How do you underwrite an apartment deal?

Underwriting is a chain of custody for one number

Most people are taught that underwriting is model building. It is not. A model is arithmetic, and the arithmetic is the easy part. Underwriting is the work of proving one number, net operating income, because every figure that matters downstream is derived from it. Price is NOI divided by a cap rate. Loan proceeds are NOI divided by a debt coverage target. Equity is what is left. Return is a function of all three. Get the NOI wrong and every decimal place after it is decoration.

So the sequence below is not a spreadsheet exercise. It is a series of tests, each one asking whether a number the seller handed you survives contact with a second document. The tests run in order because each one depends on the last: you cannot normalize an NOI you have not reconciled, and you cannot size debt against an NOI you have not normalized.

Every step is worked on the same property: a 24-unit building whose seller package reports $177,000 of NOI and prices it at a 6 percent cap. By step six that NOI is $168,000, the supportable value is $2,800,000 rather than $2,950,000, and the loan is $1,771,961 rather than $1,866,888. The work takes an afternoon. It moves the price by $150,000.

Step 1: Gather the actuals

Two documents start every underwrite: the trailing twelve month statement and the current rent roll. Ask for both in the same request, and ask for the T-12 broken out month by month rather than as an annual summary. The monthly columns are the entire point, because one-time items, seasonality, and a rent push made right before the listing are all invisible in a yearly total. If you have not worked with one before, what a T-12 is and what it contains covers the structure, and how to read a rent roll covers the other half.

Prefer a native export from the accounting system, AppFolio, Yardi, RealPage, Buildium, or whatever the manager runs, over a spreadsheet someone assembled by hand. A retyped summary is not a source document; it is a summary of a source document, and the gap between the two is where lines go missing. On self-managed properties a hand-built sheet is often all that exists, which is a fact about the risk, not a reason to skip the check. If you are on the seller side and need to produce a clean statement, the multifamily T-12 template and the rent roll template lay out the line items a buyer will ask for anyway.

Request the supporting documents in the same breath, because they take time to produce and you will want them before the offer, not after: the assessor's property tax bill, the insurance declaration page, the general ledger, twelve months of utility bills, the lease files with start dates and concession schedules, and a current delinquency report. Those are what turn a category total into a verified number.

One judgment call belongs here at the start. If a seller sends a proforma and an offering memorandum but slow-walks the T-12 and the rent roll, that is information. Nobody withholds financials that make the property look good.

Step 2: Reconcile the rent roll to the T-12

This is the step most buyers skip, and it is the one that does the most work. Take the rent roll's in-place rents, annualize them, and set that number next to what the T-12 says was actually collected. On the 24-unit example, 24 units at an average in-place rent of $1,400 is $33,600 a month, or $403,200 a year of scheduled rent. The T-12 collected $358,500.

The gap is the underwriting. Everything else on this page is arithmetic; this $44,700 is the question. It blends vacancy, concessions, bad debt, units occupied by tenants who are not paying, and any rent increases that happened partway through the year. Those five causes have wildly different implications. Rent increases mid-year mean the trailing twelve understates the property and the gap closes on its own. Bad debt means the rent roll overstates the property and the gap never closes. You cannot tell which from the total, so make the seller decompose it line by line. The rent roll template's reconciliation worksheet lays the four buckets out side by side so the seller has to assign every dollar.

This step is also the answer to the occupancy claim in the marketing. An offering memorandum that says "95 percent occupied" is quoting physical occupancy, which counts heads. The collection gap counts dollars. A unit with a tenant who has not paid in four months is fully occupied and worth nothing, and the two figures diverge exactly where the risk lives. As a working threshold, a gap above roughly 8 to 10 percent needs the line-item split before you treat the rent roll's rents as achievable.

Step 3: Normalize NOI to a run rate

A T-12 records what happened. Underwriting needs what recurs. Normalizing is the process of turning the first into the second, and it cuts in both directions: some adjustments help the seller, some help the buyer, and a buyer who only makes the adjustments in their own favor is not underwriting, they are negotiating. Three moves carry this property, and the full mechanics are worked line by line in the five-step T-12 analysis workflow.

