If Your Data Quality Is Low, Your AI Strategy Will Blow (Or You Might Be a Data Redneck)

Ron Shevlin

February 27, 2026

Bankers want to discuss AI strategy. They’re excited about chatbots, machine learning models, generative AI tools, and AI agents that promise to revolutionize everything from customer service to underwriting. What they don’t want to talk about, however, is their data mess.

The Inconvenient Truth About AI Strategy

AI doesn’t create insights from thin air. It amplifies what you already have, including your weaknesses. Feed it messy data, and you get messy results. Feed it incomplete data, and you'll get incomplete insights. Feed it data from siloed systems that don't talk to each other, and you'll get AI that can't see the full picture.

But having good data doesn't matter if you're not actually using it effectively. According to a study from Cornerstone Advisors, most banks are nowhere near where they need to be. We evaluated banks’ data quality across five functional areas: 

Credit analysis. Tied for highest-scoring function, even the quality of data use for credit analysis has significant gaps, as most banks struggle with incorporating alternative data sources and tracking loan performance.
Data access and analysis. Data quality and integrity monitoring are shortcoming, as is the integration of structured and unstructured data.
Operations. Fraud detection and loss prevention analytics were strong points but root cause analysis of service issues and predictive maintenance for IT infrastructure are improvement areas.
Sales and marketing. This was the lowest scoring function, with two-thirds of institutions falling below 50 points out of 100, meaning their data usage was below even the “developing and/or improving” level.
Strategic planning. Many bankers believe that their board members have achieved a strong level of data fluency and literacy, but the use of predictive analytics in strategic planning is seriously lacking.

Only a quarter of the financial institutions participating in the study qualified as “high performers.” What it means: Three-quarters of financial institutions are building AI strategies on a flawed foundation.

Why Your AI Strategy Depends on Data Quality

If your data quality score is low for any given function, AI initiatives related to that function will fail. Specifically, without high data quality:

Strategic planning is fantasy. AI models can simulate financial scenarios and forecast loan demand — but only if you're feeding them clean, structured data on deposit flows, branch performance, loan delinquency rates, demographics, interest rates, and economic trends. A robust data warehouse and well-managed business intelligence layer aren't nice-to-haves. They're prerequisites.
Marketing falls short. AI can personalize product recommendations and target campaigns. But these efforts are only as good as your underlying customer data. Do you have unified customer profiles? Are marketing engagement metrics integrated with transaction data? Can you differentiate between an SMB and a sole proprietor based on behavioral signals? If behavioral data isn’t integrated into your CRM system, your AI marketing tools will simply amplify the chaos.
Credit analysis is incomplete. AI can improve underwriting by incorporating cash flow data, bill payment history, and real-time payroll deposits. This could open doors to serving thin-file borrowers and gig economy workers. But banks need to re-architect their credit models to accept non-traditional data — and most aren’t there yet. Without robust governance ensuring explainability and fairness, AI-driven lending isn't just ineffective, it's a regulatory time bomb.
Operation analysis is subpar. AI can automate reconciliations and flag errors in real time, but only if your systems know what stage a loan application is in, if teller transaction volumes are accurately tracked, and if your core systems and digital channels are properly integrated. Most community banks and credit unions fail on all three counts.

You Might Be a Data Redneck If…

Wondering if your data quality is killing your AI dreams? You might be a data quality redneck if:

Your dashboards are “curated” by Excel wizards.  If reports depend on one person’s Google Sheet magic, that’s not a system. That’s a vulnerability.
You can’t drill down.  Seeing a deposit outflow trend is nice. Not being able to slice it by age, product, branch, and tenure is disqualifying.
Your business users don’t trust the data. If your head of lending has a “shadow P&L,” your data quality is low.
Strategic planning feels like archaeology. If pulling three years of data for the board requires interns, prayers, and a FedEx envelope, you’re not ready for AI.
You have multiple sources of truth — and none of them are real-time.  Enough said.

The Cultural Problem Nobody Wants to Admit

It isn’t just technology that separates the high data quality performers from everyone else — it’s culture. Among high performers:

72% consider information a strategic asset, in contrast to just 3% of the low performers.
More than half say data is a key driver of strategic decision-making. Zero percent of low performers can say the same.
Almost half of high performers foster a culture around data usage. None of the low performers do.

You can’t “AI” your way out of a culture problem and you can't buy your way out with the latest vendor solution.

Getting Your Data EQ High So Your AI Strategy Will Fly

To set your AI strategy up for success:

Create a unified performance model. Banks should build a common performance model — one that aligns products, branches, teams, and customer segments across the institution. Not just a chart of accounts. A unified data language. Without this, AI agents are flying blind, and so is your CFO.
Surface insights, not spreadsheets. Deploy operational intelligence platforms that sit on top of your core, LOS, CRM, etc., but doesn’t replace them. The goal: enable business users to ask questions and get answers, not wait three weeks for an analyst to “pull a report.” Strategic data access is the next-gen BI.
Tie data quality to business outcomes. Want executive buy-in? Show how poor data quality leads to missed growth goals, failed marketing campaigns, or compliance risk. Improving branch-level data visibility can uncover millions in growth gaps.
Plan AI use cases around data maturity. Don’t build an AI copilot for frontline bankers if your product mapping still uses four-letter codes from 2003. Start with high-data quality areas (like deposits, lending, or digital engagement) and expand from there. AI needs data with depth, recency, and structure. Otherwise, it’s just guessing.
Deal with the people side of the problem. Identify the “impeders” — the execs who don’t believe data is a strategic asset, don’t foster a culture of data usage, and don’t drive their functions toward higher data quality. They need to change or be changed. Next, address the skill gaps. Too many marketers are weak in analytics. Asking them to optimize digital channel spend or develop predictive churn models is a setup for failure.

The Bottom Line on Data and AI Strategy

Banks and credit unions that rush into an AI strategy without improving their data quality are going to learn an expensive lesson: garbage in, garbage out — at faster and greater scale.

The institutions that will win in the AI era will be those that do the unsexy, unglamorous work of getting their data house in order — not the ones with the biggest AI budgets.

Ron Shevlin is chief research officer at Cornerstone Advisors. He was named one of the World's Top Fintech Influencers in 2026 by FinTech Magazine. Tune in to Ron’s What’s Going On In Banking podcast and follow him on LinkedIn and X.