The AI Imperative: Avinash Kaushik Declares 25-75% Savings on Agency Contracts, Reshaping Marketing’s Future

A seismic shift is upon the marketing industry, driven by the rapid convergence of artificial intelligence, integrated ad platforms, and real-time data ecosystems. This profound transformation, according to digital analytics luminary Avinash Kaushik, necessitates an immediate and radical renegotiation of agency contracts, promising brands potential savings of 25% to a staggering 75% starting this month.

Kaushik, a veteran strategist with 16 years at Google and senior roles at Intuit and DirecTV, now advises as Chief Strategy Officer at Human Made Machine. His pronouncements are not mere speculation but are backed by meticulously constructed models and data-driven arguments, challenging marketers to confront an uncomfortable truth: much of the work agencies currently bill for is now efficiently handled by machines. This paradigm shift, detailed in his latest "Marketing < > Analytics Intersect" newsletter, "Pay Less, Grow More, Agencies in an AI-Era," extends far beyond performance marketing, impacting SEO, content, and even geo-marketing functions.

Main Facts: A New Era of Marketing Efficiency

Avinash Kaushik’s compelling thesis is built upon the simultaneous convergence of three critical forces, creating what he terms a "we are not in Kansas anymore" moment for the entire agency landscape:

1. Broad, General AI Capabilities: Artificial intelligence has transcended its previous limitations of excelling at narrow, specific tasks. Modern AI is now broadly intelligent, capable of understanding context, making inferences, and performing complex operations across diverse data sets.
2. Platform-Integrated Intelligence: Major advertising platforms (like Google, Meta, etc.) are not just leveraging AI; they own the foundational AI models that underpin their ecosystems. This means advanced intelligence is inherently wired into the very tools agencies use daily, from campaign setup to optimization.
3. Real-Time System Interoperability: The contemporary marketing technology stack is no longer a collection of siloed tools. Every system now communicates with every other system in real-time. This interconnectedness doesn’t just make AI smarter in theory; it makes AI specifically smarter about a client’s unique account, data, and objectives.

The practical implication of this convergence is that a significant portion of work traditionally performed by human agency teams, and consequently billed for, is now automated or dramatically streamlined by these intelligent platforms. Kaushik argues that clients should anticipate substantial savings on existing scopes of work. However, he is quick to clarify that this is not a call for smaller agency relationships, but rather a demand to reallocate financial resources towards genuinely new, high-value strategic work that requires uniquely human judgment and creativity, work often overlooked or under-resourced in outdated contracts.

For SEO, content, and geo-marketing teams, the message is equally direct: tasks like keyword research busywork, manual rank tracking, and templated technical audits, which once justified significant portions of monthly retainers, are now largely commoditized by AI-driven crawlers, advanced analytics anomaly detection, and automated monitoring tools. The core challenge for both brands and agencies, therefore, becomes recognizing this shift and restructuring partnerships to reflect the new reality.

Kaushik proposes a new contract structure designed for the AI era, advocating for agency fees to be split into three distinct categories:

1. Lean Base Retainer (40-50%): Covering essential governance, strategic steering, and crucial data engineering functions. This acknowledges the ongoing need for human oversight and foundational infrastructure management.
2. Project Fees (30-40%): Allocated for work that demands genuine human judgment, creative concepting, complex strategic analysis, and portfolio strategy – tasks where AI is a tool, not a replacement.
3. Outcome Incentive (15-25%): Directly tied to verified business results such as incremental profit or demonstrable revenue lift. Critically, Kaushik warns against linking incentives to platform-reported metrics like ROAS, which platforms have an inherent incentive to inflate. This echoes the sentiment for SEO, where incentives should be tied to organic revenue or Citation Share of Voice lift, rather than vanity metrics like keyword rankings or deliverable counts.

Chronology: The Evolution from Manual Labor to Algorithmic Orchestration

To fully grasp the magnitude of Kaushik’s argument, it’s essential to trace the historical trajectory of agency-client relationships and the gradual, then sudden, impact of automation and AI.

