After ten years of stacking tools, the uncomfortable verdict is in: the technology is not the bottleneck, the gap between insight and action is. New research from eClerx finds that 78 percent of marketing leaders say their martech investment fails to deliver the ROI they expected. The report names the culprit an activation gap, the distance between generating an insight and being able to act on it. That reframes the whole martech problem, and it explains why the industry’s next sales pitch is already arriving.

What the data says

The eClerx 2026 research puts a number on a feeling every CMO knows. Despite years of spending, most organizations cannot fully leverage the data they already collect. The shortfall is not insight, modern stacks generate plenty, it is activation: the capability to turn an insight into a segment, a campaign, a decision, fast enough to matter. When 78 percent of leaders report the investment underdelivering, the issue is not that they bought the wrong dashboards. It is that the dashboards do not connect to action.

How the gap got built

The activation gap is a direct product of how martech was bought. Stacks grew tool by tool, each solving one problem with its own data store, and the result is a pile of capable systems that do not operate as one. Data sits in a CDP, insight is generated in an analytics tool, and the campaign runs in a third platform, with humans copying conclusions between them. Every handoff is latency, and latency is where ROI leaks. The reflex response, buy another tool, usually widens the gap rather than closing it.

Advertisement

300 × 250

Why agentic AI is being sold as the cure

This is the backdrop to the wave of agentic marketing launches. Kana raised funding explicitly to close the activation gap with agents, MadTech is building an infrastructure layer to let agents act across platforms, and warehouse-native players like Hightouch and GrowthLoop are pushing from data into execution. The pitch is consistent: stop generating insights humans cannot act on, and let agents act on them directly. It is a coherent answer to the eClerx finding, and it is why measurement, data and agents are converging, a shift we traced in the dissolution of the standalone CDP and the arrival of the agentic marketing stack.

The market is reorganizing around activation

The consolidation wave is the same story from the supply side. When an analytics company buys a customer data platform, as Supermetrics did with Relay42, or a warehouse-native vendor launches ad creation, the logic is to collapse the handoffs that create the activation gap by owning more of the path from data to action. The vendors have read the same research the buyers have. The bet across the sector is that the winners will be whoever shortens the distance between an insight and the campaign it should trigger.

Newsletter

Get the week's best tech coverage.

Free. Read by thousands of HR, tech, and business leaders.

The catch nobody should miss

An agent that acts on siloed or wrong data does not close the activation gap. It automates it. The reason insights do not drive action today is rarely that humans are slow, it is that the data is fragmented, definitions disagree across tools, and nobody trusts the number enough to act. Point an autonomous agent at that same mess and you get faster wrong decisions, at scale. The activation gap is a data-and-governance problem wearing a tooling costume, and no agent fixes a broken source of truth.

What it means for the marketing leader

Read the eClerx number as a prompt to audit activation, not acquisition. Three moves. First, stop scoring your stack by how many capabilities it has and start scoring it by how fast an insight becomes an action, that cycle time is the real KPI. Second, fix the source of truth before you add agents, because autonomy amplifies whatever data quality you already have. Third, when a vendor pitches agents to close the activation gap, ask what data the agent acts on and who governs it, the answer tells you whether you are buying a solution or a faster version of the problem. The tools were never really the point. The connection between knowing and doing always was.