The Measurement Paradox

More data than ever. Less confidence than before. Something's broken in marketing measurement and it's costing real money.

I’ve lost count of the number of marketing conferences where someone confidently declares we’re living in the “golden age of data”. Graphs go up and to the right. Everyone nods. Then you talk to actual marketers afterwards and get a very different story.

New research from EMARKETER and TransUnion—196 US marketers surveyed in July 2025—confirms what many of us suspected. Yes, 62% have some confidence in their metrics. But dig deeper and the picture gets uncomfortable. More than half (54%) report zero change in confidence year-on-year. Another 14% say it’s actually declined.

We should be getting better at this. The tools certainly keep improving. Instead, confidence has flatlined. In some cases, it’s going backwards.

Data everywhere, insight nowhere

The problem isn’t lack of data. It’s that the data doesn’t talk to itself. Nearly half of marketers (49.5%) cite siloed or incomplete data as the main reason they question their measurement. Cross-channel deduplication? A nightmare for 48%. Walled gardens that won’t share their data? 41% struggle with that daily.

Jeremy Rose at Bayer puts it well: “The key to unified measurement is unified data, and that starts with breaking down the walls between systems that were never designed to work together.” He’s not wrong. Your CRM, advertising platforms, and analytics tools all speak different languages. Critical information gets lost in translation, and nobody seems particularly motivated to fix it because—let’s be honest—everyone’s protecting their little empire.

Confidence is lowest where it matters most: influencer marketing (44.4% lack confidence), in-store activity (38.3%), social platforms (35.2%). The channels driving real business? Those are exactly the ones we measure worst. Brilliant.

Trust issues

Here’s where it gets expensive.

Six in ten marketers say their internal stakeholders question their metrics regularly. Another 20% aren’t even sure if non-marketing colleagues trust their reports. And 12% know their reporting is barely trusted at all.

When the CFO doesn’t trust your numbers, you don’t get the budget. Simple as that. Over a quarter of marketers (28.6%) report that between 11% and 20% of their budget has been reallocated or put at risk because of measurement doubts. That’s not a rounding error. That’s real money walking out the door.

The correlation is stark: marketers whose stakeholders trust them see their budgets protected. Everyone else is fighting for scraps. Trust isn’t a nice-to-have. It’s budget protection.

Show me the money (or else)

With budgets under pressure, marketers have responded predictably: by obsessing over ROI. Two-thirds (67.4%) say proving incremental return has become more pressing. It’s now the top measurement challenge, full stop.

Brian Silver at TransUnion frames it starkly: “Marketers are always being asked to do more with less. And even though many marketing organisations are realising the value of metrics beyond just return, incremental ROI is still one of the most important measures of a campaign’s success, especially when budgets are shrinking.”

Nearly 30% face moderate to significant cuts to measurement budgets next year. In that environment, you either prove performance or you lose funding. Brand building, long-term value, customer lifetime metrics—all lovely concepts. But when the axe falls, it’s ROI that saves your neck.

I’m not saying this is ideal, by the way. The short-term ROI obsession creates its own problems—underinvestment in brand, gaming of metrics, channel bias towards bottom-funnel tactics. But I understand why it’s happening. When you’re bleeding budget, you grab for the nearest lifeline.

The AI hype (and reality)

Half of marketers have adopted or plan to adopt AI for automated reporting. The pitch is seductive: automate the boring stuff, surface insights faster, do more with less. And there’s evidence it works—marketers whose stakeholders trust them are more likely to be using AI. It amplifies credibility rather than replacing it.

But let’s not kid ourselves. Research from BearingPoint found that whilst every C-level executive expects AI-driven insights by 2028, only 9% are using AI for performance analytics right now in 2025. That figure might triple by 2028, but we’re still talking about a three-year gap between boardroom promises and actual deployment.

I’ve sat through enough vendor demos to know the pattern. The technology looks amazing in a controlled demo. Then you try to implement it with your messy, real-world data and discover it requires six months of data cleaning, three system integrations, and a data scientist you don’t have on staff. Suddenly that “plug and play” AI solution needs a project team.

Rethinking the stack

Over a quarter of marketers are dissatisfied with their current measurement tech. That’s driving real change in how money gets allocated.

Platform attribution is still most common (65.8% use it), but it’s being supplemented aggressively. Incrementally testing: 52%. Marketing mix modelling: 49.5%. Nearly half plan to increase MMM spending over the next year, whilst 35% will invest more in multitouch attribution.

