Deal Underwriting in Real Estate: A Step-by-Step Developer's Guide
A practical walkthrough of the development underwriting process — from market assumptions through returns analysis — and where AI is compressing the timeline.
Underwriting a real estate deal is the analytical foundation of every development decision. Done well, it translates market conditions and project assumptions into a clear picture of risk and return. Done poorly, it's the reason projects get built that should never have left the spreadsheet.
This guide covers the full underwriting process for development transactions: what goes in, how the model is built, and where the judgment calls live.
What Deal Underwriting Actually Is
Underwriting is the process of analyzing a deal's financial viability. It translates physical, market, and financial inputs into projected returns — IRR, equity multiple, cash-on-cash yield — and tests those projections against different scenarios.
For a development deal, underwriting is distinct from acquisition underwriting. You're not buying stabilized cash flow. You're building a financial model around a construction timeline, a lease-up assumption, and a long-run stabilized value that doesn't exist yet.
Step 1: Establish the Market Context
Before touching the financial model, you need the market inputs:
Market rent and rent growth. What comparable product is renting for today, and what's the trajectory? How tight is vacancy in the relevant submarket?
Supply pipeline. What's under construction or planned that competes with your project? When does it deliver?
Exit cap rate. What are stabilized assets in this product type and geography trading at? Where are buyers today?
Absorption pace. How long does it realistically take to lease up this product type in this market?
Getting these inputs right matters more than the precision of the model. A 10-basis-point cap rate error on a $200M stabilized asset is $20M in exit pricing. Market context is where deals go wrong first.
Step 2: Build the Development Budget
The development budget covers every cost from land through stabilization:
Land cost. Purchase price, closing costs, carrying costs during the predevelopment period.
Hard costs. Shell and core construction, tenant improvements. Hard costs vary significantly by product type — data centers and life sciences carry far higher per-square-foot costs than industrial or standard office.
Soft costs. Architect, engineer, legal, permits, financing fees, developer overhead. Typically 15-25% of hard costs depending on project complexity.
Contingency. 5-10% of total construction cost is standard. Higher for complex technical builds or fast-moving cost environments.
Financing carry. Construction loan interest through the development and lease-up period.
Every number in the budget needs a source. Industry cost guides from Turner & Townsend and Rider Levett Bucknall provide benchmarks for hard costs. Soft costs and financing terms come from the deal's specific advisors and lenders.
Step 3: Build the Pro Forma
The pro forma projects income and expenses over the hold period. For a development deal, this includes:
Revenue build. Base rent per square foot multiplied by leasable area, phased by absorption schedule. Layer in free rent periods, stepped rents, and any above- or below-market terms from pre-leasing.
Operating expenses. Property taxes, insurance, management fees, maintenance — typically underwritten on a market basis unless you have property-specific data.
Net operating income (NOI). Revenue minus operating expenses, before debt service.
Stabilized value. NOI divided by the exit cap rate. This is what the asset is worth when fully leased and operating.
The difference between stabilized value and total cost is your development spread. If that spread isn't wide enough to justify the development risk over acquiring an existing asset, the deal doesn't pencil.
Step 4: Structure the Capital Stack
How a deal is financed affects returns at every level of the stack:
Senior debt. Construction loans typically cover 50-65% of total capitalization. Interest is floating-rate during construction, often converting to fixed-rate permanent financing at stabilization.
Mezzanine or preferred equity. Fills the gap between senior debt and common equity. Carries a yield requirement that compresses common equity returns.
Equity. The sponsor's contribution and any co-invest or JV equity. Common equity sits last in the stack and takes the most risk.
Target return thresholds for each tier need to be established before you run the model. LP equity in development typically targets 15-20% IRR. Senior debt is priced to credit risk. The development spread has to cover the full cost of capital or the deal doesn't work.
Step 5: Run Sensitivity Analysis
No assumption in a development underwrite is certain. Sensitivity analysis tests the deal under different scenarios:
Cap rate shift. What happens to returns if the exit cap rate is 25 or 50 basis points above base case?
Rent variance. What if market rents are 10% lower at delivery than projected?
Cost overrun. What if hard costs come in 10% over budget?
Lease-up delay. What if absorption takes six months longer than modeled?
Deals that work are ones that still pencil under realistic downside scenarios. If the IRR only works at base case, the risk-adjusted return is worse than it looks.
Where AI Is Compressing the Timeline
AI is being deployed at three points in the underwriting process:
1. Market input gathering. Pulling comp sets, vacancy rates, and cap rate data from aggregated sources — work that previously required analyst hours across multiple databases.
2. Pro forma population. AI populates standard pro forma templates from market data and budget inputs, generating a first-pass model for senior review.
3. IC memo assembly. Synthesizing market context, deal summary, financial outputs, and risk factors into a structured investment committee document.
What AI doesn't do: make the judgment calls. Assumptions around market rent, cap rate, and absorption pace still require experienced analysts who understand the deal-specific context. The productivity gain is in data assembly and document generation — not in replacing the analytical judgment that determines whether a deal gets done.
Development teams using AI in underwriting report cutting first-pass modeling time by 60-80%. That's real, and it compounds across a deal pipeline. But the judgment that separates good deals from bad ones hasn't moved to the machine.