The Pre-AI Era (Before ~2020): For decades, agencies thrived on manual execution and human-intensive processes. Campaign setup involved meticulous keyword research, audience segmentation, manual bid adjustments, and extensive campaign architecture. Reporting was a laborious process of data extraction, spreadsheet manipulation, and deck creation, often consuming significant human hours. Agencies justified their retainers based on the volume of these tasks, the perceived "expertise" in navigating complex platforms, and the sheer time investment required. Strategic insights were often gleaned from post-campaign analysis, with optimization cycles being relatively slow and reactive. Human intuition, while valuable, often drove decisions that today’s AI would deem inefficient.

Early Automation and Narrow AI (2020-2023): The first wave of automation brought some efficiencies. Platforms introduced basic smart bidding options, automated reporting dashboards, and rudimentary AI-driven creative testing. Agencies began to see parts of their workload streamlined, but human oversight remained heavy. The "AI" was often narrow, excelling at one specific task (e.g., optimizing for conversions within a defined budget) but lacking broader contextual understanding or the ability to autonomously adapt to complex, dynamic market conditions. Agencies adapted by focusing on managing these tools and interpreting their outputs, still largely billing for the execution and synthesis of data. The idea of "AI learning cycles" was nascent, and human intervention, even if suboptimal, was often seen as necessary to "course-correct."

The Convergent "Now" Moment (Late 2023 – Present): The landscape has fundamentally transformed in recent months. The critical shift, as Kaushik outlines, is the convergence. General AI models like large language models (LLMs) and advanced machine learning are no longer isolated; they are deeply embedded within the core infrastructure of advertising platforms. These platforms can now autonomously handle sophisticated tasks:

• Dynamic Audience Segmentation: AI can identify and target audiences with far greater precision and at a scale impossible for human teams, adapting in real-time to behavioral shifts.
• Intelligent Campaign Architecture: Campaigns can be built and optimized by AI, leveraging vast datasets to predict performance and allocate resources efficiently.
• Autonomous Bidding and Pacing: AI algorithms now consistently outperform human operators in managing bids and pacing, learning from millions of data points and adjusting in milliseconds. Kaushik’s observation that AI has outpaced human pacing decisions since late 2024 underscores this point.
• Real-Time Data Synthesis and Reporting: Interoperability means all systems talk to each other. AI can not only collect data but also analyze it, identify anomalies, generate insights, and even narrate what happened and why, presenting it in accessible dashboards without human translation.

This "now" moment signifies that the core operational mechanics of digital marketing, once the bread and butter of agency retainers, have largely become the domain of intelligent machines. The role of the human has shifted from execution to strategic direction, governance, and the pursuit of entirely new avenues of growth that AI can facilitate but not originate.

Supporting Data and Expert Perspectives: Unpacking the Numbers

Kaushik’s argument is not abstract; it’s grounded in specific, quantifiable reductions in effort and cost for traditional agency tasks. He dissects the old agency scope into clusters, revealing a consistent pattern of AI-driven efficiency.

Account Architecture, Keyword & Audience Structuring, Campaign Build-out: This cluster, historically accounting for roughly a fifth (20%) of a typical contract’s cost, is ripe for significant reduction. Kaushik estimates an astonishing 80% shrink in this area. The rationale is clear: where human teams once painstakingly segmented audiences, structured campaigns for "control," and manually researched keywords, platform algorithms now leverage vast datasets to perform these tasks with superior accuracy and speed. AI can identify nuanced audience clusters, predict optimal keyword performance, and build out campaign structures that adapt dynamically, far exceeding human capacity. The "control" sought by manual segmentation is often counterproductive, hindering AI’s ability to learn and optimize across broader data sets.

Manual Bid and Pacing Adjustments: This core optimization function, once a daily ritual for agency specialists, is another area where AI has unequivocally surpassed human capability. Kaushik states that AI has been outperforming human pacing decisions since late 2024. This is a critical point: an agency that still dedicates significant hours to manual bid adjustments is not just inefficient; it’s actively undermining the system. Kaushik makes a sharper point that deserves widespread attention: an agency attempting to "rescue" a dip during an AI learning cycle is not helping; it is actively sabotaging the algorithm’s ability to learn. Every manual intervention resets the clock, forcing the AI to restart its learning process, leading to suboptimal performance and wasted spend. The true value now lies in understanding when and how to guide the AI, not in overriding its continuous optimization.

