If you feel like the pay-per-click (PPC) advertising world has been turned on its head lately, you’re not alone. In the past year or two I’ve watched PPC advertising undergo some of the most radical changes I’ve seen in my career. Both Meta (formerly Facebook & Instagram) and Google – the two juggernauts of digital ads – have upended their old strategies. Meta is moving away from the hyper-granular audience targeting we all grew up with, in favor of broad reach powered by machine learning. Google, on the other hand, is pouring gas on its new Performance Max campaigns and automated tools, even as traditional search ads face an identity crisis in the age of AI answers. It’s an exciting, perplexing, and sometimes slightly terrifying time to be a PPC marketer!
In this post, I’ll break down exactly what’s changing on both platforms and why. Just imagine we’re chatting over coffee about why your Facebook ads and Google ads suddenly feel so different.
Meta: From Micro-Targeting to Broad Strokes
I remember the days when running Facebook and Instagram(now Meta) ads felt like being a kid in a candy store of targeting options. You could pinpoint users by insanely granular attributes – from the obvious demographics to interests as niche as “people who love Belgian chocolate and live with two dogs.” Back in 2017, Facebook was the platform for ultra-precise targeting, letting advertisers reach highly specific audiences with ease. Fast forward to today, and that world has basically vanished. In the last few years, Meta has drastically changed its advertising approach, taking more and more targeting control out of our hands and entrusting it to its algorithms.
Why Meta Blew Up Its Old Targeting Model
So, what happened? In short: a combination of public pressure, privacy changes, and technological evolution. A few watershed moments forced Meta’s hand. First, there was the fallout from the Cambridge Analytica scandal in 2018, which raised alarms about Facebook’s data usage. In response, Facebook began removing many granular interest and behavior targeting options – especially anything that could be deemed sensitive (e.g. targeting by race, religion, political affiliation. Over time, thousands of niche interest segments vanished, and advertisers lost a lot of their former targeting superpowers.
The second (even bigger) hit came from Apple. In 2021, Apple’s iOS 14.5 update let users easily opt out of app tracking, and approximately 90% of iPhone users said “no thanks” to being tracked. For Meta, this was devastating. Overnight, they could no longer see most of the behavior of iOS users off of Facebook, which meant conversion tracking and lookalike audiences took a nosedive. (Up to 90% of your audience on Apple devices became effectively invisible for targeting or retargeting. Custom audience match rates plummeted – typically now only 25–40% of a customer list can be matched to a Facebook profile, versus much higher in the past. In short, the data that fueled Facebook’s targeting machine was greatly diminished.
Faced with these realities, Meta had to pivot its advertising model. The solution? Lean hard into artificial intelligence. If the system had less explicit user data to go on, Meta decided it should model and predict who to show ads to, rather than relying on advertisers to manually choose every detail.
Over the last couple of years, Meta has steadily taken controls away from us advertisers (some of us went kicking and screaming) and given more decision-making to its algorithms. Precise, advertiser-defined targeting is no longer the name of the game – instead, Meta’s algorithms find the right people at the right time based on conversion goals and content, with far less manual input. As one marketing expert put it, “precise, advertiser-controlled audience targeting is no longer a key feature” on the platform, as Meta now relies more on its modeling and machine learning to serve ads.
The New Normal: Advantage+ and Broad Audiences
The clearest example of this shift is Meta Advantage+ campaigns (formerly called broad targeting). This essentially means you let Meta decide who sees your ads – you provide the creative and maybe a broad stroke or two about who you think is relevant, and Meta’s AI does the rest.
If you’re cringing at the idea of ceding that much control, I get it – I was skeptical at first too. But the results often speak for themselves. Meta now openly encourages advertisers to use Advantage+ “broad” targeting, meaning very few filters on the audience. In fact, Meta’s own best practices say that unless you have a very specific niche to hit, you should keep your targeting as open as possible and let the algorithm figure it out. They even recommend audience sizes of 2 million+ people if possible, so the algorithm has a huge data pool to learn from. The philosophy is simple: more data points = better machine learning optimization.
