There are now more than 15,000 marketing technology tools available to enterprises. In 2011, there were roughly 150.
That represents a 10,000 percent increase in options promising to automate, optimize, and disintermediate work that agencies have performed for decades. And yet global advertising agency revenues are projected to reach $584 billion by 2032. The U.S. digital agency market nearly doubled between 2019 and 2023. The broader professional services sector is on track to approach $3 trillion by 2034.
If software is eating the world, it is not consuming agencies.
This is not a story about agencies failing to adapt or clients behaving irrationally. It is a structural story about how large organizations actually buy, how they manage risk, and why human judgment continues to command a premium even as tools improve. The persistence of agencies alongside an explosion of software is not a paradox. It is a signal that technology has changed the shape of work without eliminating the conditions that make intermediaries valuable.
For more than a decade, analysts have predicted consolidation in marketing technology. Each year, the expectation was that platforms would absorb point solutions, simplify stacks, and reduce the need for external expertise. Each year, the opposite occurred.
The marketing technology ecosystem expanded at roughly a 42 percent compound annual rate for over a decade. In 2024 alone, nearly 28 percent more tools entered the market, the majority of them powered by AI. Approximately three quarters of new entrants relied on generative AI, primarily for content creation and optimization.
What is most telling is not the growth rate but the composition. The long tail of specialized tools has remained stable at roughly half of all marketing technology in use for seven consecutive years. Enterprises continue to buy point solutions not out of novelty seeking, but because their needs are genuinely specific. Horizontal platforms struggle to fully address edge cases tied to industry, regulation, data architecture, or internal process design.
Each new tool introduces integration complexity. Each integration requires configuration, governance, and expertise. Expertise requires people. The proliferation of marketing technology has not reduced the need for human intermediaries. It has increased it by expanding the surface area where things can break.
The standard pitch for marketing platforms is efficiency. Faster execution, lower costs, and measurable return on investment are positioned as self-evident advantages. In controlled environments, these benefits are often real. In large organizations, they collide with a different reality.
The average B2B buying committee now includes ten to eleven stakeholders, up from five or six historically. In large enterprise purchases, that number can exceed twenty. Buying cycles for complex solutions regularly extend to eleven or twelve months. According to Forrester, 86 percent of B2B purchases stall at some point in the process, and 81 percent of buyers report dissatisfaction with the provider they ultimately select.
These dynamics are not caused by poor software. They are caused by organizational complexity. When nearly 80 percent of enterprise purchases require CFO approval, when more than 90 percent of executives demand formal business cases, and when most expect demonstrable ROI within six months of implementation, the buying environment rewards familiarity and perceived safety.
In this context, better and cheaper is rarely decisive. What matters is who can navigate internal politics, align incentives across functions, and sustain confidence when outcomes are uncertain. Agencies have been optimized for this role over decades.
Trust is not an abstract concept in B2B purchasing. It is measurable and it carries economic weight.
Surveys consistently show that the most trusted sources for technology decisions are colleagues and existing vendors, trusted by roughly 79 to 82 percent of buyers. Vendor sales representatives rank significantly lower. Research also shows that 86 percent of enterprise buyers shortlist only products they were already aware of before beginning formal evaluation. Approximately 90 percent of deals are won by vendors in the initial consideration set.
This creates a structural barrier for new entrants, including platforms seeking to displace agencies. Performance metrics matter, but they rarely determine who is evaluated in the first place. Word of mouth remains the dominant discovery mechanism and the most trusted by nearly three quarters of B2B decision makers.
There is also a documented trust premium. Buyers are materially more likely to pay higher prices and to recommend partners they trust. This premium compounds over time, reinforcing incumbent relationships even when alternatives appear objectively superior on paper.
The mid 2010s saw a surge of enthusiasm for in housing marketing capabilities. Programmatic advertising appeared especially susceptible to disintermediation. Several large brands reported early cost savings, sometimes in the tens of millions of dollars, within weeks of bringing operations inside.
The broader trajectory, however, is more instructive. In 2017, 35 percent of marketers reported expanding in house programmatic capabilities, up sharply from the prior year. By 2023, only 32 percent operated programmatic in house, with just 17 percent planning to add those capabilities. The movement plateaued rather than accelerated.
Brands built internal capabilities without eliminating external partners. Complexity increased without a corresponding reduction in dependency. Even among organizations that invested heavily in in housing, 96 percent continued to believe agencies should manage multiple aspects of programmatic operations.
In housing functioned as a capability strategy, not a replacement strategy. It shifted the boundary of work without removing the need for intermediaries who could manage scale, volatility, and accountability.
One of the least discussed functions of agencies is risk absorption. When executives approve marketing strategies, they are not only buying execution. They are buying shared accountability.
If a campaign underperforms, agencies help explain the rationale, contextualize results, and absorb part of the blame. If performance deteriorates materially, leadership has an external partner to challenge, restructure, or ultimately disengage without destabilizing internal teams. This is not cynicism. It is a rational response to incentive structures in large organizations.
Research identifies a substantial segment of relationship oriented buyers who prioritize risk reduction over performance optimization. These buyers exhibit slow switching behavior even when dissatisfied. This pattern, often described as defensive purchasing, favors partners who reduce perceived career risk for decision makers.
Self service platforms concentrate risk internally. Agencies distribute it. As long as executive incentives remain tied to visible outcomes and reputational exposure, this distinction will matter.
The case for technology replacing services weakens further when implementation is examined. Organizations, on average, utilize less than half of their SaaS licenses. Large enterprises waste an estimated $21 million annually on unused software. The typical company manages hundreds of applications, adds several new ones each month, and faces hundreds of renewal decisions every year.
Failure rates are not isolated. Studies suggest that a majority of ERP implementations fail to meet stated objectives on first attempt. CRM systems show similarly wide failure ranges, with poor user adoption cited as the primary cause. Many employees report that new tools are introduced without sufficient training or process alignment.
Integration compounds the problem. Most enterprise applications remain disconnected. IT leaders routinely cite integration as a primary barrier to AI adoption. Each failed implementation reinforces the value of partners who can bridge the gap between purchase and performance. Agencies have built durable businesses precisely in this space.
Public disclosures from major advertising holding companies reveal consistent patterns. Some firms pursued aggressive technology acquisition strategies, growing revenue and headcount simultaneously. Others emphasized productivity improvements, reducing staff while limiting revenue decline.
What did not occur was large scale substitution of people with technology. Investments in AI have been substantial, often reaching hundreds of millions of dollars annually, but they have been framed as augmentation rather than replacement. Agencies expanded analytics capabilities, created new technical roles, and launched proprietary platforms without materially compressing margins.
Net margins remained relatively stable. Technology expanded the scope of services rather than eliminating the need for them.
Technology complexity tends to grow faster than technology simplicity. Each new capability introduces new dependencies, learning curves, and failure modes. In environments where complexity increases, demand for human intermediaries rises rather than falls.
Agencies function less as vendors and more as infrastructure. They manage complexity, absorb risk, and sustain relationships that allow large organizations to act. Software changes how work is done, but it does not eliminate the organizational dynamics that make intermediaries valuable.
For executives evaluating agency relationships, the implication is not that agencies are immune to disruption. Their roles will continue to shift. Hybrid models will expand. Capabilities will move in house selectively. But the structural conditions that sustain agencies are not disappearing.
The tools will continue to improve. The need for people who know how to integrate them into messy organizational reality will persist.