How AI has made bad measurement worse

For a decade or more, CMOs have been told that better technology would finally solve measurement. First, it was attribution. Then it was omnichannel dashboards. Now it’s AI.

But the uncomfortable truth is that AI didn’t fix measurement, it has simply made untrusted measurement more dangerous, because it can create false confidence and lock in the wrong decisions faster.

The pressure on CMOs has never been higher. Boards expect growth that is faster and more efficient. While the need to become “AI-powered” has become increasingly prevalent. CEOs expect marketing to be accountable, predictive, tech-savvy, and resilient.

Yet most marketing organizations are still operating on measurement systems built for an era when the web was the center of the customer journey, channels were fewer, and signals were easier to interpret.

When the foundation is shaky, adding AI doesn’t just accelerate decisions; it accelerates the wrong ones and makes them harder to unwind.

Fragmented customer journeys

What makes it so difficult, of course, is that modern customer journeys are fragmented across mobile apps, web, CTV, retail media, offline touchpoints, and emerging platforms that didn’t exist five years ago.

Mobile has now become the gravitational center of consumer behavior, and it is where measurement has been most stress-tested by privacy changes and signal loss.

Yet many measurement systems still treat mobile as just another channel rather than the connective tissue that links the entire journey. The result is data that appears comprehensive on the surface but is riddled with blind spots beneath. Conversions appear disconnected. Paths seem linear but aren’t.

Performance signals over-index on what’s easiest to measure rather than what is driving outcomes. CMOs are seeing reports that don’t line up with reality. This needs to change.

Confidence is not accuracy

AI systems are remarkably good at creating confidence. They produce forecasts, recommendations, and optimizations that can feel precise and authoritative on the surface. Dashboards can look smarter. Outputs can feel sophisticated. But confidence is not accuracy. In practice, false confidence just lead to worse decisions faster.

The problem is that when key signals are missing, AI tools will fill in the gaps with assumptions. Those assumptions get reinforced over time. Budgets shift, and strategies get locked in. Teams trust the outputs because they look advanced, even when they’re grounded in only partial truth.

In other words, AI can give marketing leaders a false sense of certainty at the exact moment they need clarity most. Once teams operationalize those outputs, the feedback loop can become self-reinforcing, making it harder and more expensive to unwind errors.

Most conversations about AI in marketing focus on tools, models, and capabilities. But the foundational question should be is our measurement infrastructure producing data we can trust? Trust is about fidelity. But it can be difficult to see how customers move across environments.

This is why so many early AI initiatives didn’t meet expectations. The technology didn’t fail, but the underlying measurement infrastructure was never designed for autonomous or semi-autonomous decision-making.

Measurement is not a supporting metric. It’s the foundational infrastructure that determines whether AI becomes an accelerator of a liability.

Mobile at the core

One of the most persistent misconceptions in marketing measurement is that omnichannel means treating all channels equally. In practice, this means understanding how they connect and where behavior occurs. For most consumers that center of gravity is mobile.

Mobile is where identity is strongest, engagement is deepest, and intent is most clearly expressed, even when the final transaction happens elsewhere. It’s where discovery, comparison, loyalty, and repeat behavior increasingly live.

It’s also where marketers learned to measure with fewer deterministic identifiers, tighter consent expectations, and constant platform change.

In too many stacks, mobile measurement is still based upon web-era assumptions and reporting conventions adapted to apps, rather than mobile-grade standards built for privacy constraints.

Yet without a reliable anchor point that holds up under privacy constraints, omnichannel measurement becomes a patchwork of proxies and assumptions.

When mobile is treated as an afterthought, teams end up optimizing to what their platforms can most easily observe, not what customers do.

A fork in the road

CMOs need to start asking harder questions about measurement: Where are our biggest blind spots across channels and devices? Which decisions rely on modelled assumptions rather than observed behavior? What data do we treat as a “source of truth”? And are our systems designed to support automation?

From there, the focus should shift to strengthening the measurement infrastructure on which AI sits. AI works best when it is atop systems built for today’s complexity rather than retrofitted onto frameworks designed for yesterday’s simplicity.

That means designing measurements that connect journeys end-to-end, not channel by channel.

In a world where AI increasingly shapes decisions, measurement becomes the control layer to know what to trust, what to question, and when to intervene. As decision-making becomes more autonomous, the cost of getting that wrong only rises.

CMOs face a fork in the road. Treat measurement as the foundation for AI-driven marketing, anchored in mobile-grade standards that hold up under privacy pressure. Or keep stitching together channel reports and feed AI a partial view of reality. I know which I would choose.

We've featured the best customer experience (CX) tool.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro