Marketing's Blind Spot: Why CMOs Rank ROI Measurement Last
Measurement came last in what CMOs said they would fund. The problem is not that the numbers are wrong. It is that they cover the wrong things.
Stat of the Week. Measurement came last. In BCG’s 2025 global survey of 200 chief marketing officers (CMOs), measurement ranked at the bottom of the funding priorities for improving the digital customer experience and the marketing technology stack that enables it. (BCG, June 2025.)
Measurement is unique in marketing because it can feel finished even as it becomes outdated. When this happens, everything seems fine. Reports keep coming in on schedule, the numbers look right, but each year, what gets measured quietly narrows and few people notice.
A BCG survey of 200 CMOs found that measurement was the lowest priority for investment to improve digital customer experience and technology (BCG, June 2025). It’s easy to see why. Measurement often looks complete, while other areas seem more urgent or unfinished. This thinking makes sense until you realize what happens when your measurement system slowly stops working as well as it should.
It’s easy to put measurement at the bottom of the list, and there’s a reason for that.
BCG points this out too. They say measurement is a lower priority partly because many companies have already invested heavily in first-party data and measurement, giving them a solid base (BCG, June 2025). After such big upgrades, it’s natural to shift new spending elsewhere.
But BCG also warns about the risk. They say putting measurement last 'may also indicate a blind spot, because considerable untapped potential remains,' and that the details are often dull and expensive (BCG, June 2025). For real marketing ROI, BCG says many companies still don’t invest enough in three things: connecting marketing to extra sales, agreeing on how to measure upper-funnel and nontraditional spending, and using automatic measurement to improve personalized content during campaigns.
None of these three problems are about accuracy, so better models or cleaner data won’t fix them. They’re really about what your measurement system covers, which is harder to notice. A system can be very precise about what it tracks, but if it’s tracking the wrong things, it still seems complete.
An average customer isn’t the same as a real customer.
A good example of this issue is how averages are used. Most marketing measurement depends on blended numbers, like cost per acquisition for the whole program or lifetime value for all customers, because these are the metrics leaders usually accept.
Bain’s research with B2B companies found that promoter customers have an average lifetime value three to twelve times higher than detractors, depending on the segment and industry (Bain, 2014). The main takeaway is the big gap between promoters and detractors, not their average. The difference between your best and worst customers is a multiple, not just a small percentage, and it changes so much by segment that using a single average hides important insights.
When you compare that big range in customer value to a blended cost per acquisition, the numbers stop being reassuring. If new customers can be worth so much more or less, it’s easy for a campaign to look good on paper but bring in the wrong people. This mistake happens often. For example, one industrial distributor ran a campaign that hit all its targets: cost per new account fell by 20% and volume went up. But no one checked who these new customers were. They bought once, cared only about price, and rarely came back. The same team had to serve both these new and long-term accounts. The older accounts brought in most referrals, but referrals dropped the next year and didn’t show up in any report, since there was no metric for them. By the measures used, the campaign still looked like a win.
Programs can look successful even as relationships with customers get weaker.
This problem isn’t just found in industrial distribution. It also shows up in consumer loyalty, where metrics focus on the program and not the real relationship. EY’s 2026 loyalty study makes this clear. Loyalty programs look good by standard numbers, but it’s not clear customers feel the same value (EY, 2026). Enrollment and ROI seem strong, but people join fewer programs, check rewards less, and feel less positive, even when the program works as planned. At the same time, churn is going up.
A program can look like it’s doing well and still be declining, because measurement only tracks the program itself. Enrollment counts sign-ups and ROI counts redemptions, but neither tells you if the real relationship is getting weaker. This is the same issue the distributor had, just in a different industry.
The good news is you don’t need a new platform to fix this. The three areas BCG talks about are about what you analyze, and the loyalty gap EY describes can be found in data most companies already have. What’s needed is someone willing to admit that measurement isn’t finished.
Here are four questions you should try to answer, even if you can’t right now.
You don’t need a big project to tackle these. The real benefit is finding out which ones you can’t answer yet.
Can you tell what your last campaign really added, apart from what would have happened anyway? If you can’t separate extra sales from normal sales, your ROI number is questionable at best.
Do you know how much your best and worst customers are worth, for each segment? A single average can’t show if you’re gaining or losing with new customers.
Which of your upper-funnel or nontraditional spending doesn’t have an agreed way to measure it? Not everything that’s unmeasured is unmeasurable, and the list is usually shorter than you think.
What could you change during a campaign, using measurement that updates automatically? This is where you’ll see the fastest results, and BCG’s approach assumes you have measurement that updates by itself.
Measurement often gets less funding because it never looks broken. If you treat it as something you need to maintain, your next budget talk will be based on real evidence, not just gut feeling.
Sources
BCG, “How CMOs Are Scaling GenAI in Turbulent Times,” June 2025 (published June 2, 2025; annual global survey, n=200 CMOs, fielded April and May 2025; “measurement ranked last” is a ranking of stated funding priorities, and the “blind spot” reading is BCG’s own).
Bain & Company, “Do Your B2B Customers Promote Your Business?,” February 2014 (the 3–12x promoter-to-detractor lifetime-value range is Bain’s observed range across client work, not a survey; varies by segment and industry; vintage dated in-article).
EY, “2026 EY Loyalty Study: the gap between performance and experience,” 2026 (third edition; cited directionally — the study’s published summary carries no isolated headline percentage, so no figure is quoted).