For many PPC media specialists this was hard to wrap their heads around. They used to meticulously build Facebook audiences with layer upon layer of interests and behaviors. Now, some of my best-performing Meta campaigns literally target “Everyone in the US aged 18-65” (with maybe a gender or age filter if it’s obviously relevant). It’s counter-intuitive, but Meta’s machine learning has gotten scarily good at finding the right sub-audience on its own.
One huge misconception, though, is that using broad targeting means your ads show to everyone. In reality, Meta’s algorithms still only deliver the ads to people likely to take the desired action – honing in on the target that we don’t see it working under the hood. It’s like a self-driving car: you tell it the destination (your campaign goal), and it figures out the route and driving maneuvers without you, mostly.
The data backs this up. Advertisers who “let go” and use broad or open targeting have often seen better performance. One 2023 analysis noted that open or broad targeting is now king when scaling campaigns, because feeding Meta’s AI a larger audience gives it more room to optimize and find converters. In fact, keeping targeting broad (millions of users) is explicitly cited as the best way to make machine learning work for you in Meta ads. Meta’s own documentation says something similar: the algorithm can efficiently find the most interested audience if you either leave targeting open beyond basic demographics, or only apply very minimal interests so that your potential reach stays in the millions.
Creative is the New Targeting
So if Meta is doing the heavy lifting on who to show the ad to, what’s left for us advertisers to do? In a word: creative. Targeting and audience selection used to be half the battle on Facebook. Now, much of that has shifted to making scroll-stopping, relevant ads and letting the algorithm match them to the right people.
As growth expert Amanda Berg put it, your creative is now your targeting on Meta; it’s the job of your ad content to attract the right users out of that broad audience pool. The idea is that the algorithm will try showing your ad widely, but the ad’s message and visuals will resonate strongly with certain pockets of people – those who will click or convert – and the algorithm will quickly learn to focus on those folks.
What we creatives have always known is truer than ever: good creative is always the winner.
In practice, this means I’ve been investing more time in understanding my customers and crafting spot-on messages, and less time fiddling with dozens of nuanced audience splits. But I’ve also seen cases where a broad campaign with great creative outperforms a narrowly targeted campaign because the latter was too restrictive and missed lots of potential buyers. Meta’s own algorithm simply has more data and vigilance than a human marketer can, especially after losing some manual targeting levers.
As one Meta guide put it, “Unless your targeting needs to be very restrictive, you should minimize your targeting parameters and let Meta figure most of it out for you. Its ML algorithm is just going to be better than a human’s assumptions at determining the right audience.” Ouch (my ego), but true.
To succeed now on Meta’s PPC, focus should be on a few things: ensuring the Meta Pixel and Conversions API are set up right (feeding the algorithm good data), providing a variety of strong creative assets, and then letting campaigns run long enough for the learning to kick in. It requires some faith and patience – not unlike baking sourdough, where you let the yeast do its thing.
And when results dip, rather than immediately blaming the audience targeting, you’re more likely to tweak the ad creatives or check if maybe my conversion tracking is off. It’s a different mindset, but honestly, once you embrace it, it can simplify your life. Fewer knobs to turn in Ads Manager; more focus on strategy and messaging. (Music to my ears-maybe ad quality will start to improve overall).
Of course, broad targeting isn’t a silver bullet in every scenario. If you’re a very small or local business (say a niche B2B product in a single city), giving Meta complete free rein might waste spend on truly irrelevant users. And the loss of granular targeting has made things harder for those edge cases. Higher-level campaign costs (CPMs) have also risen over time, partly because more advertisers are competing in the same broad auctions. But for the majority of consumer-facing brands I work with, Meta’s new AI-driven approach still delivers – it’s different, not necessarily worse.
