The New Metrics of Underwriting: What AI Helps You See (That Excel Doesn’t)

Underwriting has always been about asking the right questions, the ones that reveal whether a deal is worth pursuing. It’s not about a laundry list of questions but rather identifying the critical ones determining a deal’s viability. The most intelligent decisions don’t just come from what you ask; it’s about seeing the right patterns, fast.

Legacy tools like Excel have served the industry well, helping lenders model deals and document assumptions. They are familiar and flexible in the way many lenders work. However, as portfolios grow in size and complexity, and market dynamics shift more rapidly, static spreadsheets leave strategy and risk leaders with blind spots. 

This article will discuss how AI-powered underwriting can help change that in the commercial real estate (CRE) industry. We’ll review what AI underwriting metrics to look for to help scale your portfolio. As well as how teams can gain visibility into risks and performance patterns that traditional tools weren’t built to show. 

From Spreadsheets to Strategic Signals

Underwriting with AI isn’t about discarding your current process. It’s about expanding what your process can reveal.

Unlike spreadsheets, which rely on fixed templates and manual data entry, AI-powered underwriting introduces a dynamic system that pulls data from diverse sources like financial statements, market trends, borrower history, asset performance, and real-time economic indicators. It ingests information across systems, structures it in real time, and unlocks new signals that fuel sharper decision-making. This can provide a comprehensive risk assessment. Think of it like moving from a still photo to live video.

Here’s what becomes possible when AI powers underwriting:

• Ingesting and analyzing deal data at scale
• Detecting patterns, outliers, and inconsistencies instantly
• Auto-tagging and comparing rent rolls, income statements, and sponsor details
• Surfacing trends hidden within complex, multi-source datasets

With AI, you’re not just reviewing the deal before you—you’re seeing how it connects to broader market patterns and potential risks.

AI Underwriting Metric #1: Time To Decision

Speed is a competitive advantage in commercial real estate lending, but speed without control can increase risk. That’s why tracking time to decision is such a powerful performance indicator. AI enables lenders to automatically measure the time it takes to move a deal from intake to credit committee readiness. This gives leaders clear benchmarks across originators, deal types, and internal workflows. Reducing underwriting cycle times is one of the most effective ways to unlock operating leverage.

When lenders track the time to decision with AI-powered underwriting, they gain several strategic and operational benefits beyond just reducing the underwriting cycle time. They gain:

• Improved Deal Flow Management: Faster decisions enable lenders to handle more deals simultaneously and allocate resources more strategically to keep the pipeline moving.
• Enhanced Client Experience: Faster decisions mean lenders can provide borrowers with a more responsive and agile experience, building stronger client relationships.
• Competitive Advantage: Reducing cycle time can help lenders respond faster to market opportunities, potentially winning deals that slower competitors miss.
• Risk Management: Faster decision-making reduces the window for market conditions to change between underwriting and closing, helping to maintain loan profitability and mitigate risk.
• Operational Efficiency: Shorter underwriting cycles free up underwriters to focus on complex deals or strategic analysis, maximizing the value of human expertise.

This metric doesn’t just help improve operations. It allows teams to move faster without compromising judgment. 

AI Underwriting Metric #2: Confidence Scoring: Turning Uncertainty Into Insight

Traditional underwriting relies heavily on backward-looking data and standardized metrics to assess risk and make lending decisions. It is built on a framework of financial statement analysis, credit analysis, and collateral evaluation to determine loan-to-value (LTV) ratios, debt service coverage ratio (DSCR), and market analysis. 

But what happens when data is incomplete or conflicting? For example, consider an acquisition loan for a property with plans for lease-up or development, where tenant data may be missing or unknown. That’s where AI can step in to provide a confidence score, which is a quantifiable measure of certainty behind every data point. 

