Data Analysis For Real Estate

Starting with the spreadsheet most people get wrong

Data Analysis For Real Estate doesn't require a fancy platform. It requires knowing what you're trying to measure and not letting the data trick you into thinking you've found something real. I spent three years working on commercial deals where analysts would throw a bunch of metrics at a property and call it a day. The problem wasn't the math. The problem was asking the wrong question. Here's what actually happens when you sit down to analyze a property. You pull comps. You run the numbers. You get a cap rate. You make a decision. That sequence sounds right but it misses most of the important stuff. The cap rate doesn't tell you about lease rollover risk. It doesn't tell you about tenant credit quality. It doesn't tell you what happens in year four when the roof needs replacing and the tenant has an exit clause.

The fundamentals everyone skips

You need two things before you open any software: the property level data and the market level data. Property level means lease schedules, operating expenses, capital expenditure history, and vacancy records. Market level means absorption rates, rent growth trends, construction pipeline, and economic indicators for the submarket. Most people start with the property and work outward. That's backwards. The market tells you what's possible. The property tells you if you can actually capture it. I worked on a multifamily deal where the property level numbers looked solid. The cap rate came in clean. The rent roll was. What nobody checked was the construction pipeline. Three competing projects were coming online within eighteen months in the same submarket. We pulled out. The deal would have looked fine on paper and ugly in practice. The market data should always come first because it sets the ceiling on what the property can deliver.

Picking the right tool matters more than you think

Excel handles eight percent of real estate analysis. Everything else runs on either a specialized platform or a custom build. If you're analyzing single family rentals or small multifamily, Excel is fine. You don't need Arion or Yardi. If you're doing commercial, institutional grade analysis, Excel will bite you because the model gets too complex and the formulas start doing things you don't expect. I've seen NPV calculations shift by two percent when someone forgot that a discount factor was referencing a cell that had been filtered out of view. Happened to me once. Cost about forty thousand dollars in advisory fees to fix before closing. For commercial work, RealData, Argus, and ATTOM are the standard tools. Argus is the gold standard for commercial cash flow modeling because it handles lease rollups, expense recoveries, and CAM charges in a way that mirrors how leases actually work. But Argus costs money and has a learning curve that isn't trivial. You can export data from Argus into Excel for presentation purposes, but you shouldn't build your analysis in Excel and call it Argus ready. They're different ecosystems. Free option for smaller deals: CoStar gives you market data but their analysis tools are locked behind expensive subscriptions. Bigger Data Bank offers some free data with limited exports. REIS is institutional grade and expensive. The workaround I use is pulling public records for transaction data and pairing it with local municipal assessment data. It's slower but it's accurate and it costs nothing.

A real case where the data lied to me

I was analyzing a class B multifamily in Atlanta. The cap rate compression story looked good. Rents were growing faster than the metro average. The property had a renewal pipeline with ninety percent of tenants on month-to-month leases. Sounds like a rent growth machine on paper. The catch was the lease structure. Most of those month-to-month tenants were in units that had been converted from original three-bedroom layouts to two-bedroom layouts as part of a renovation that happened five years prior. The conversion hadn't been permitted. The units were technically illegal. When I dug into the building permits through the county records, the violation showed up. Not in any listing. Not in the property disclosures. In the permit office. We renegotiated the price by six hundred thousand dollars after finding that. That's why public record research matters more than the MLS data you're given. Step one is defining the investment thesis. What are you trying to prove? Are you buying for cash flow? For value add? For land play? The thesis determines everything that comes after it. If you're buying for cash flow, you care about stabilized NOI and debt service coverage ratio. If you're buying for value add, you care about rent gaps and renovation costs. If you're buying for land, the property itself is irrelevant. Don't analyze it like it matters. Step two is data collection. Pull the rent roll. Pull the operating statements for the last three years. Pull the tax records. Pull the zoning map. Pull the flood plain data. Pull the environmental reports if the property has any industrial adjacent uses. This takes time but it's the step most people rush. I usually spend two to three days on data collection for a mid-market commercial deal. The due diligence period is thirty to sixty days. Rushing this part is how deals fall apart later.

Step three is the pro forma. Build it bottom up. Line item every revenue source. Line item every expense. Don't use market averages for expenses. Use the actual property history adjusted for inflation and known changes. If the property replaced the HVAC in 2022, don't budget a HVAC replacement in year one of your hold. Most pro formas fail because they use standardized expense ratios instead of property specific data. A 1980s built office building doesn't have the same operating cost structure as a 2010 built one. The data shows this clearly if you look at it. Step four is scenario testing. Run best case, base case, and worst case. Not optimistic, neutral, pessimistic. Those are different things. Best case assumes no vacancy increase and rent growth at the top of the market range. Worst case assumes your key tenant leaves, expenses run two percent above forecast, and exit cap rates widen by fifty basis points. I always run a sensitivity on exit cap rates because that's where most models break. A fifty basis point change in exit cap can swing your IRR by a full percentage point on a ten year hold. Step five is the decision matrix. Write down what would make you walk away. Not what would make you happy. What would make the deal unacceptable. I once walked away from a deal that looked profitable because the primary tenant had a credit rating below BBB. The property was eighty percent occupied by that tenant. One default and the deal goes negative. The numbers on paper were fine. The concentration risk was the real issue. Writing the walk away criteria forces you to confront the thing you're ignoring.