Google: Automation Ascendant and Search in Flux
Over on the Google side of the PPC universe, we’re experiencing another shake-up. For years, Google Ads (formerly AdWords) was all about those search keywords – bidding on queries to show text ads on Google’s search results. And that’s still a core piece, but Google has been busy expanding a new AI-driven campaign type known as Performance Max (PMax).
At the same time, the classic search ad itself is undergoing an identity crisis as AI chatbots and direct answers start to change how people use Google. As someone managing clients’ Google Ads, I’ve basically had to rethink where to allocate budgets: do I double down on search, or follow Google’s nudge and pour more into these new automated campaigns? The answer seems to be trending toward the latter. One of today’s largest search engines knocking on Google’s door is ChatGPT.
Meet Performance Max: Google’s AI-All-the-Things Campaign
Let’s start with Performance Max. If you’re unfamiliar, PMax is essentially Google’s “black box” campaign that runs everywhere – Search, Display Network, YouTube, Gmail, Maps, Discover feed – you name it. You give Google a bunch of assets (headlines, images, videos), your goals (e.g. target CPA or ROAS), and it uses its mighty AI to figure out who to show ads to, which channels to use, and how to bid. PMax uses AI to run ads across all Google channels, including Search, Display, YouTube, Gmail, Discover, and Maps, aiming to find customers wherever they are. It’s the Google equivalent of broad targeting on Meta, but arguably even more automated – it’s not just choosing the audience, it’s also deciding which creative to show and on which Google property at any given time
Google introduced Performance Max in 2021/2022 as a way to consolidate and automate campaign types (it replaced things like Smart Shopping and Local campaigns). At first, many PPC pros were wary – giving Google that much control felt like giving the casino dealer their wallet and hoping they return it with interest. But over time, Google has aggressively improved PMax and pushed advertisers to try it.
And guess what? Many of us are seeing excellent results. In fact, in several accounts I’ve seen, PMax campaigns are driving better return on ad spend than the traditional search campaigns. I’ve heard other advertisers echo the same. There’s data to back it up too: for instance, retail and consumer electronics brands saw 19% higher ROAS from Performance Max campaigns in 2023 compared to their other automated campaigns on a major social platform. And Google loves to tout that using the “Power Pair” of Search + PMax together can yield 27% more conversions at similar cost than search alone – basically arguing that PMax captures incremental conversions beyond what search ads get you.
Why is PMax Often Outperforming Regular Search Campaigns?
-
Cheaper Inventory: PMax has access to display and video placements that are much cheaper per click than the high-stakes search ad auctions. According to one analysis, PMax CPCs are roughly 50% lower than search CPCs on average. As Optmyzr’s Navah Hopkins explained, because PMax incorporates visual content (YouTube videos, image ads, etc.), it can garner clicks at a discount compared to pricey search keywords. If you’re struggling with the rising cost of search ads, shifting budget to visually-rich PMax campaigns can deliver solid results for less money – as long as you feed it great creative and track conversions properly. In my experience, this has absolutely held true. Some of our clients’ PMax campaigns find lots of cheap conversions on YouTube or Gmail ads that we might have missed if we only ran search ads.
-
All-Channel Presence: PMax finds users wherever they roam in Google’s ecosystem. These days a customer might see a YouTube video ad, then later search for the brand, then visit the website – a multi-touch journey. PMax can target that user across different touchpoints automatically, whereas a search campaign alone only covers when they actively search. This means PMax can both drive demand and capture demand. It’s like having a full marketing funnel in one campaign.
-
Automation and Reach: Google’s AI is optimizing bids and placements in real-time in PMax. It might decide to show more ads on Tuesday when performance is hot, or allocate more budget to, say, a specific audience segment it identifies as converting well. Humans managing manual campaigns simply can’t react that fast or consider as many signals. Google claims its AI bidding can analyze myriad signals (time of day, device, user behavior patterns, etc.) to maximize results for each auction. The result is often more conversions for the same spend. I’ve personally seen smart bidding (which is built into PMax) boost conversion volume noticeably once it learns – sometimes achieving 10-20% more conversions at similar cost, which aligns with Google’s statements that AI-driven optimization yields ~14% more conversions at the same CPA in Search campaigns.
