Why martech stacks are getting worse

The 2025 MarTech Replacement Survey revealed a seeming contradiction: organizations are replacing key platforms more than at any time in the past three years, yet their stacks continue to grow.

In the 2025 survey, 59.9% of respondents said they had replaced a marketing technology application in the past year. That’s down from 69.8% at the 2022 peak — a 10 percent drop in three years. CRM turnover has reached its lowest level in the history of the survey. Marketing automation is down from 31.1% in 2024 to 19.4% in 2025. Email platform switching decreased from 24.3% to 13.7%.

Those numbers seem to indicate that change is slow, and slowing down fast.

But the stack data tells a different story. When I broke down respondents who said they had switched platforms, nearly two-thirds said their total number of applications had increased in the past year.

Specifically, 62.9% of switchers added apps to their stack. About 37.9% added one or two tools. Another 21% added three to five. Four percent added six or more.

Only 22.6% saw their stack shrink. About 14.5% remain low.

That intensity – fewer changes, more applications – is the trend in martech right now.

Replacement without reduction

If organizations change several systems, shouldn’t their stacks be stable? Not if substitution and addition are disconnected – and that’s exactly what the data says is happening.

The traditional switching cycle is thought of as alternating: one platform out, one platform in. The modern pattern looks a lot like layering. Organizations keep their core systems in place – CRM, marketing automation, email infrastructure – and add specialized tools at the edges. The categories with the most churn – SEO tools, analytics, marketing automation, and project management platforms – are also where point solutions are expanding rapidly, making it easy to add without subtracting.

That pattern makes sense given the market environment. Replacing an automated marketing platform or CRM involves the cost of migration, retraining, redesigning workflows, and integration work. Adding a point solution for SEO analytics or a lightweight project management tool is relatively inexpensive.

The result is a martech ecosystem shaped less by profit and more by accumulation.

Incorporation tax

The problem with hoarding is that it doesn’t scale well.

Every new application added to the stack creates more integration points, more data silos, and more room to work with complexity. Our survey found that integration capabilities and data integration are among the top selection criteria for replacement platforms – cited by 37.1% and 42.7% of respondents, respectively. Organizations understand the responsibility of integration.

But understanding it and avoiding it are different things. When 62.9% of switchers are still adding more tools than they’re removing, the gap between the integration intent and the stack’s reality seems to be widening.

This creates a kind of operational hazard. The complexity of the stack often converges silently. Organizations add tools in response to specific needs – better SEO performance, granular analytics, a specific AI feature – without a corresponding effort to prune or integrate. Over time, the stack becomes difficult to manage, integrate, and secure.

Decreasing returns may increase this volatility. When organizations become reluctant to change core platforms – due to cost, AI uncertainty, or high switching costs – they are more likely to fill the gaps with additional tools rather than deal with them through platform migration.

The rise of integrative architectures – headless platforms, API-first tools, and the shift to Scott Brinker’s “integrative canvases”, where everything is “close and flexible” – has supercharged this evolution. In a March 2026 research report with Databricks, Brinker argued that martech architecture is moving away from rigid stacks and toward a model where applications and AI agents connect to a universal data layer. “This is not a rip-and-replace proposal,” he wrote. “A three- to five-year vision for architecture.”

That helps explain the survey data. Integration architectures enable additional tools that are almost seamless – bolt on a new analytics platform via API, connect a headless CMS to the end, connect an AI agent – without touching the core. But they don’t make replacements cheap. If anything, they make it harder to argue, as the old “change platform to get new skills” concept has been replaced by “add what you need to what you have”.

What this means in the next section

If this pattern holds, the next few years in martech will be defined less by platform churn and more by stack management.

Vendors that help organizations integrate – platforms that absorb interdependent functionality, integration layers that reduce connectivity complexity, tools that make existing stacks more manageable rather than adding another tile to a mosaic – may be better positioned than pure-play point solutions.

For workers, the message is very different. Substitution has not been difficult to implement – it has become difficult to justify. Cost now dominates selection decisions (cited by 50.8% of switchers). Evaluation cycles are extended – almost two-thirds of replacements spent three months or more being considered, and 35.5% took more than six months. When the marginal benefit from switching platforms diminishes, and the costs of switching remain high, the statistics increasingly favor staying the same.

But endless accumulation creates its own problems. All new tools add integration points, data silos, and workspaces. Organizations that manage this well will treat stack growth as a deliberate choice rather than an automatic outcome.

The survey data does not suggest that martech is slowing down. It suggests that it’s getting messier — it’s slower to replace, faster to add, and harder to counter. That is not a contradiction. It’s the new normal.