The pitfalls that cost me money

Using trailing twelve month data for seasonal markets. I analyzed a self storage deal in Arizona using TTM data that included the spring peak months. The winter numbers told a different story. Revenue dropped forty percent between April and November. The annualized pro forma was wildly optimistic. Always check monthly breakdowns before annualizing anything. Ignoring the debt structure. Cap rate analysis assumes all cash. If you're leveraging at six percent and the property only cash flows at four percent cap, you're underwater on debt service. I've seen people fall in love with a cap rate without checking if the cap covers the debt service. It's basic but it happens constantly. Debt service coverage ratio below one point zero is a red flag regardless of what the cap rate says. Overweighting recent comps. If a neighborhood has seen a wave of new development, the recent comps are inflated by construction costs and developer concessions. Using them as comparables for an older property gives you a distorted value estimate. I had a situation where the comparable sales suggested a property was worth fifteen percent more than it actually was because the comps included two brand new buildings with luxury finishes. The subject property was a 1970s build with cosmetic updates. The difference was material. Adjust comps for age, condition, and quality, not just location.

What the numbers can't tell you

No analysis captures tenant sentiment. A property can have perfect occupancy and terrible tenant morale. Morale problems lead to non-renewals that show up all at once rather than gradually. That's called a cliff renewal risk and it destroys pro formas. I learned this on a student housing deal where the university changed its housing policy overnight. Half the market lost its primary demand driver. No financial model could have predicted that because it wasn't in the data. Regulatory risk is another blind spot. Rent control measures, zoning changes, environmental remediation requirements. These appear suddenly and they aren't in any historical dataset. I track local municipal meeting minutes for the markets I'm active in. It's tedious but it catches issues that financial models miss. A proposed zoning change can eliminate a value add strategy before you even start the deal. Climate risk is becoming a real factor in underwriting. Flood insurance costs are rising. Some regions are seeing insurer pullouts. If you're analyzing a property in a flood zone, get the actual insurance quote, not the FEMA map estimate. The insurance market has tightened significantly since 2022 and many properties that insured cheaply five years ago now face premiums that are three to five times higher. I had a coastal commercial deal where the insurance renewal came in at $42,000 annually versus the $8,000 the seller's accountant had been paying. That changed the underwriting completely.

A practical workflow for getting started

If you're new to this, start small. Pick one property type you understand. Single family rental is the easiest entry point because the data is accessible and the transactions are frequent. Pull twelve months of rent rolls from a property you're considering. Calculate the actual income per unit. Compare it to the market rate from Zillow or Redfin. Note the gap. Pull the operating expenses from the tax assessor. Calculate the actual cap rate. Then build a simple pro forma with three scenarios. The formula you need most is the debt service coverage ratio: NOI divided by annual debt service. Anything below 1.25 is tight. Below 1.10 is dangerous. I keep a simple spreadsheet with DSCR, cash-on-cash return, and internal rate of return as my three key metrics. That's enough to make a decision. You don't need twenty variables to know if a deal works. For commercial, the same logic applies but the numbers are bigger and the consequences are steeper. Get a mortgage broker to give you actual debt terms for the property type and market before you build the pro forma. Lenders price differently based on property class, borrower experience, and market conditions. The rate you see on a website isn't the rate you'll get. I always budget debt costs three to five basis points above the posted rate because that's what actually happens in practice.

The tools I recommend starting with: Excel or Google Sheets for basic analysis, CoStar or Reis for market data if you have access, and the local county assessor's website for property specific data. That's it. You don't need software that costs more than your first deal's due diligence budget. The analysis quality comes from your questions, not your tools.

When data analysis fails completely

It fails when the market is dislocated. War, pandemic, regulatory crackdown, natural disaster. During the pandemic, commercial office valuations became meaningless because there was no comparable transaction data. The market didn't have a price discovery mechanism. The same thing happened in multifamily during the 2008 crisis when transaction volume dried up. If there are no sales in the last twelve months, any valuation you produce is a guess dressed in formulas. In those situations, you fall back to replacement cost and income capitalization with heavy haircuts. You also slow down. Waiting for the market to reprice is often better than buying into a market that doesn't know its own price. Data analysis is a tool, not a truth machine. It gives you a structured way to think about a deal. It doesn't replace judgment. The best analysts I know treat their models as hypothesis generators, not decision makers. The model tells you what needs to be true for the deal to work. Your job is to figure out whether those things are actually true.