How to Use AI for Rent Roll Analysis

The rent roll is the single most important document in a commercial real estate transaction. It is the bridge between the leases and the financials — a summary of who pays what, for how long, and under what terms. Every assumption in your underwriting model traces back to it.

It is also the document most likely to contain errors, omissions, and optimistic interpretations. Sellers prepare rent rolls to support their asking price. Rents may reflect future escalations rather than current in-place amounts. Vacancy might be understated by including signed-but-not-yet-commenced leases. Lease expirations might conveniently exclude a tenant who gave notice but hasn't formally vacated.

The buyer's job is to take the seller's rent roll and pressure-test every line against the source leases. This is mechanical, repetitive work — exactly the kind of work AI handles well. But the analysis that sits on top of it — identifying concentration risk, stacking exposure, and mark-to-market gaps — is where the real value lives.

This guide walks through both layers.

Phase 1: Rent Roll Validation

Before you analyze the rent roll, you need to confirm that the numbers in it are actually correct.

Extracting the Rent Roll

If the seller provides the rent roll as a PDF (common) or embedded in an Offering Memorandum, the first step is getting it into a structured format you can work with.

"Extract the rent roll from Offering_Memorandum.pdf (it starts on page 14). Create an Excel file called Rent_Roll_Seller.xlsx with the following columns: Unit/Suite Number, Tenant Name, Lease Start Date, Lease Expiration Date, Square Footage, Monthly Base Rent, Annual Base Rent, Rent Per Square Foot, Lease Type (NNN/Gross/Modified Gross), and any notes on concessions or free rent periods."

For a clean, well-formatted rent roll, this extraction is straightforward. For a rent roll embedded in a messy OM with footnotes, merged cells, and inconsistent formatting, expect to do some cleanup. AI handles the extraction — you verify the edge cases.

Cross-Referencing Against Source Leases

This is the step that catches real money. The seller's rent roll says Tenant A pays $5,200/month. The executed lease says $4,800/month with a 3% annual escalation that brings it to $5,095 in the current year. That $105/month discrepancy, multiplied across the remaining lease term and capitalized at a market cap rate, can represent tens of thousands of dollars in overstated value.

"I have the seller's rent roll in Rent_Roll_Seller.xlsx and 18 executed lease PDFs in the 'Leases' folder. For each tenant, extract the current base rent from the source lease (accounting for any escalations that have occurred since commencement) and compare it to the rent shown on the seller's rent roll. Create a reconciliation tab in the workbook that shows: Tenant Name, Rent Roll Amount, Lease-Verified Amount, Variance ($), and Variance (%). Highlight any variance greater than 2% in yellow."

This is where AI provides the most direct ROI in the entire underwriting process. Manually cross-referencing 18 leases against a rent roll takes a full day. AI does it in minutes. The output is not a final answer — you still need to check the flagged variances — but it tells you exactly where to look.

Common Discrepancies to Watch For

AI will surface the numbers. You need to understand what the discrepancies mean.

Escalations not yet effective. The seller may show the post-escalation rent even if the escalation date hasn't occurred yet. This inflates in-place income. Check whether the rent shown is the actual current rent or the rent that will be effective after the next escalation.

Free rent or abatement periods. A tenant may be in a free rent period that the rent roll does not reflect. The lease shows $0 for months 1–3, but the rent roll shows the stabilized rent. This is technically correct (the contractual rent exists) but economically misleading for underwriting current cash flow.

Amendments superseding original terms. The rent roll may reflect the original lease terms while an amendment changed the rent, the expiration date, or both. If the seller provides amendments separately from the original leases, the AI needs to process both and give the amendment priority.

Percentage rent or overage rent. For retail properties, the rent roll may include percentage rent projections that are based on the seller's sales assumptions, not actual reported sales. These should be separated from base rent in your analysis.

Phase 2: Stacking Analysis

Once you trust the rent roll data, the next layer of analysis is understanding the lease expiration profile — what happens to the income stream over time.

The Lease Expiration Schedule

"Using the validated rent roll in Rent_Roll_Validated.xlsx, create a lease expiration schedule. Group tenants by expiration year and show: Number of Leases Expiring, Total Square Footage Expiring, Total Annual Rent Expiring, and Percentage of Total Rent Expiring. Include a row for month-to-month tenants. Flag any single year where more than 25% of total rent expires."

The 25% threshold is a judgment call — adjust to your risk tolerance. But the principle is important: concentrated lease expirations create re-leasing risk. If 40% of your income expires in 2028, you need to underwrite that year assuming some tenants leave and some renew at potentially different rates.

Tenant Concentration Risk

"From the validated rent roll, calculate each tenant's share of total annual rent. Flag any tenant that represents more than 15% of total rent. For flagged tenants, also note: remaining lease term, renewal options (if available in the lease data), and whether they are investment-grade or unrated."

A building where one tenant represents 35% of income is a fundamentally different risk profile than a building with 20 tenants each at 5%. AI surfaces this instantly from the rent roll. The analysis that follows — whether that tenant is likely to renew, what the space re-leases for if they leave, how long the downtime would be — is your job.