• Take out one-time income. Other income totals $31,500, but $12,000 of it is an insurance claim payout booked in a single month. Recurring other income is $19,500. Capitalizing a claim payout at a 6 percent cap means paying roughly $200,000 for money that arrives once.
• Take out one-time expense. One month shows $21,000 of repairs against a $3,000 median for the other eleven. The invoice says building re-pipe, so the extra $18,000 is a capital item, not an operating cost. This adjustment moves NOI up, in the seller's favor, and making it is what gives your other adjustments credibility.
• Put in the cost that is missing. The statement books no management fee because the owner self-manages. The next owner will pay a manager whether or not the previous one did, so a market fee goes in: about 4 percent of adjusted effective gross income, or $15,000 a year here.

Then compare the trailing three months against the prior nine before you finish. On this property the last three months average $32,500 of collected rent against $29,000 for the prior nine, a 12.1 percent lift. That is not automatically fake, but it is unproven, and unproven revenue gets discounted. The rule that follows is worth memorizing: the T-3 tells you what to investigate, not what to capitalize. You value the normalized twelve months and you go read the leases that produced the recent three.

Net of all of it, the reported $177,000 of NOI becomes a verified $168,000. Nine thousand dollars sounds like a rounding error until step six converts it.

Step 4: Screen for red flags

Normalization fixes what you can see. Screening looks for what you cannot. These are pattern checks, and they run against two different documents: the statement itself and the story wrapped around it.

On the statement, the recurring patterns are consistent enough to automate. An expense ratio below roughly 30 percent of effective gross income is almost never an efficient building; it means categories are missing. Property taxes booked as one lump instead of accrued monthly distort every intermediate month. A missing insurance line overstates NOI by exactly the premium. Fewer than four expense categories is too coarse to underwrite. A month with no data, a negative-NOI month, other income that grows while the unit count stays flat, and a trailing-three rent lift more than 10 percent above the prior nine all belong on the list. RTOM publishes its full methodology as 15 T-12 anomaly detectors, with the thresholds written out, and no listing goes active until the critical ones are cleared or documented.

On the marketing side, the tells are different. A headline cap rate computed on proforma NOI rather than actual NOI. An occupancy figure with no economic occupancy next to it. Comparable sales chosen for their price rather than their similarity. Expense assumptions expressed per unit with no source. A rent comp set that quotes asking rents at new lease-ups instead of effective rents at the subject's vintage. The catalog of these is in offering memorandum red flags. None of them is proof of anything by itself. A cluster of them tells you how much of this package to take at face value.

Step 5: Separate the actuals from the proforma

A proforma is not dishonest. It is a business plan, and every value-add deal has one. The failure is structural, not moral: the proforma and the actuals get merged into a single column, and the buyer ends up paying today's price for tomorrow's performance. Keep them apart. One column is the verified history, which is what the seller gets paid for. A second column is the plan, which is what you might create, funded by capital you have not spent yet and executed on a schedule that has not happened yet. The side-by-side on this same property shows how far the two columns can drift.

Discipline in this step is mostly about what you refuse to do. Do not capitalize projected rents. Do not underwrite an expense ratio no building in the submarket achieves. Do not assume the tax bill stays where it is, because a sale often resets it, and appraisal districts tend to move assessments toward the transaction price even in nondisclosure states. Do not treat renovation upside as free; it costs capital, it costs downtime, and it arrives on a lag. Price the plan by discounting what it produces and subtracting what it costs, then decide how much of that value you are willing to hand the seller in advance. The honest answer on most deals is: not much.

Underneath both columns is the same requirement, which is that the historical numbers have actually been checked against source documents rather than accepted from a PDF. Verifying a T-12 is the mechanical version of that: tie collected rent to bank deposits, tie the tax line to the assessor, tie insurance to the declaration page, tie the rent roll to the leases. A verified actual column and a clearly labeled plan column is what a lender, an equity partner, and your own future self all need.

A deal is a T-12 and a rent roll, not a narrative

Nothing in the six steps above is clever. It is reconciliation, subtraction, division, and the willingness to ask a seller an uncomfortable question. What makes it rare is that the industry's default packaging runs the other direction: the offering memorandum leads with the story, the proforma sits in the front of the deck, and the trailing twelve is an appendix. A buyer who reads in that order is being handed conclusions before evidence.

Underwrite the actuals. Price the plan separately, and price it low, because you are the one who has to execute it. Insist on documents that can be traced to a source, and treat anything that cannot be traced as an assumption with a number attached to it. A seller with a clean property loses nothing by proving it, and a verified statement is worth real money at closing precisely because unverified numbers carry a risk discount that shows up as a wider cap rate.

That is the entire premise RTOM is built on. Listings are computed from real T-12 financials, run through the published detector set, and marked verified only when the ledger supports the NOI on the page. The narrative is optional. The documents are not.