Most marketing teams spend six and seven figures on campaigns they can’t actually measure, more commonly than any CMO wants to admit. Marketing budgets have flatlined at 7.7% of company revenue, per a 2025 Gartner survey, with 59% of marketing leaders saying they lack the budget to execute their strategy, according to a 2025 CMO study. Budgets are shrinking while expectations climb. And yet, a staggering portion of that spend goes toward technology and campaigns whose actual return nobody can pin down with confidence.
The Attribution Mirage
Marketing attribution models are supposed to answer the most fundamental question in the discipline: which dollar made us money and which one didn’t? Most of the time, they answer a different question entirely: which touchpoint happened to be closest to the conversion event?
Last-click attribution, which still underpins much of how companies evaluate channel performance, ignores every interaction that builds awareness and trust before the final click. Multi-touch models try to distribute credit more fairly, but as analysts have argued, the subjective choices embedded in how those models are scored and weighted introduce compounding bias. If you asked ten senior marketers at the same company to build attribution models for an identical customer journey, you’d get ten meaningfully different answers.
The deeper issue is that correlation keeps getting mistaken for causation. An ad platform reports strong return on ad spend, but research from this study from lifesight.io points out that platforms like Google and Meta have a vested interest in demonstrating their own effectiveness, and their reported figures often inflate true impact. The numbers look precise, but that precision is an illusion. When I moved from luxury retail to pharma to SaaS, the same pattern repeated across all three industries: teams making confident budget allocation decisions on data that wouldn’t survive even mild scrutiny.
The CAC Blind Spot
Customer acquisition cost is the metric most companies cite when they want to prove marketing efficiency. It is also the metric most companies calculate wrong. The standard formula is deceptively simple: total marketing and sales spend divided by new customers acquired. But most teams exclude indirect costs like tool subscriptions, onboarding expenses, and the salaries of people who support acquisition without formally sitting on the marketing team. This kind of omission can understate true acquisition costs by 40–60%, according to this study from Marketer.com.
Then there is the timing mismatch. In B2B, especially at companies like CRM Messaging where I work in growth marketing and corporate development, the sales cycle can stretch across months. A lead generated from a campaign in March may not convert until June. Dividing March’s spend by March’s conversions produces a number that is directionally misleading and operationally useless.
For CMOs, this means CAC is often treated as a precise metric when it is, in reality, structurally flawed without full cost inclusion and proper time alignment.
The Space Between Data Science and Marketing
One reason these problems persist is organizational. In most companies, the analytics team and the marketing team operate as separate fiefdoms. Data scientists build sophisticated models and hand off dashboards. Marketers glance at the dashboards, extract the number that supports the decision they already wanted to make, and move on. The feedback loop between model output and marketing action barely exists.
I’ve seen this from both sides. My background is in computer engineering and business analytics, but I’ve spent my career in marketing and business development roles. That cross-functional perspective has made one thing painfully clear: the gap is rarely about capability. Both teams are usually competent. The breakdown happens in translation. Data scientists optimize for model accuracy; marketers optimize for campaign speed. Nobody optimizes for the handoff.
Bridging that gap starts with shared ownership of metrics. When I’ve worked on growth marketing and capital raise projects simultaneously, the discipline of finance has been instructive. Investors don’t accept vague proxy metrics. They want to see unit economics, cohort analysis, and retention curves. If marketing teams held themselves to the same standard that investor decks demand, the quality of measurement would improve overnight.
Customer Lifetime Value That Actually Predicts Something
Customer lifetime value should be the metric that connects acquisition spend to long-term revenue. In practice, most CLV calculations are backward-looking snapshots that flatten the complexity of customer behavior into a single number. The traditional formula of average order value multiplied by purchase frequency multiplied by customer lifespan produces a figure that looks useful and collapses under pressure. While 81% of organizations track CLV, only 37% use those insights to drive strategic decisions, per a 2025 customer analytics survey. The remaining organizations collect the data and then ignore it.
Predictive CLV models are more promising but require genuine investment in data infrastructure. Probabilistic approaches like BG/NBD models, which estimate the likelihood that a customer is still active and forecast their transaction frequency, can work with minimal data. Machine learning approaches can incorporate richer signals like acquisition channel, engagement patterns, and product category behavior. Both approaches demand clean, well-integrated data, and that is where most organizations stall.