How does AI provide a confidence score? Blooma AI provides a confidence score by analyzing a combination of borrower data, property data, and market data using advanced AI models. Here’s how it works: 

• Data Aggregation and Analysis: Blooma pulls data from financial statements, property performance metrics, market comparables, and borrower credit profiles to build a comprehensive risk profile. 
• Risk Assessment Models: AI models evaluate factors like DSCR, LTV ratios, historical cash flows, and projected rental income, quantifying financial stability and potential risk.
• Stress Testing and Scenario Analysis: Blooma simulates various stress scenarios, such as interest rate spikes, tenant defaults, and market downturns, to assess how adverse conditions could impact asset performance and loan repayment. 
• Predictive Analytics and Pattern Recognition: AI analyzes historical loan outcomes and market trends, identifying patterns associated with both positive and negative loan performance to refine the risk assessment. 
• Probability Scoring: Each data point is weighted, factoring in the baseline risk assessment and stress test outcomes, resulting in a confidence score reflecting the likelihood of successful loan repayment under current and stressed conditions. 
• Human-in-the-Loop Validation: Underwriters review the AI-generated confidence score, providing contextual insights or adjustments, ensuring that the final score is data-driven and expert-validated. 

Confidence scoring isn’t just about risk detection. It’s about building a more transparent underwriting process. As McKinsey notes, applying data analytics to underwriting improves decision quality and drives risk transparency across institutions.

AI Underwriting Metric #3: Portfolio Risk Matrix: Mapping Risk Exposure Across The Portfolio

Underwriting isn’t just about assessing individual deals. It’s about understanding how those deals fit into the larger risk landscape. The portfolio risk matrix is an AI-powered metric that visualizes asset-level risk across a lender’s entire CRE portfolio, offering a dynamic, data-driven view of exposure patterns.

The portfolio risk matrix aggregates data at the asset level, mapping risk based on property type, loan structure, borrower profile, and geographic location. It quantifies how risks are distributed across the portfolio, allowing lenders to identify potential concentration risks or emerging vulnerabilities. Here’s how it works:

• Data Visualization: Blooma aggregates asset-level data, such as property type, geographic location, borrower profile, and loan performance, and presents it in an interactive map format.
• Risk Layering: Multiple data layers are applied to highlight risk concentrations based on criteria like loan type (e.g., bridge loans, permanent loans), property type (e.g., multifamily, retail), or borrower creditworthiness.
• Scenario Analysis: Users can apply various stress scenarios (e.g., economic downturns, interest rate hikes) to see how different assets or regions might be impacted, identifying potential exposure points.
• Actionable Insights: The map flags assets or borrower segments with heightened risk, enabling lenders to proactively address potential vulnerabilities through targeted strategies, such as refinancing, tenant outreach, or loan restructuring.
• Real-Time Updates: Blooma continuously updates the exposure map as new data is ingested, providing a dynamic view of portfolio health and emerging risks.

Imagine a lender with a CRE portfolio that’s heavily concentrated in multifamily properties across two metro areas. The portfolio risk matrix reveals:

• 40% of loans in one metro area are bridge loans with near-term maturity dates—a potential risk in a rising interest rate environment.
• 30% of assets show risky vacancy rates, suggesting potential cash flow concerns. 
• 25% of properties are backed by a single sponsor, increasing exposure to a borrower’s financial health.

With these insights, the lender can proactively mitigate risk by diversifying new originations across geographic areas and asset classes, target high-risk loans for refinancing, and monitor borrower performance more closely in over-concentrated segments. 

By layering structured deal data with external market inputs, AI creates heatmaps and exposure dashboards that answer questions like:

• Are we over-concentrated in a weakening submarket?
• Is sponsor risk diversified or overly reliant on a few players?
• How do projected rent or value shifts impact portfolio health?
• How would a 5% dip in market rents affect overall performance? 

As highlighted by CompStak, CRE underwriting today benefits from data-driven benchmarking of regional cap rates, occupancy levels, and lease comps.

Traditional underwriting assesses risk deal-by-deal, but the portfolio risk matrix provides a broader, more strategic view. It not only helps lenders pinpoint where risk is concentrated but also tracks how that risk evolves over time, enabling proactive asset management and more informed decision-making

AI Underwriting Metric #4: Exception Rate: Quantifying Operational Friction

Exceptions are part of any underwriting process. They’re deals that don’t fit the mold, such as missing documentation, unusual DSCR calculations, or inconsistent sponsor information. Traditionally, these outliers are handled case-by-case, but AI takes it further by transforming exceptions into a measurable metric: the exception rate. 