Reporting: This often time-consuming and costly activity used to eat up close to a third (33%) of a contract’s cost. Kaushik believes 60% of that can now disappear. Weekly decks, twice-a-week status meetings, and hand-typed commentary on numbers that already reside in a real-time dashboard are vestiges of an outdated model. AI-fronted data tools can now explain "what happened and why" without a human translating a spreadsheet. This doesn’t eliminate the need for reporting entirely; rather, it shifts the agency’s role. Instead of compiling data, their real job becomes deciding what the machine should optimize toward, interpreting high-level strategic implications, and communicating actionable insights that require human context and foresight.

The original article’s author extends these principles directly to SEO and content agencies. Tasks like manual keyword research, often voluminous and time-consuming, are now largely automated by AI tools that can identify semantic relationships, user intent, and competitive landscapes with unprecedented efficiency. Similarly, manual rank tracking has been superseded by automated tools, and templated technical audits can be largely generated by AI-driven crawlers and monitoring systems that identify issues at scale. The "platform already does this" pile for SEO is substantial, encompassing much of what agencies have historically billed as core services.

This shift underscores the crucial distinction between activity-based compensation and outcome-based compensation. As Kaushik warns against paying agencies a "percent of media spend" due to inherent conflicts of interest (agencies incentivized to spend more), the parallel for SEO is clear: paying for deliverables (page counts, audits shipped, blog posts published) rather than for organic revenue actually generated or a verified Citation Share of Voice lift is a relic of the past. A contract built around activity always rewards more activity, not the judgment to do less and achieve more. In an AI-mediated search environment, judgment, strategy, and creative problem-solving are the only things a machine cannot fully replicate.

Official Responses and Industry Reactions: Navigating the Disruption

While direct "official responses" from agencies might not yet be widely publicized regarding Kaushik’s specific call to action, the industry is undoubtedly feeling the tremors of this AI-driven transformation.

Anticipated Agency Pushback: Many traditional agencies, heavily invested in their current operational models and billing structures, may initially resist Kaushik’s recommendations. They might argue that "human touch," nuanced strategic oversight, or bespoke creative ideation cannot be automated. They may emphasize the complexity of client relationships and the need for personalized service, suggesting that AI, while powerful, lacks the emotional intelligence or intuitive leaps of human strategists. Some might also highlight the ongoing need for human expertise in prompt engineering, AI tool selection, and data validation.

The Divide Between Traditional and Forward-Thinking Agencies: This moment will likely accelerate a bifurcation within the agency landscape. Agencies clinging to activity-based billing and manual execution will find themselves increasingly vulnerable to clients demanding greater efficiency and demonstrable ROI. Conversely, forward-thinking agencies will embrace Kaushik’s framework, proactively repositioning themselves as "AI strategists," "data architects," "creative powerhouses," and "growth engineers." They will invest in upskilling their teams in AI governance, advanced analytics interpretation, and complex strategic planning. Their value proposition will shift from "doing" to "guiding," "innovating," and "optimizing the machine."

The Imperative for Agencies to Move Beyond "Selling Hours": The core challenge for agencies is to redefine their value. If a machine can perform tasks faster and more accurately, the agency’s worth cannot be tied to the hours spent on those tasks. Instead, it must be tied to the unique human capabilities that AI cannot replicate:

• Complex Strategic Analysis: Interpreting market shifts, competitive landscapes, and consumer behavior beyond what AI can infer from data.
• Creative Concepting and Storytelling: Generating truly novel ideas, crafting compelling narratives, and building brand equity that resonates emotionally.
• Ethical Oversight and Brand Safety: Ensuring AI use aligns with brand values, regulatory compliance, and ethical considerations.
• Interpersonal Relationships and Client Management: Building trust, navigating organizational complexities, and fostering genuine partnerships.
• Future-Proofing and Innovation: Identifying emerging trends, experimenting with new technologies, and guiding clients into uncharted territories.

