Built on Borrowed Time - Part I: Paid Media in Marketing Strategies That Don't Break Every Time a Platform Changes
Updated: Aug 14
AI keeps changing the rules and many marketing teams are playing catch up. What should we actually be paying attention to?
The Disruption That Already Happened
The restructuring arrived quietly and then all at once. What shifted in the process was how content gets produced, how platforms allocate reach, how buyers find and evaluate information, and what it costs to compete for attention.
Most marketing strategies running today were built for a different environment, and the teams feeling this the most are often the ones who moved fastest by generating more content and running more campaigns. What most burnt-out marketing teams agree on is that despite all the tools and new techniques, pipelines seem more complex and harder to build, attribution is harder to defend, and the returns on spend grow more difficult to explain each quarter.
The tools were never the problem. What changed was the underlying economics of every channel, and whether the marketing mix was balanced enough to absorb it.
There is also a new pressure point emerging on top of all of this. Agentic AI systems, capable of doing autonomous research, comparing options, and completing transactions on a user's behalf, are inserting themselves between businesses and their buyers. Nearly three-quarters of users are already involving AI somewhere in their purchase journey, and 70% say they are at least somewhat comfortable letting an AI agent make decisions for them. The funnel most teams are still optimizing for was designed around a buyer who no longer behaves the same way.
Paid, owned, and earned remain the right frameworks for understanding how AI has changed the risk profile and the strategic importance of each pillar.
The Diagnostic: Your Channel Mix Is Now a Risk Map
Before looking at each pillar individually, it's worth identifying the structural issue most teams are facing.
For many organizations, marketing budgets lean heavily toward paid. Paid media, used well, is genuinely powerful. It can amplify what's already gaining traction and produce results that owned and earned channels take months or years to match. For marketing teams, paid is also easier to justify. Results show up fast and measurement is straightforward. When you can point to last-click revenue and a ROAS figure, you'll have a much easier conversation in a budget meeting than talking about compounding organic value over 18 months.
AI, however, is changing the playing field fast. Paid is becoming harder to read. Platforms optimize differently now, attribution is harder to explain, and the transparency that made paid feel so controllable has eroded. Owned channels carry more weight because first-party data is becoming a more significant input for AI-driven personalization, and most organizations are still early in building that capability. Earned is shifting too, as AI-powered search and agentic discovery start to change how people find and trust information.
If your team has been over-relying on paid for immediate results and underinvesting in the other two, you are probably among those feeling the current disruption most intensely.
What follows is a picture of where things are likely to land. Mapping your channel mix helps you understand where your pipeline is most fragile, and where the next algorithm shift or platform change impacts you the most.
Paid: More Powerful, Less Yours
The black box problem in paid media is not new. The industry has just been slow to name it clearly.
Meta Advantage+, Google Performance Max, and their equivalents represent a genuine change in how paid advertising operates. The targeting decisions, placement choices, audience expansion logic, and creative selection that marketers once controlled directly have been absorbed into AI systems optimizing toward platform-defined outcomes. AI-driven campaigns frequently outperform manually managed ones on the headline numbers.
The problem here is what you give up in exchange.
Most experienced marketers who have spent their careers in hands-on campaign management find themselves in a position of both relief and discomfort. Trading manual control for stronger performance comes at a real cost. The insights that used to come from making decisions, the ability to question why something worked, the feedback loops that built up over months and told you something real about your audience or your messaging are now being handled by systems that don't explain themselves. For senior marketers who built their expertise managing campaigns end-to-end, what remains is a result and a growing sense of uncertainty about what produced it.
Paid media has always been exposed to changes in spending patterns, and for performance marketers, the basis of the job has always been improving results by gaining knowledge through testing. What AI-optimized platforms have added is a new kind of vulnerability. Platforms change their optimization logic continuously and without warning. Advertisers find out when performance shifts unexpectedly and are prompted to adapt instantly.
The deeper issue here is what this does to organizational learning. Teams running Performance Max at scale often know something is working but cannot explain how, or how to replicate it for another client, or how to rebuild it if an algorithm update collapses performance overnight. In simple terms, we are more and more at the mercy of the platforms.