Now, PMax Isn’t Perfect or Magical
It has some downsides: lack of transparency (you don’t get detailed breakdowns of where your ads showed or which search queries were matched, at least historically – Google has added some reporting due to demand, like showing top search themes now). Also, PMax can sometimes cannibalize your branded search traffic or show ads on oddball placements.
It truly requires good creative assets; if you give it poor ads, it will still run them, just everywhere.
And one must keep an eye on impression share – one study noted PMax campaigns can suffer impression share loss if they overlap with search campaigns for the same terms, meaning neither gets to dominate an auction. In practice, you have to strategically decide when to run a search campaign versus let PMax handle certain keywords to avoid them competing. It’s a bit of a balancing act (and frankly, Google could make this easier to manage).
The Trend Is Clear: Google Is Banking On Automation
So far it’s delivering stable or improved returns for many advertisers. Even during a time (2023-2024) when search CPCs were spiking ~20% year-over-year, PMax campaigns managed to keep ROAS essentially flat and conversion rates stable. It’s not a huge improvement, but impressive given rising costs. In other words, PMax held the fort on performance despite inflation in ad prices, whereas if we’d stuck to just search, we might have seen ROI dip more. Google’s own reps often recommend, “let PMax find easy wins for you while your search campaign focuses on core terms,” and many are coming around to that philosophy.
The AI Impact: Is Search Still the Same?
All these Google Ads changes aren’t happening in a vacuum – they’re partly a response to how user behavior on Google is changing. We have to talk about the elephant in the room: generative AI and the rise of answer engines. In plain terms, more people are starting to skip the traditional search results page and instead ask questions to AI chatbots (like ChatGPT, Bing Chat, or Google’s own Bard).
Even those who use Google Search are now often met with an AI-generated summary at the top of the page (Google’s new Search Generative Experience Gemini) that attempts to answer their query without them clicking any results. As a marketer who relies on people clicking things (ads, ideally!), this trend makes me a bit nervous – and excited – all at once.
Let’s throw some numbers around: Gartner predicts that by 2026, the volume of classic search engine queries will drop by 25% due to users shifting to AI chatbots and virtual assistants for answers. That’s a huge shift in how people find information. We’re already seeing early signs: for the first time in 22 years, Google’s own search volume is no longer growing unabated. In fact, Apple recently confirmed that Google’s search volume declined year-over-year, as users (especially younger ones) turn to AI tools like ChatGPT and others. Google still handles billions of searches, but the monopoly on information retrieval is starting to crack. Google’s global search market share even dipped below 90% for the first time in a decade, which is basically unheard of.
What Does This Mean For PPC?
Well, if fewer people are searching on Google or if those who search aren’t clicking through results (because an AI already gave them the answer), then the traditional search ad has fewer bites at the apple. We’ve been seeing the rise of “zero-click searches” for a while – queries where the user doesn’t click any result because they got the info they needed on the results page. AI answers turbocharge that phenomenon by often providing a full answer right at the top. A concise AI summary can negate the need to click an ad or an organic link. As a result, fewer eyeballs on search ads, fewer clicks – potentially.
Christine Hollinden summed it up starkly: “AI answers reduce ad clicks and revenue. Google’s AI overviews are accelerating clicks away from traditional ads and links”. In other words, the more Google itself answers the user’s question, the less the user needs to scroll and engage with the paid listings or even the organic ones. I’ve noticed this in my own behavior – if I ask Google “What’s the weather in Tampa?” and it just tells me, I don’t end up clicking the weather website or any ads. Multiply that by billions of questions.
Google Isn’t Taking This Lying Down.
