Advertising’s AI Apocalypse: The Modern Performance Agency
In 2025, the ad industry is in flux. Two high-profile leaders have made stark predictions:
Sam Altman (OpenAI): Roughly 90–95% of what agencies, strategists, and creative professionals do today will be handled by AI—instantly and at minimal cost.
Mark Zuckerberg (Meta): A fully automated ad ecosystem where a business sets its goals, connects a credit card, and Meta’s AI handles creative generation, targeting, messaging, positioning, and landing page optimization—entirely in-platform.
Google has been pursuing a parallel path for well over a decade, embedding automation into bidding, targeting, and creative workflows. The cumulative effect is clear: AI-driven platforms are enabling more brands to manage campaigns in-house, reducing reliance on agencies. AI is moving traffic and changing consumer behavior and the most strident critics portend the death of agencies and perhaps even all advertising.
As a longtime agency guy, I get it. AI is disrupting nearly every operational layer. But disruption is not new; agency roles have been evolving for decades. So what is next and how can we survive?
The Evolving Role of Agencies
The agency imperative to take on new responsibilities is a familiar one. When I started display in 2007 and search in 2009, our daily work was nearly unrecognizable compared to what we do today. We were bidding on CPMs. No one touched creative or offered SEO. Most of my time went into reporting and changing bids. We didn’t send agendas. Brand positioning rarely entered the conversation. There was virtually no discussion about incrementality, the halo effect, or marginal return.
There was a time where automated bidding was seen as an existential threat too. Many believed automated bidding would make agencies obsolete. To be fair, third-party bidding software companies like Marin Software or Kenshoo either went out of business or pivoted completely. But the role of the third-party media buying agency survived through scope of work expansion. Different agency types—branding, creative, go-to-market, digital-only—began to merge capabilities in order to remain relevant and competitive.
AI is now accelerating this shift.
Large vs. Small Agencies
Previously, high-quality creative required specialized teams: a creative team, a data team, a CRM specialist, the SEO team, the social team, the display folks, and client service managers.
Large agencies had the ability to build all of these functions. Smaller agencies used to be forced to choose a specialty and focus on that, but it has become much easier to offer the same level of service without 6 different teams. Similarly, large agencies offered upper-funnel media at discounted rates through cross-client media purchases and that advantage has become diluted as media consumption has evolved from TV to streaming, radio to podcasts, and print to display/social.
In other words, the traditional advantages of larger agencies have eroded.
Platform Incentives and the Need for Independent Oversight
Companies like Meta and Google operate with inherent conflicts of interest: they control both ad delivery and measurement.
Meta’s incrementality measurement has improved with new bidding options, but it still relies on probabilistic models that are not always transparent. Google Ads representatives routinely push products before they are mature and nearly always recommend higher spend.
Agencies provide a counterweight by identifying over-inflated results, separating incremental lift from baseline, and scaling budgets cautiously, with healthy skepticism toward all “helpful” advice from the big media platforms.
Without transparency or experienced marketers defending brands from poor budget decisions, advertisers risk overinvesting in placements that appear strong in aggregate but underperform at the margin.
Google’s Performance Max – Google’s heavy-handed move into campaign automation – symbolizes this dynamic with its many pitfalls for marketers who take it at face value. The core of the model is that it bundles high-quality inventory with lower-quality placements, limiting advertiser control. This can make blended results look fine while hiding low marginal profitability and low incrementality. Advertisers who use Facebook for prospecting, meanwhile, will always aggressively retarget past purchasers if they don’t manually establish guardrails in the form of deliberate exclusions.
Data Strategy as a Competitive Advantage
Even if Google and Facebook do make the move into autonomous automation, there will be ways for marketers to drive differentiated performance and scale, namely in what they choose to optimize toward and how they measure results.
If you optimize for lead volume, you’ll get leads plus a flood of spam. Passing the right quality signals back to algorithms is critical. This includes proxy metrics that correlate with long-term value, not just immediate conversions.
Statistical skills and human oversight are both essential to segment data in business-relevant ways and to leverage those segments effectively in ad campaigns. Agencies that excel here deliver lasting value.
Positioning, Differentiation, and Strategic Interpretation
Brand positioning begins with competitive analysis – understanding your advantage relative to alternatives. Generative AI optimizes for CTR, not differentiation. And platform AIs work with your competitors.
Competitor audits, brand value articulation, and unique positioning still matter. Interpreting CRM data strategically – either to double down on strong segments or to improve weaker ones – requires a blend of marketing strategy and analytics that AI struggles to achieve.
The Human Resource Equation
AI and platform automation make it easier for small in-house teams to run campaigns, but expertise across SEO, paid search, social, and analytics is still rare. Agencies provide access to specialists whose combined experience boosts performance.
For large companies, in-housing may reduce cost for some functions. For small and mid-sized firms, hiring an agile agency is often more efficient – and nearly always more flexible – than staffing every role internally.
Is Account Management Solved?
I do program audits weekly and I still frequently see basic, expensive errors from Fortune 100 companies employing big-name agencies.
A partial list:
Campaigns competing against themselves
Creative that fails to explain the product or audience
Overspending on low-quality inventory (e.g., Google Search Partners)
Majority of spend going to non-relevant queries
Severe targeting mismatches (e.g., a campaign for 25-year-old men spending 90% on women over 60)
I’ll be worried when I don’t see those errors anymore, but until then, I know agencies like mine can deliver value many times greater than our management fees.
Looking Ahead
Assuming automation will eventually take over many of the advertising controls in Google, Facebook, and beyond, the role of agencies will evolve toward:
Product management
Financial modeling (e.g., NPV of marketing investments)
Long-term brand building and competitive advantage maintenance
Testing emerging channels early
Data science and first-party data integration
With the right talent, tools, and mindset, agencies can still thrive.
Here is a list of core cacpities that all agencies should offer in 2025. Let me know if I missed one!