Keep in mind that Google and Meta are businesses optimizing for their own margins. They control the auction, the algorithm, and the reporting. Their incentives and yours are related, but they are not always the same.
The New Age of Google Search
Paid search has always been evolving, but the impact of AI Overviews on click-through rates is on another level. Seer Interactive tracked over 25 million impressions across 42 organizations between mid-2024 and September 2025. Paid CTR on queries where an AI Overview appeared fell from 19.7% down to 6.34%. In July 2025 alone, it dropped from roughly 11% to 3% in a single month as Google expanded AI Overview coverage into commercial queries. Search spend grew 9% year-over-year in Q1 2025, while click growth came in at 4%. More budget, fewer clicks, harder math to defend.
For most businesses, the rearrangement of the search results page has been extremely disruptive and in many cases has had a direct impact on revenue. Some have responded by increasing Google Ads budgets, hoping to recover the visibility they lost. What they found is that paid search is also a changing landscape. Manual control still exists, but AI-driven campaign types like Performance Max consistently outperform it on headline numbers. For anyone thinking critically about this from a business perspective, a question worth sitting with is this: if a vendor is actively pushing everyone toward a product where they control both the optimization logic and the pricing, what happens when the economics shift in their favour?
Google's 2025 transparency updates partly addressed the visibility problem. Channel-level reporting, full search term visibility, and asset-level insights all arrived. What they introduced was a more complex version of the same underlying problem. Marketers receive more data than before, but the expertise required to act on it meaningfully is significant, and the levers to redirect budget, including time, based on what that data shows are limited. For most teams, the strategic result has stayed the same.
Meta's Andromeda: When Creative Becomes the Targeting
Meta's Andromeda ad retrieval system has quietly changed the underlying logic of how paid social works. Previously, audience targeting was the primary lever. You defined who you wanted to reach, and creative served that audience. Andromeda reverses that relationship. Creative quality has become the dominant performance driver, with the system using creative signals to determine who sees the ad, rather than the reverse.
The consequence is significant for teams that still build paid social programs audience-first. Audience segmentation and persona work based on research remain extremely relevant (needs to be discussed in another post), but they are increasingly inputs to creative development and overall strategy rather than campaign controls. If the creative is weak, audience selection will not compensate for it. In this new system, the teams adapting fastest are treating every creative asset as a performance hypothesis and building testing volume as a core capability.
Meta is not the only social platform using AI and machine learning in advertising. The pattern also mirrors what we are seeing in Google Search. Meta is also pulling execution and optimization further inside its own systems, reducing the surface area where marketers have meaningful control. There have also been reported cases of unexpected AI-powered modifications to creative assets, including background changes, format shifts, and static images converted to video without explicit approval. Meta has acknowledged this and has stated that advertisers have the ability to opt out of AI creative testing at any time, and that Meta runs these tests with a small share of ad impressions to help improve performance.
The same structural risks apply: sudden shifts in algorithm logic, pricing changes, and platform-level decisions that advertisers have no visibility into until they show up in performance data.
None of this is an argument for walking away from paid media. It remains the fastest way to reach a specific audience at a specific moment. The argument is for treating it as what it actually is: a distribution mechanism, not a growth engine. It amplifies what is already working. It does not build the things that make it work. Positioning, message clarity, conversion architecture, and audience trust. Those existed before AI and they need to come from somewhere else.
If your paid investment is not building anything durable, not an audience you can retain, not data you own, not a brand that earns organic return, then the moment you stop spending, you are back to zero. That is dependency, not strategy.
Paid media's fragility doesn't exist in isolation. The reason platform dependency hurts as much as it does is that most organizations haven't built the infrastructure that would absorb the shock. The direct audiences. The first-party data. The earned credibility that grows on its own terms, independent of what any platform decides to do next. Part two of this series looks at owned and earned media through the same lens. Not as secondary channels, but as the foundation that determines whether your marketing holds when the ground shifts underneath it.
Comments