They’re in a bit of a paradoxical position: they’re building the very AI that might undermine their ad model, but they’re also finding ways to insert ads into the AI answers. In Google’s experimental Search Generative Experience, you’ll sometimes see sponsored links or shopping ads within the AI summary box. Sundar Pichai (Google’s CEO) even noted that putting content (including ads) inside the AI overview yields higher click-through rates than ads in the old layout. Essentially, Google is redesigning the search page so that the AI answer includes ad opportunities – ensuring they still make money even if the format of results changes. As an advertiser, this means in the future I might be crafting ads or content that get pulled directly into AI answers. The whole notion of “ranking #1” may matter less than “being the chosen snippet the AI reads off.” Wild, right?
In the meantime, Google is also making search campaigns more AI-powered to adapt to these changes. One big push has been Broad Match keywords with smart bidding. Historically, we loved Exact Match for precision and avoided Broad Match because it would match our ads to sometimes crazy irrelevant queries. But guess what: Broad Match ain’t so broad anymore – it’s gotten much smarter, thanks to AI understanding of context and user intent. Google uses machine learning to evaluate the user’s whole search intent (including their past searches) and will only trigger broad keywords when it’s confident the intent matches. The result: broad match keywords now often perform almost as well as exact match in many cases. In Q2 2025, non-brand broad match keywords for Google advertisers had nearly closed the performance gap with exact matches – the ROAS difference was the smallest it’s ever been.
Broad match even delivered a higher average order value in e-commerce campaigns than exact match in one analysis. That was unthinkable a few years ago! Google basically wants us to trust their AI on keywords, just like Meta wants us to trust their AI on audiences. As Google itself noted, “as AI chatbot usage proliferates, optimizing for natural language, long-tail searches is becoming more important, and broad match is a great way to cover those bases”. In other words, broad match helps you catch the weird, voice-assistant type questions that users are throwing at the search engine. And those are on the rise thanks to AI-influenced behavior.
What’s a PPC Marketer to Do?
Whew, we’ve covered a lot – from Meta broad targeting to Google PMax to AI search disruption. As someone who lives and breathes this stuff daily, I can say one thing with certainty: change is the only constant in digital marketing (cheesy but true!). The key is adapting your strategy to these shifts rather than clinging to the old ways. Here’s what I’d suggest to fellow advertisers:
-
Embrace the Automation – Smartly: Both Meta and Google are steering us toward AI-driven automation. Rather than fighting it, test it for yourself. Try Meta’s Advantage+ campaigns or broad targeting if you haven’t, even alongside your usual interest-targeted ad sets. On Google, spin up a Performance Max campaign (or ask your agency to do so) and see how it contributes. You might be pleasantly surprised by the efficiency. Of course, monitor closely. Trust, but verify. Look at the types of conversions coming in, check any available reports to ensure the automation isn’t going off the rails (e.g., burning budget on low-value placements). But overall, don’t ignore these new tools. They’re likely here to stay, and early adopters often reap rewards.
-
Feed the Algorithms Quality Ingredients: If the machines are doing more of the cooking, we need to supply the best ingredients. For Meta, that means killer creative and accurate conversion data. Since you can’t specify the audience as much, your ad has to do the heavy lifting of attracting the right people. Invest in better ad creatives – videos that instantly grab attention, copy that speaks to your customer’s pain points, visuals that stop the scroll. Also, make sure your pixel or Conversions API is tracking the outcomes properly (purchase, sign-up, etc.) so Meta knows what success looks like. On Google, feeding the machine means providing varied ad assets for PMax (headlines, descriptions, images, videos, fill all the asset slots you can) and setting sensible goals (e.g. a realistic target CPA). And ensure your conversion tracking is solid there too. If the AI is optimizing for “purchase” but half your purchases aren’t being tracked due to a glitch, it’s like flying with bad coordinates.
-
Leverage Your First-Party Data (But Expect Diminishing Returns): With all the privacy changes, your own customer data is gold. Upload those email lists and phone numbers as custom audiences in Meta, use them to create lookalikes, and import customer match lists in Google too. They still work – albeit not as magically as before, because match rates are lower post-iOS14 and cookie loss. You might need larger lists than before to be effective (Meta suggests having at least 10k matched users in a custom list to get good lookalike performance. But your buyer list, past lead list, etc., is proprietary info that can guide the algorithms to find more like them.
