The Limits of Tracking in Financial Services
Measurement in financial services is different because the conversion often happens where platforms cannot see it. A prospect clicks a display ad, spends a few weeks comparing options, then walks into a branch and opens an account. The conversion happens inside a core banking system that the ad platform has no visibility into. Customer journeys are longer, highly impacted by macro & micro economic factors beyond media, and brand halo (or drag) from other lines of business is significant.
And it’s not just one platform you’re dealing with. Meta, Google, and programmatic DSPs all run different attribution models with different parameters, so when you try to reconcile performance across channels, you’re not viewing the whole picture. It’s like letting each player’s agent report their stats. Add them up and the team scored more runs than the scoreboard shows.
ROAS may look like a catch-all metric, but it creates blind spots and pushes teams to optimize toward signals that are incomplete and often misleading. In this article, we break down why that is, and what a more defensible measurement framework for financial services actually looks like.
Why a Pixel Fix Won’t Solve This
The instinct is to treat this as a tracking problem: add more pixels, better tagging, cleaner data pipelines. But none of that touches the actual issue. The consideration cycle in financial services is long, the channel path is fragmented, and the conversion quite often happens offline. These aren’t data quality problems you can engineer your way out of. They’re structural features of how financial products get sold.
What a finance-grade framework has to do, instead, is answer the question finance (and any category that converts offline.) actually cares about: did the media cause this outcome, or would it have happened anyway?
What Finance-Grade Measurement Needs to Do
A measurement framework worth defending in an executive review has to do four things: establish a single authoritative source of conversion truth, answer causal questions about media contribution, provide a view over time that supports budget decisions in a repeatable manner, and act as a guide for performance tracking and optimization decisions.
The first requirement sounds obvious but is often the most misunderstood. Mapping a unified, deduplicated conversion relationship, built from authoritative internal sources like a CRM or core banking system rather than pixel fires, is the foundation everything else depends on. Without it, any analysis built on top is measuring noise.
From there, the causal piece is where most measurement frameworks fall shortest, and it’s the second requirement: incrementality. Knowing that conversions went up during a campaign is not the same as knowing the campaign caused them. The question worth asking is what would have happened anyway. No single method answers that for every brand. If you need to separate media from economic conditions or brand halo, geo or matched-market testing is usually the strongest option, provided you have enough regional volume to read a result. If you need to split budget across channels, MMM is the better fit, provided you have two or more years of clean spend and outcome data. Brand lift and panel studies work for upper-funnel questions or low-volume tests, though they may not reach the significance a budget decision needs. Attribution models like MTA still help with in-flight optimization, but they describe the path a customer took rather than what caused the conversion. Every method gives up something, whether speed, precision or cost, and choosing which tradeoffs to accept is a strategist’s job, not a vendor’s.
One of our largest banking clients reports last-touch-attributed account opens as its primary performance number, and the consistency is valuable: every team uses the same metric for channel allocation and creative testing. But last-touch alone can’t say whether the media caused those opens. We run regular lift studies and geo tests to measure what share of those opens are truly incremental, and that ratio adjusts the reported number before it reaches a budget decision.
With the framework and methods chosen, you need a process that gets results to the people who act on them. That means a regular reporting cadence, clear owners for each decision, and agreed optimization triggers, so insights don’t just sit in a dashboard or a PDF. Every report, test and optimization becomes repeatable, and teams of any size stay accountable to the same bottom line. Because everyone acts on the same triggers, teams stop optimizing toward the wrong goals, or worse, competing against each other.
The Metrics That Hold Up in Executive Reviews
Once the framework is in place, the north star metrics shift. Incremental account opens, cost-per-incremental acquisition (not platform-reported CPA), incremental ROAS (not the platform-reported version), and budget efficiency ratios derived from incrementality outputs are the ones that actually travel well across the marketing and finance divide.
The framing that tends to land in executive reviews is simple: can we show that a dollar of media spend generated revenue that would not have happened otherwise, and can we put a dollar value on that? It moves measurement out of the marketing org and into a language finance already speaks. It’s a capital allocation question, and those get taken seriously.
Bottom Line
The right incrementality method depends on the question you’re answering, but any of them beats the platform grading itself. For marketing leaders in financial services facing pressure to justify spend with something more defensible than a dashboard screenshot, the path forward isn’t better attribution. It’s a different measurement philosophy, one that starts from the question finance actually asks: would the money have been better spent somewhere else?


