Banks Own More Data Than Ever. What Makes Marketing Decisions Still So Tough?
By Mike Dawson, Lumin Digital
Simple Subscribe
Subscribe Now!
The data conversation in financial services has circled the same territory for about a decade: dashboards, business intelligence vendors, extract-transform-load pipelines, and warehouse architecture. While the tools have improved, the underlying capability gap at most banking institutions has not, and it is hampering financial marketing.
Unanswered essentials: Every digital banking team should be able to answer two questions:
• Did my campaign work?
• Where are my users getting lost?
But most cannot, and the gap between asking those questions and being able to answer them has less to do with ambition or budget than with architectural decisions made years before the questions even came up.
Every institution runs marketing campaigns through its digital banking platform and email and other outbound channels. Most can tell you how many people received a message, how many opened it, but very few can tell you, with confidence, what those people actually did inside the digital banking experience.
Common unanswered questions:
• Did they sign up for the offered product?
• Did they change their behavior in a measurable way?
• Did the revenue or retention impact justify the marketing investment?
Key marketing breakdown: Without end-to-end measurement that connects campaign activity to actual outcomes within the platform, marketing spend stays disconnected from the business results it should produce.
Need to Know:
- How quickly bank and credit union marketers can move from a passive approach to data to a proactive analysis and application hinges on information technology architecture.
- Institutions need to assess where they stand in three key stages of data use maturity. That status dictates how well they can use data today.
- Moving to a more sophisticated stage may require multiple infrastructure upgrades or a wholesale rethinking of the company’s approach.
The same problem appears in the workflow data. Every digital banking platform has standard journeys: bill pay, transfer setup, account opening, loan application, dispute initiation.
Most institutions have rough completion data — how many users started, how many finished. But very few have the behavioral granularity to know where users dropped off within a workflow.
Was it the first screen? A specific field? A confusing piece of language? Did certain customer or member segments struggle in ways others did not?
Key challenge: Such questions are practical entry points to the bigger strategic issue every financial institution is working through right now: How do we make the digital channel measurably contribute more to the bottom line?
You cannot improve what you cannot measure. Institutions that cannot answer such questions with precision are running their digital banking platforms in partial darkness.
Assessing Your Bank’s Data Maturity
The clearest way to assess where an institution stands on its data journey is a three-stage maturity model. The stages differ in what they enable. The practical implications of each are distinct.
• Stage one: Getting started.
Institutions at this stage primarily work with dashboards and standard reports — out-of-the-box KPIs, auto-populated views, prebuilt reporting templates. The data exists, but it is only consumed in packaged form. The questions an institution can ask are limited by what the dashboard supports.
This is where most institutions are, in the early phases of digital banking maturity, and the dashboard layer is where most decisions are made today.
But the ceiling is real. Standard dashboards cannot easily answer the key questions. That’s because end-to-end measurement and behavioral granularity require data structures that the prebuilt view does not deliver.
• Stage two: Building momentum.
Institutions at this stage are pulling raw data out of the digital banking platform into their own data warehouse or business intelligence tool, joining it with other institutional data and running their own queries. The data they receive needs to be timely, structured for analysis, and comprehensive enough to support the kinds of questions a curious analyst might ask.
This is the stage where campaign attribution and questions about user friction can be answered. That is because the analyst can shape the data to fit the question, rather than fit the question to the dashboard.
However, at this stage constraints remain. Typically this includes data feeds that are not complete. Another issue is latency — the time it takes for data to become available after it is generated or requested.
A once-daily batch of CSV (comma-separated values) data supports some analysis. But a near-real-time feed of structured data supports considerably more.
Institutions at Stage two are doing real analytical work.
• Stage three: Leading the pack.
Institutions here are working with near-real-time data pipelines, direct query access via mechanisms like Delta Sharing, behavior-based personas, and machine learning models that run on their digital banking activity.
The questions these institutions can answer have moved from retrospective to predictive: What is likely to happen, and what should we do about it before it does?
