How AI raised the measurement stakes for CMOs
Opinion – Week in Focus
Stop adding AI-powered tools and start strengthening the measurement infrastructure on which AI will sit, argues the CMO of AppsFlyer.
For a decade or more, CMOs have been told that technology would finally solve measurement. First, it was attribution; then, omnichannel dashboards; now, it is AI. But here’s the uncomfortable truth: AI didn’t fix measurement; it simply made untrusted measurement more dangerous.
The pressure on modern CMOs has arguably never been higher. Boards expect growth that is both faster and more efficient, while the need to demonstrate teams are “AI-powered” has become increasingly prevalent.
Yet, most marketing organisations are still operating on measurement systems built for an era where the web was the centre of the customer journey, and signals were fewer and easier to interpret. When the foundation is this shaky, adding AI just accelerates wrong decisions and makes them harder to unwind
Fragmented journeys
Modern customer journeys have fragmented across mobile apps, web, CTV, retail media, offline touchpoints, and various emerging platforms that didn’t even exist five years ago.
Mobile has become the gravitational centre of consumer behaviour, and is where measurement has been most stress-tested by 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 drives outcomes.
CMOs sense this gap. They see it when reports don’t line up with reality, performance shifts but explanations lag, and when teams argue over which numbers are right.
Unfortunately, AI increasingly sits atop that gap.
A false sense of confidence
AI doesn’t reason, it infers. And it infers based on the data it’s given. However, if that data is incomplete, biased toward certain channels, or missing core behavioural signals, AI compounds the issue rather than correcting it.
AI systems can create a false sense of confidence. They produce forecasts, recommendations, and optimisations that feel precise and authoritative.
Dashboards look smarter, decisions feel faster, and outputs feel more sophisticated. But confidence is not accuracy. In practice, that false confidence simply leads to worse decisions faster.
Most conversations about AI in marketing focus on tools, models, and capabilities. But for CMOs, the foundational question is simpler. Is our measurement infrastructure producing data we can trust to drive AI-led decisions? And are we vetting that info?
Trust is about fidelity. So CMOs need to ask themselves whether they can see how customers move across environments, not just within them.
Can they connect exposure, engagement, and outcomes across devices and channels? Can they distinguish real behaviour from modelled guesswork?
The new centre of gravity
One of the most persistent misconceptions in marketing measurement is that omnichannel means treating all channels equally.
However, for most consumers, that centre of gravity is mobile. It is where identity is strongest, engagement is deepest, and intent is most clearly expressed, even when the final transaction happens elsewhere.
Mobile is where discovery, comparison, loyalty, and repeat behaviour increasingly live. It’s also where marketers learned to measure with less. Fewer deterministic identifiers, tighter consent expectations, and constant platform change.
Unfortunately, in too many stacks, mobile measurement is still a retrofit on top of web-era assumptions and reporting conventions that were not built for privacy guardrails. But that choice has consequences.
Without a reliable anchor point that withstands privacy constraints, omnichannel measurement becomes a patchwork of proxies and assumptions.
What should CMOs do now?
CMOs need to sequence AI adoption correctly. That starts with asking harder questions about measurement:
* Where are the biggest blind spots across channels and devices?
* Which decisions rely on modelled assumptions rather than observed behaviour?
* What data do we treat as a “source of truth,” and why?
* Are our systems designed to support automation, or just reporting?
From there, the focus should shift from adding tools to strengthening the measurement infrastructure on which AI will sit.
AI works best when it sits on top of systems built for today’s complexity rather than retrofitted onto frameworks designed for yesterday’s simplicity.
That means prioritising data that reflects real customer behaviour, especially on mobile. It means designing measurements that connect journeys end-to-end, not channel by channel. And it means preparing teams to work with AI as a decision partner rather than merely as a reporting tool.
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.
One path leads to faster decisions that you can defend, the other produces faster decisions you cannot explain, until the quarter is over and the spend is gone.
Ran Avrahamy is the CMO of AppsFlyer