they include more varied search behavior; they include more data. While there are still many privacy problems with audiences, they do not give away personally identifiable information as easily. By grouping users together into audiences or cohorts, you hide individuals in the crowd, but you preserve the shared traits that unite members of the audience. 

Let’s recap quickly. Google is expanding the reach of its keyword match types to get more data and anonymize individual users. At the same time, it’s building up a new tracking and targeting mechanism focused on groups of users and cohorts. 

My prediction is that audience targeting will only become a bigger part of our PPC strategies. Google will continue to incentivize advertisers to leverage first-party audiences and FLoC cohorts to feed its bid strategies and limit access to user data. 

It’s still unclear how cohort targeting will look, but we know how our first-party data looks. Your customer lists and Google Analytics audiences will be pretty much the same, unless you’re hyper-focused on behavioral segments. In any case, this next evolution of PPC is going to challenge us to start using these audiences to our benefit.

Performance Impacts 

Let’s be clear; this change will hurt performance in the short term. And our returns might not rebound. When we lose granularity, we should expect profitability to take a hit too. 

On the plus side, some of the scale of this performance dip is in our hands. If we’re gearing up for changes to third-party audience tracking, then it follows that companies with solid data collection have an edge. If you have better, more robust first-party audiences, you can leverage those in your ads program. If you don’t, you might watch the rug slide out from under you.

Still, there are some things we should all be doing to limit drastic changes to performance in the coming year. It all comes down to first-party data collection and audience organization.

Actions To Take Now

These are the things I’ve begun to do in my accounts.

  • Observe All Remarketing Audiences in All Campaigns
      1. Upload all of your first-party audiences to Google Ads and set the targeting to “observation.” This will show you just how many existing users your ads are reaching. For your brand terms, it will give you a gauge of how likely users are to re-engage or how much spend you might be wasting on users who already converted.
  • Create Customer Look-Alike Lists
      1. This is something I’ve only recently begun to try as a primary tactic. If you have enough customer data to do it, create in-platform look-alike lists to prospect users who behave similarly to your existing customers. While this capability will change over time, the important thing is to get the learnings -- and the data -- into your bid strategies while you can.
  • Measure Everything
      1. Tagging, email capture, and first-party data storage have always been king. That’s doubly true now. Your marketing and sales team need to be in constant communication, either directly or through the CRM. Be sure your leads are properly scored, and that you’re doing everything you can to make specific customer lists. Also be sure your Google Analytics audiences are populating with relevant users at all stages of the funnel.
  • Give FLoC A Chance
    1. Once we have this new behavioral targeting tool, let’s not look down our noses at it. Though it will probably be lackluster, it’s worth a test. If cohorts are targetable in ads, observe them. See how their performance stacks against your historical audiences. With the right settings, and the right layers, they might just do okay.

The Gospel Of Google

Our role as advertisers is shifting towards something akin to “algorithm managers.” We manage inputs, and the machine behind the curtain churns out performance: deus ex machine learning. 

I don’t see these trends abating any time soon. If we really dive into it, I’m preparing for a Google Ads experience that looks a little more like this...

  • All keywords are broad match (or smart match).
  • Keywords become guiding suggestions for the bid strategy, not concrete targets. 
  • Display ads & YouTube are rolled into each campaign by default.
  • Audience targeting (or FLoC cohort target) becomes mandatory to launch a campaign.

And that’s okay. It frustrates every fiber of my marketing being to lose specificity, but it’s okay. It doesn’t change the fundamentals of good marketing. 

I miss the exact keyword, and I hope Google doesn’t mirror Facebook too much with its heavy audience focus. But we shouldn’t mourn these changes. We should adapt. The onus is now on us to target our audience as effectively as possible without violating their right to privacy. 

John Smith

John Smith

John is a Paid Media Manager at Uproer, where he works to build paid search strategies for clients in the e-commerce and SaaS spaces. He's drawn to the ideas, channels, tactics, and emerging trends that tackle big issues in marketing. And he approaches SEM with a focus on data privacy, incrementality, and social impact. When he's not knee-deep in a spreadsheet, John volunteers with local climate organizations and helps spread their message through search.

See More Insights

SearchLite - Don't Sleep on Category Page Content

This month’s SearchLite intro was written by Content Manager, Skye Sonnega Hey everyone, We recently livened up this dull Minnesota winter with two fantastic additions to the Uproer team! Eric Davison joined as a Senior SEM Analyst, and Jenny Hudalla joined as a Content Specialist. These folks are bright, hardworking, Minnesota-local, and coming in

Read More

Sr. Manager, SEO & Operations

Dave Sewich

Dave made an accidental foray into digital marketing after graduating from the University of Minnesota Duluth and hasn’t looked back. Having spent the first part of his marketing journey brand-side, he now works with the Uproer team to help clients realize their goals through the lens of search.

When not at work, you’ll find Dave staying active and living a healthy lifestyle, listening to podcasts, and enjoying live music. A Minnesotan born and raised, his favorite sport is hockey and he still finds time to skate once in a while.

Dave’s DiSC style is C. He enjoys getting things done deliberately and systematically without sacrificing speed and efficiency. When it comes to evaluating new ideas and plans, he prefers to take a logical approach, always sprinkling on a bit of healthy skepticism for good measure. At work, Dave’s happiest when he has a chance to dive deep into a single project for hours at a time. He loves contributing to Uproer and being a part of a supportive team but is most productive when working solo.

Founder & CEO

Griffin Roer

Griffin discovered SEO in 2012 during a self-taught web development course and hasn’t looked back. After years of working as an SEO consultant to some of the country’s largest retail and tech brands, Griffin pursued his entrepreneurial calling of starting an agency in May of 2017.

Outside of work, Griffin enjoys going to concerts and spending time with his wife, two kids, and four pets.

Griffin’s DiSC style is D. He’s driven to set and achieve goals quickly, which helps explain why he’s built his career in the fast-paced agency business. Griffin’s most valuable contributions to the workplace include his motivation to make progress, his tendency towards bold action, and his willingness to challenge assumptions.