So what does it measure? Exception rate quantifies the percentage of deals flagged with exceptions within a specific timeframe. AI-powered underwriting systems track and categorize these exceptions, revealing patterns that can drive operational improvements. 

By tagging deal anomalies and learning from repeat behavior, AI can reveal:

• Risk Detection: A rising exception rate could indicate a trend of incomplete documentation from a specific broker or asset type.
• Process Optimization: Identifying where exceptions are most common helps teams pinpoint inefficiencies. Understanding which exceptions tend to get approved and which stall deals can reveal opportunities for process improvements. 
• Model Calibration: If underwriters consistently override certain exceptions, it may signal a gap in AI training data or risk assessment criteria. It can also reveal which internal reviewers are more or less conservative in decision-making.

Let’s say a lender processes 100 deals in a month, and 20 are flagged as exceptions due to incomplete rent roll data, inconsistent financials, or discrepancies in DSCR calculations. The Exception Rate is 20%. 

Upon deeper analysis, AI may reveal that:

• 50% of those exceptions are tied to deals from one broker, indicating a potential training gap.
• 20% are overrides where underwriters adjusted AI-generated recommendations, highlighting areas where the model’s accuracy could be refined.

By transforming exceptions into a quantifiable metric, AI helps underwriters not just react to outliers but proactively address them, turning exceptions from a manual process into actionable insights. 

AI Underwriting Metric #5: Risk Velocity Analysis: Tracking Shifts In Risk Over Time

Risk is not static. It evolves as market conditions shift, tenants vacate, or loan terms near maturity. Risk velocity analysis assesses how quickly a risk can impact a CRE asset or portfolio once it materializes. It evaluates the speed and severity of potential risks, helping lenders prioritize which risks need immediate attention versus those that can be monitored over time.

Risk velocity analysis tracks changes in core risk metrics such as DSCR, occupancy rates, tenant health, and loan-to-value (LTV) ratios over time. Unlike the portfolio risk matrix, which provides a snapshot of current risk, this metric shows the pace and direction of change. Here’s an overview of how it works: 

• Data Collection: AI continuously pulls data from financial statements, rent rolls, market reports, and borrower profiles to establish baseline metrics.
• Trend Analysis: The system monitors these metrics monthly to identify shifts that indicate increasing or decreasing risk.
• Anomaly Detection: AI flags significant deviations from historical baselines, such as a rapid drop in DSCR or a spike in tenant vacancies.

For example, consider a lender’s retail loan portfolio that has experienced a 15% drop in occupancy over the past two quarters, along with a 0.3 decline in DSCR. Risk velocity analysis would:

• Flag this as a concerning trend, prompting closer monitoring or intervention. 
• Highlight correlated risk factors, such as regional economic downturns or tenant bankruptcies.
• Let the lender run projections on how further declines might impact overall portfolio stability.

While the portfolio risk matrix helps lenders see where risks are currently concentrated, risk velocity analysis reveals how those risks are shifting—and how quickly. Providing a forward-looking view of risk enables proactive mitigation strategies and more informed decision-making.

Smarter Metrics, Sharper Strategy With Blooma 

What separates today’s top-performing CRE lenders isn’t just gut instinct. It has better visibility.

AI underwriting metrics give strategy and credit leaders a sharper edge, rooted in real-time, portfolio-level insights. It’s not about replacing underwriters—it’s about equipping them with more precise data, deeper context, and faster confidence in every decision.

In this market, foresight is everything. And underwriting with AI is how you get there. By using Blooma’s solution, you can:

• Cut deal processing time by up to 85%
• Improve decision quality with better data and analytics
• Grow your commercial lending business without a proportionate increase in staff

Curious what new metrics your team could be tracking? Learn more about how Blooma makes AI in underwriting powerful, practical, and built for CRE.

See it in action now with our on-demand demos.