At Just Wines in Australia, I saw firsthand what happens when you actually tie CLV to operational decisions. We increased average order value by 70% and scaled a dormant eBay channel from AU$16,000 to AU$150,000 in monthly revenue. Those results didn’t come from a single insight or a clever campaign. They came from building the analytics infrastructure to understand which customers were worth acquiring, what made them stay, and where the economics broke down.
The shift toward analytics didn’t start with sophisticated modeling. It started with cleaning up fragmented data. Order history lived in one system, campaign data in another, and marketplace performance data sat in spreadsheets maintained manually. We focused first on integration: consolidating customer purchase data, channel attribution signals, and pricing information into a single reporting layer that the commercial team could actually use.
The tools themselves weren’t exotic. We relied on SQL-based reporting, BI dashboards, and structured cohort tracking rather than complex machine-learning models. The breakthrough came from discipline, not technology. Once we could see repeat purchase behavior clearly, we began segmenting customers by acquisition source, purchase frequency, and price sensitivity. That allowed us to shift promotions away from blanket discounts toward targeted bundles and pricing tiers that increased order value without eroding margins.
There was resistance at first. Sales teams worried that stricter measurement would slow campaigns down. Marketing teams feared analytics would limit creativity. But once the data started surfacing obvious inefficiencies, adoption became much easier.
For CMOs, this means CLV should evolve from a static reporting metric into a predictive tool that directly informs acquisition, retention, and pricing strategy.
Building an Analytics Infrastructure That Earns Trust
The technology for better marketing measurement exists. The problem is adoption, integration, and organizational will. Gartner’s 2025 Marketing Technology Survey reports that only 49% of martech tools are actively used, and just 15% of organizations qualify as high performers in terms of meeting strategic goals and demonstrating positive ROI from their tech stack.
Fixing this requires less technology and more discipline. Three things I’ve seen work across different industries and company sizes:
First, agree on definitions. If the marketing team and the finance team cannot agree on what counts as a “new customer” or how to calculate CAC, no amount of tooling will produce trustworthy numbers. I’ve worked in organizations where marketing and finance used different time windows for the same metric and arrived at acquisition costs that differed by 30%. The argument over whose number was right consumed more energy than fixing the underlying campaigns would have.
Second, invest in the plumbing before the dashboard. The analytics platforms and visualization tools are only as useful as the data pipelines feeding them. If your ad platform data, CRM, and billing system don’t talk to each other, the reports you produce will always contain gaps.
Third, measure at the cohort level. Aggregate metrics obscure the story. Segment your acquisition costs by channel, customer type, time period, and product. When I led growth marketing for a B2B messaging platform, the aggregate numbers looked healthy. The segmented view revealed that one acquisition channel was subsidizing two others that were quietly hemorrhaging money. We cut one channel entirely and reallocated the budget. Pipeline quality improved within a quarter.
While I was leading growth marketing at CRM Messaging, segmentation showed that a high-volume acquisition channel (third-party aggregator listings) was generating leads at nearly twice the blended CAC while converting into qualified pipeline at less than half the rate of direct search and referral channels. On the surface, the channel looked productive because it produced a steady flow of leads. At the cohort level, it was quietly undermining efficiency.
We reduced spend on that channel by roughly 60% over a quarter and redirected the budget into intent-driven search campaigns and partner co-marketing programs that historically produced smaller volumes but stronger conversion rates. Within two quarters, marketing-sourced pipeline value per dollar spent improved by roughly 20–25%, and the sales team reported noticeably higher lead quality.
The lesson wasn’t that one channel is universally bad. It was that aggregate reporting had been masking a structural inefficiency for months. Once we looked at performance by cohort rather than in aggregate, the decision became obvious.
Marketing budgets are not going to get bigger anytime soon. The companies that will outperform are the ones that stop treating analytics as a reporting function and start treating it as the core operating system of their marketing. That means fewer vanity metrics, fewer tools bought on promises, and more honest engagement with the numbers. The data to make better decisions is already sitting in most organizations’ systems, but they must act on what it actually says, instead of what they hope it confirms.