-
Rethink Metrics and Attribution: As these platforms take more control, it becomes a bit harder to attribute exactly which ad led to what (especially with Google’s PMax lumping everything together). Focus on overall outcome. If your total conversions and ROI are improving with these new campaign types in play, that’s a win, even if you can’t trace every conversion to a single keyword like in the old days. Both Meta and Google now rely a lot on modeled conversions and probabilistic attribution due to signal loss. It’s not perfect, but it’s the new reality.
-
Stay Agile and Keep Testing: The dust hasn’t fully settled. Maybe in a year, we’ll have even more AI in the mix (hello, Google’s next AI search updates, and Meta’s rumored AI chat ad targeting?). The best approach is to stay curious and keep experimenting. Try new beta features. A/B test different strategies (e.g. one campaign with broad targeting vs one with custom targeting). Share notes with fellow marketers, like I’m doing now. And if something stops working, don’t panic, iterate. For example, if your search campaigns are suddenly tanking because an AI answer stole your thunder, maybe shift budget to PMax or try new keyword angles that complement the AI results (like more specific queries or competitor keywords).
-
Optimize for Humans and AI: This is a bit tangential to PPC, but worth noting. As AI answers become more common, consider how your brand can be the answer that AI provides. This crosses into SEO/content strategy (often called “Answer Engine Optimization”). For instance, publishing high-quality content that establishes your expertise might get your brand cited in an AI response to a user’s question – which could indirectly drive them to search for you or trust your ads when they see them. Google’s algorithms (and likely Meta’s in their own sphere) will keep favoring content that is authoritative and trustworthy. So maintain those fundamentals: good content, clear messaging, strong brand presence. It all feeds into the ecosystem that makes your ads more effective when the AI is choosing winners.
Conclusion: Riding the PPC Revolution
The world of pay-per-click advertising is in the midst of a revolution driven by AI and changing user behavior. Meta’s once-famous targeting knobs have faded into the background, replaced by an AI engine that finds your buyers if you give it the freedom (and some great ads to work with). Google’s search-centric approach is expanding into an all-encompassing AI-powered marketing machine, from Performance Max campaigns to AI answers on the search page itself.
As an advertiser, I won’t lie – it’s a bit of an ego check.
The algorithms are doing more of the work we used to take pride in doing manually. But rather than spelling doom for us, I see it as an opportunity to level up our own roles. We get to spend less time on tedious setup (who really loved splitting out 800 ad sets by every tiny interest anyway?) and more time on strategy, creativity, and analysis. That’s always been my secret weapon.
At the end of the day, success in this new PPC landscape still comes down to understanding your customer and offering something valuable to them. That part hasn’t changed. What’s changed are the tools and tactics to connect with that customer. Instead of manually steering every aspect of the campaign, we’re becoming more like orchestrators – setting the right conditions, feeding the right inputs, and letting our algorithmic “team members” perform. And when they perform well, the results can be spectacular (and our clients/bosses are happy!). When they don’t, we troubleshoot and tweak the inputs, or sometimes just give it a bit more time to learn.
Personally, I’m finding this era both challenging and invigorating. I’ve had moments of triumph, like when a broad-targeted Meta campaign significantly beat my carefully targeted one – I was both annoyed (curse you, algorithm, you were right!) and thrilled to see better ROI. But hey, marketing has never been static. This is just the next evolution.
So, here’s to embracing the new PPC. Pour yourself a strong coffee (or maybe a glass of wine), and dive in. Test that new campaign type. Brainstorm that next thumb-stopping ad concept. And maybe chat with your friendly neighborhood AI (they’re here to help… usually). The platforms may change and the algorithms may rise, but creative, adaptable marketers will always find a way to thrive. After all, if an AI can learn, so can we.
Thanks for reading. May your CPCs be low and your ROAS high!