These institutions are starting to operate the way digital-native financial services companies have operated for years, with data as an active strategic input to decisions, not as a quarterly report on them.
Read next: Banks are Sitting on a Data Goldmine and Watching Their Competitors Dig It Up
Why Data Architecture Matters More than Tooling
The conventional framing for advancing data maturity is a tooling decision: “We need a data warehouse, a business intelligence platform, and a machine learning pipeline.”
Those tools are necessary. But the harder constraint is whether the underlying data can flow cleanly enough into the tool to make it useful.
• On a digitally unified banking platform, data flows are coherent by design. The same platform that produces the dashboards at Stage one produces the raw feeds at Stage two and the near-real-time pipelines at Stage three.
Each stage is an incremental expansion of the same foundation. The move from Stage one to Stage two is a configuration upgrade. The move from two to three is an infrastructure expansion.
The institution advances by adding and upgrading capabilities, not by rebuilding old plumbing.
• On an assembled stack — a digital banking environment stitched together from acquired or third-party components — the data flows are fragmented by default.
The signals that drive Stage one dashboards live in different systems than those needed for Stage two warehouse analysis. The behavioral data that would support friction analysis lives in one component; campaign attribution data lives in another; transaction context lives in a third. Reconciling them across integration seams is its own engineering project, separate from any analytical goal.
Each stage of the maturity journey requires a separate integration build. Moving from Stage one to Stage two is a multi-month data engineering effort. In many cases, the bump from Stage two to Stage three is an undertaking on the level of platform replacement.
Key point: That architectural difference determines whether the data maturity journey is a series of incremental investments or major capital projects. It is also why institutions on fragmented stacks tend to stay at Stage one; the cost of advancing is embedded in the platform architecture, invisible — until they try to move.
Read more: The Hidden Cost of Bad Data: Financial Institutions Lose Millions Without Knowing It
What Stage Three Looks Like in Practice
To make the third stage concrete at an institution operating at this level, marketing campaigns are measured end-to-end.
When an offer goes out, the team can see open rates while also noticing downstream behavior within the platform —product signups, revenue impact, the segments that responded, and those that did not.
User friction is measured at the workflow level. When transfer setup completion drops, the institution can identify which step caused the drop, how long users spent there, and which member or customer segments were most affected.
Product teams prioritize fixes against actual behavioral data. Behavior-based personas update in near real time, so marketing and product operate against the current segmentation. Machine learning models surface attrition risk, cross-sell opportunities, and anomalous behavior — feeding decisions about retention spend, product recommendations, and fraud response before the moment passes.
Reality check: Most institutions are not there yet.
• The ones operating on a unified platform are typically one or two infrastructure upgrades away from this level of capability.
• The ones on fragmented stacks are typically a major rebuild cycle away and many do not realize that until they’ve priced it.
Read next: The Hidden Cost of Siloed Data in Financial Services
Assessing Future Technology in Light of Data Maturity
If you are evaluating your digital banking platform or considering a replacement, the data maturity question deserves a place in that conversation.
Yet it is rarely asked in the right way.
Most institutions ask: “What does this platform’s reporting look like?” But the right question is: “Can this platform carry us from where we are today through the next two stages without requiring us to rebuild the foundation each time?”
It can be argued that a platform that supports the full journey from Stage one to Stage three through incremental capability upgrades is, over a five-year horizon, substantially less expensive and more capable than one that requires re-engineering at each transition.
The two questions every digital banking team should be able to answer — did my campaign work, and where are my users getting lost —can be answered today at institutions that made the right architectural call.
At others, they are answerable too, but at a higher cost and on a longer timeline.
Bottom line: The data conversation in financial services has moved from tool selection to platform architecture.
Most institutions have not fully caught up to that yet. The ones that do will spend their energy answering questions rather than building the infrastructure to ask them.
Read next: Tracking Customers’ Spending Patterns Can Drive Better Personalization