Silver sees the writing on the wall: “The days of monolithic measurement are over. The most effective measurement strategies are going to feature AI-enabled data collection and data management, which will serve as a foundation for bringing together core methodologies, like MMM, MTA, and incrementally testing.”

Translation: stop looking for one perfect tool. Use multiple methods that cover each other’s blind spots. Platform attribution for near-term optimisation. MMM for the big picture. Incrementally testing for causal proof. It’s more complex, sure. It also produces better answers.

The quarterly trap

Nearly half of marketers adjust their media strategy only once a quarter. Once. A. Quarter.

Consumer behaviour shifts weekly. Competitors move constantly. Your supply chain has problems. The algorithm changed. But you’re still on a quarterly review cycle because that’s what the calendar says and nobody’s questioned whether it makes sense anymore.

Always-on experimentation is the alternative. Incrementality tests, lift studies, geo-holdouts—these let you isolate what’s actually causing results rather than just watching correlations. Each test adds to your evidence library. Over time, you build real knowledge about what works.

Just show your working

Measurement tools are only valuable if people trust them. And too many marketers undermine that trust by presenting results like they’re carved in stone whilst quietly hoping nobody asks about methodology.

Be explicit. Share your data sources, assumptions, limitations. Show how you arrived at the numbers, not just what they say. Rose emphasises this: “One of the biggest mistakes is to have multiple different models saying contradictory things when it comes to performance. The most important thing we can do is to provide a single source of truth.”

Here’s the paradox: acknowledging uncertainty strengthens credibility. Stakeholders know nothing’s perfect. When you pretend otherwise, they get suspicious. When you’re honest about limitations, they’re more likely to trust the numbers.

It’s cultural, not just technical

Silver argues—correctly, in my view—that this isn’t just a technical problem. “Innovation in marketing measurement starts with a strong data foundation,” he says. “But the real transformation happens when organisations pair that foundation with cultural change. That means breaking down silos between marketing, analytics, IT, and finance so measurement isn’t stuck in a corner but embedded into every stage of decision-making.”

One practical approach: establish a measurement council. Cross-departmental, meets regularly, reviews metrics together, aligns on what success looks like. This creates shared ownership. When finance, IT, marketing, and leadership all contribute to defining KPIs, everyone has skin in the game. The numbers become “ours” rather than “marketing’s problem.”

Of course, getting these councils to work requires political capital and patience. The first few meetings will be painful—different departments have different incentives and they’ll all try to game the metrics in their favour. But push through that phase and you eventually get alignment. Or everyone gives up and it becomes another zombie meeting that nobody takes seriously. I’ve seen both outcomes.

What actually needs to happen

The plateau in measurement confidence won’t fix itself. The challenges are real—fragmented data, sceptical stakeholders, shrinking budgets. But they’re solvable if organisations are willing to invest properly.

What’s needed: infrastructure that actually connects systems. Multiple measurement methodologies that balance each other out. Experimental frameworks that provide proof, not just correlation. And cultures that value transparency over spin.

Some specific things that work:

Set aside 5-10% of media budget for experiments each quarter. Small tests compound over time into real knowledge.

Use AI for speed—scenario modelling, anomaly detection—but keep humans in the loop to pressure-test assumptions.

Document your methodology clearly. Every report should come with a brief explaining data sources, assumptions, error margins. Makes you credible and harder to dismiss.

Build that cross-functional measurement council. It’s painful to set up but worth it if you can make it work.

The bottom line

Marketing’s legitimacy depends on measurement. Without it, we’re just making expensive guesses. With it, we’re a disciplined growth engine that earns resources and trust.

The current plateau isn’t inevitable. It’s a reckoning with real challenges. But those challenges are solvable—if organisations invest in infrastructure, embrace multiple methodologies, foster transparency, and build cross-functional alignment.

The organisations that figure this out won’t just defend their budgets. They’ll earn the resources to seize opportunities their competitors miss. They’ll move faster because they know what works. They’ll waste less because they can spot failures quickly.

In an age of data abundance, the scarcest resource isn’t information. It’s the wisdom to interpret it well. That comes from combining perspectives, testing rigorously, acknowledging limitations, and being transparent about what you know and don’t know.

Get that right, and the confidence will follow. The budgets too.

You can download the TransUnion report Here

---

Thanks for reading all the way to the end of the article! This post is public so feel free to share it, and if you have not done so already sign up and become a member.

---

This article is an abridgement. The full long-form version is available to view here https://www.themediastack.co.uk/p/the-measurement-paradox