Weighted Average Lease Term (WALT)

"Calculate the Weighted Average Lease Term remaining for the property, weighted by annual base rent. Also calculate WALT weighted by square footage. Show both numbers."

WALT is a standard metric that lenders and investors use to assess income durability. Rent-weighted WALT tells you how long the revenue is locked in. SF-weighted WALT tells you how long the space is occupied. A building with high rent-weighted WALT but low SF-weighted WALT has a few long-term, high-paying tenants and many short-term tenants — a specific risk profile worth understanding.

Phase 3: Mark-to-Market Analysis

The rent roll tells you what tenants are paying. The market tells you what the space is worth. The gap between the two is either upside or risk.

In-Place vs. Market Rents

"I have submarket rental comps in Market_Comps_Q1_2026.pdf. For each tenant in the validated rent roll, compare their in-place rent per square foot to the average asking rent for comparable space in the submarket (adjust for lease type — compare NNN to NNN, Gross to Gross). Create a column showing the percentage above or below market for each tenant. Summarize the portfolio-level mark-to-market: what is the aggregate in-place rent vs. aggregate market rent?"

Tenants paying below market represent potential upside at renewal — you can underwrite rent bumps. Tenants paying above market represent risk — they may push back at renewal or leave for cheaper space. Neither is inherently good or bad, but you need to know where you stand before building your pro forma.

Loss-to-Lease

"Calculate the total annual Loss-to-Lease for the property — defined as the difference between aggregate market rent and aggregate in-place rent for tenants currently paying below market. Express this as both a dollar amount and a percentage of current gross income."

Loss-to-lease is the embedded upside in the property. It is real, but it is not guaranteed — you only capture it if tenants renew and you successfully negotiate market-rate renewals. Sellers will present loss-to-lease as a reason to pay a premium. Buyers should view it as optionality, not income.

Phase 4: Building the Output

The analysis above produces the inputs for your underwriting model. The final step is assembling it into a format your team can use.

"Compile the validated rent roll, lease expiration schedule, tenant concentration analysis, WALT calculations, and mark-to-market analysis into a single Excel workbook with the following tabs: 1) Validated Rent Roll, 2) Lease Expiration Schedule, 3) Tenant Concentration, 4) Mark-to-Market Summary, 5) Assumptions & Notes. Format for print — headers on every page, property name and date in the footer."

This workbook becomes the foundation for the rest of your underwriting. The cash flow model pulls from it. The investment memo references it. The lender reviews it. Getting it right at this stage prevents errors from compounding downstream.

The Tools

Purpose-built CRE platforms (Dealpath, Prophia, Reonomy) offer rent roll analysis as part of their broader acquisition workflows. If you are evaluating deal flow at volume — multiple properties per week — the structured pipeline and integration with your deal management system justifies the cost.

General-purpose agents (Claude Cowork, ChatGPT) are well-suited for deal-by-deal analysis. Drop the OM, leases, and comps into a folder, run the prompts above, and get a working rent roll analysis in a single session. The flexibility to handle whatever file formats the seller provides — PDF rent rolls, scanned leases, Excel exports — is the advantage.

Spreadsheet templates still work if you already have a proven rent roll analysis template. AI can handle the extraction and validation steps while you maintain control of the model logic in your own workbook.

Where This Breaks Down

Rent roll analysis is one of the cleaner use cases for AI in CRE — the data is structured, the math is well-defined, and the output format is standardized. But there are limits.

Multi-property portfolios with inconsistent formats. If you are analyzing a 12-property portfolio and each property's rent roll uses a different format, column structure, and naming convention, the AI will need more guidance. Batch processing works best when the inputs are consistent.

Mixed-use properties. A property with office, retail, and storage components may have fundamentally different lease structures for each use type. Retail leases with percentage rent, office leases with expense stops, and storage leases with month-to-month terms all need different handling. The AI can process all of them, but you need to prompt for each structure explicitly rather than treating the rent roll as homogeneous.

Verbal or informal lease arrangements. In smaller properties — particularly older multifamily and retail strip centers — some tenants operate on handshake agreements or expired leases that have rolled to month-to-month without documentation. These won't appear in any PDF for the AI to read. If the rent roll includes tenants without corresponding lease documents, flag them for direct verification with the seller.

The deeper challenge, as with every AI-assisted CRE workflow, is that the tools are general and the work is specific. A general agent can extract a rent roll and calculate WALT. It cannot tell you whether the anchor tenant's dark clause makes the lease expiration schedule irrelevant, or that the submarket is about to absorb 200,000 square feet of new competitive supply that will compress your mark-to-market upside.

Purpose-built CRE coworkers — like those from Lumetric — are designed to close that gap. AI that already understands rent roll structures, knows what to flag in a stacking analysis, and produces the output in the format your acquisitions team expects. Not a general agent you teach CRE to. A CRE analyst that already knows the job.