Client Responsibility in Data Ownership: A crucial prerequisite for clients to effectively renegotiate contracts is absolute data ownership. As the original article emphasizes, access to GA4, Search Console, log files, and any AI citation tracking tools must reside firmly in the client’s hands, not the agency’s. Without this leverage, clients lack the transparency and control necessary to audit agency performance, validate Kaushik’s claims, and implement a new, outcome-driven compensation model. Agencies that resist full data transparency are signaling a reluctance to adapt to the new reality.

Implications and Future Outlook: Reshaping the Marketing Ecosystem

The implications of Avinash Kaushik’s analysis are far-reaching, promising to fundamentally reshape the relationships between brands and their marketing partners, as well as the very nature of marketing work itself.

For Brands and Marketers: A New Era of Empowerment and Efficiency

Brands stand to gain immensely from this shift. The opportunity for significant cost savings on commoditized work means budgets can be reallocated to higher-impact initiatives, such as deeper market research, innovative product development, truly differentiated creative campaigns, or internal upskilling of marketing teams. This also presents an opportunity for increased ROI, as agency compensation becomes directly tied to verified business outcomes rather than activity metrics.

However, this empowerment comes with responsibility. Brands must invest in internal data literacy and ensure robust data ownership to effectively monitor agency performance and drive strategic decisions. They will need to become more sophisticated buyers of agency services, articulating their needs not in terms of tasks, but in terms of strategic challenges and desired business outcomes. The focus will shift from simply "managing" an agency to forging genuine strategic partnerships with firms capable of delivering unique judgment and innovation in an AI-augmented world. This could lead to a leaner, more agile internal marketing function that collaborates with a network of specialized external partners.

For Agencies: Adapt or Become Obsolete

For marketing agencies, Kaushik’s pronouncements are not an existential threat but a clarion call for urgent, profound adaptation. Those that fail to evolve beyond selling hours and manual execution risk rapid obsolescence. The survivors and thrivers will be those that embrace a new value proposition focused on:

• Strategic Advisory and Governance: Guiding clients through the complexities of AI adoption, data strategy, and ethical implementation.
• Advanced Data Engineering and Analytics: Building bespoke data pipelines, creating predictive models, and extracting proprietary insights from disparate data sources.
• Uniquely Human Creativity and Innovation: Focusing on high-level creative concepting, brand storytelling, and experiential marketing that AI can assist but not originate.
• Complex Problem Solving: Tackling ambiguous business challenges that require human intuition, cross-functional collaboration, and strategic foresight.
• Portfolio Strategy and Experimentation: Helping clients navigate a diverse marketing technology landscape, identify new growth vectors, and conduct rapid, data-driven experiments.

This will necessitate a significant reskilling of agency talent, shifting from tactical execution roles to more analytical, strategic, and creative functions. Agencies must become transparent about their AI capabilities, demonstrate how they leverage automation to achieve efficiency, and clearly articulate the human judgment they bring to the table. The future agency will be a partner in "growth engineering," not merely a vendor of tasks.

The Broader Marketing Ecosystem: A More Intelligent and Accountable Future

The broader marketing ecosystem is poised for a significant transformation. We can anticipate a move towards a more efficient, outcome-driven industry where resources are allocated based on demonstrable impact rather than historical precedent. This may lead to a redefinition of marketing roles, with greater emphasis on data scientists, AI ethicists, strategic communicators, and creative technologists.

The increased reliance on AI also brings ethical considerations to the forefront. Ensuring AI algorithms are unbiased, transparent, and used responsibly will become paramount, requiring collaboration between brands, agencies, platforms, and regulatory bodies. The industry will need to grapple with questions of data privacy, algorithmic fairness, and the potential for AI-driven manipulation.

Ultimately, Avinash Kaushik’s concluding thought encapsulates the choice facing the entire industry: "You can pay for the past, or you can pay for the present." This isn’t merely a financial decision; it’s a strategic imperative. Agencies and brands that continue to write checks for outdated models will find themselves increasingly behind the curve, while those that embrace the AI era’s efficiencies and refocus on uniquely human value will unlock unprecedented growth and innovation. The future of marketing is not about eliminating agencies, but about transforming them into indispensable partners for strategic growth in an intelligently automated world.