Why Rent Roll Data Is Where Deals Go Wrong
Every acquisitions analyst has been there: it's Thursday evening, IC is Monday morning, and you're staring at a 47-page PDF rent roll from a broker OM that somehow uses four different column formats depending on which leasing agent filled it out. Column headers shift between "Unit" and "Apt #" and "Suite" and "Space ID." Rent figures are sometimes monthly, sometimes annualized, occasionally missing entirely. The occupancy column has entries that read "M-T-M," "Month to Month," and "MTM" for the same lease type.
The manual process of cleaning, normalizing, and re-keying that data into your underwriting model isn't just a time sink — it's a source of real errors. Transposition mistakes, missed concessions, skipped vacant units. An analyst working from an unstructured rent roll at 10 PM before a weekend site visit isn't doing their best work.
This guide walks through what a properly abstracted rent roll looks like, which fields actually drive your underwriting conclusions, and where the gaps between what brokers provide and what lenders require tend to appear.
The Minimum Viable Rent Roll: Fields That Move the Model
Not every rent roll field carries equal weight in an underwriting model. Some fields matter for legal diligence and lease review; others are operationally useful; a smaller set directly drives your cap rate and DSCR calculations. The following fields are the ones that, if missing or wrong, will produce a materially incorrect model.
Unit identification and mix
For multifamily: unit number, building/phase (on multi-phase deals), unit type (studio / 1BR / 2BR / 3BR / penthouse), and square footage. For commercial: suite number, floor, rentable square footage, and usable square footage if the lease distinguishes them.
Unit mix matters beyond square footage. A 200-unit apartment building underwritten at average rent per unit will produce a very different model than one underwritten at actual 1BR/2BR distribution — particularly when the subject's unit mix is skewed toward smaller units relative to its comp set.
Lease status and occupancy flags
Occupied, vacant, down unit, model unit, employee unit — each category feeds a different line in your physical and economic vacancy assumptions. A rent roll that lists every unit as "vacant" without distinguishing between market-ready vacant, down/offline, and held-off-market for employee use will produce a misleading physical vacancy figure.
One scenario that catches teams frequently: a 96-unit multifamily asset in a mid-Atlantic market showing 94.8% physical occupancy on the rent roll — but three units flagged as "employee/mgmt" are excluded from the denominator. Actual leasable occupancy is closer to 91.4%. That gap, left undetected, shifts the effective gross income assumption enough to move DSCR from 1.27x down to 1.21x — which for an agency loan with a 1.25x DSCR floor is the difference between qualifying and not.
In-place rent and lease expiration
In-place monthly rent (not annualized, not estimated — actual contract rent), lease commencement date, and lease expiration date form the core of your T-12 income reconciliation. Market rent can be modeled separately, but you need clean in-place figures to calculate the in-place vs. market spread and to stress-test near-term rollover exposure.
Lease expiration bucketing — what percentage of leases roll in Year 1, Year 2, Years 3-5 — directly informs your reversion cap rate assumption and your refinance risk analysis. A building with 60% of leases expiring in Years 1-2 in a softening market is a fundamentally different underwriting problem than one with staggered 3-5 year terms.
Concessions and effective rent
Contract rent and effective rent diverge whenever concessions are in play. Free rent months, lease-up specials, and one-time move-in credits all reduce effective rent relative to contract rent. On a rent roll, these are often missing entirely — they appear in individual lease files but don't get rolled up into the summary document a broker sends with the OM.
Missing concessions on a 40-unit lease-up property can represent 3-5% of annualized gross potential rent. That's not noise — that's the difference between a deal penciling and not penciling at today's rates.
What "Clean" Actually Looks Like: A Structured Output Standard
A clean, model-ready rent roll has a consistent schema. Every unit is a row. Every relevant field is a column. Status codes are normalized to a defined vocabulary. Dates are in ISO 8601 format or at minimum in a consistent MM/DD/YYYY structure. Rent figures are in a single consistent unit (we prefer monthly, but the critical thing is consistency).
The fields that should exist in a normalized commercial rent roll output:
- Unit / Suite ID — normalized identifier
- Floor / Building — for multi-building or multi-story assets
- Unit Type / Space Type — 1BR / 2BR / office / retail / warehouse, per asset class
- RSF — rentable square footage
- Occupancy Status — normalized: Occupied / Vacant-Market-Ready / Vacant-Down / Model / Employee
- Tenant Name — or blank/anonymous for multifamily privacy compliance
- Lease Start / Lease End — ISO dates
- Contract Rent (monthly) — in-place rent per month
- Effective Rent (monthly) — after concession adjustments if available
- Rent PSF (annual) — calculated field: (Contract Rent × 12) / RSF
- Lease Type — NNN / Modified Gross / Gross / FSG (for commercial)
- Reimbursement Structure — expense stop, base year, or full NNN cap detail
- Renewal Options — count and terms if present
The T-12 Reconciliation Problem
The rent roll is a point-in-time snapshot. The trailing twelve-month income statement is a flow document. Getting these two to agree is where most diligence weekends get complicated.
Common discrepancies to hunt for: the rent roll shows a unit occupied at $2,400/month but the T-12 shows annualized unit revenue of $21,600 (implying a partial-year tenancy or a mid-year rent adjustment). Vacancy loss on the T-12 that doesn't reconcile to the unit-level vacancy schedule. Other income line items — parking, storage, laundry — that aren't broken out in the rent roll but represent meaningful revenue.
The reconciliation isn't a formality. Lenders will ask for it. Sellers who can't explain T-12 to rent roll discrepancies are either running sloppy books or hiding something. Either way, the acquisition team should know the answer before it comes up in loan underwriting.
Where Broker OMs and Actual Rent Rolls Diverge
We're not saying broker OMs are dishonest — most aren't, in any deliberate sense. The divergence tends to come from selective framing rather than fabrication: an OM will often show "stabilized" occupancy figures that exclude concession periods, use forward-looking market rent assumptions rather than in-place rent, or aggregate unit mix in ways that obscure the presence of a problematic subset of units.
The three most common OM-vs-actual discrepancies seen in live diligence:
- Occupancy rate inflation — OM shows 95% occupied; rent roll reveals 4 down units and 2 employee units excluded from the denominator.
- Market rent as effective rent — OM underwrites to market rent on vacancies; actual lease-up comps in the submarket show 8-12% concession burn-off on new leases.
- Expense reimbursement overstating income — OM assumes full NNN reimbursement; actual lease abstracts show a mix of gross stop and modified gross leases with reimbursement caps.
None of these are unique situations. They're the normal state of broker OM economics, and the acquisitions analyst's job is to find them before the LOI goes out, not after.
Getting the Data Structured Faster
The traditional approach — download the PDF, extract to Excel manually, normalize columns, build a validation layer — works, but it's a 4-8 hour task on a complex rent roll. Multiply that by the number of deals in active diligence at any time and the math gets uncomfortable quickly.
Structured extraction tools that parse rent rolls directly from PDF or Excel and output to a normalized schema can compress that timeline significantly. The value isn't just speed — it's consistency. When the same extraction logic runs on every deal, discrepancy flags appear in the same place every time. An analyst reviewing an extracted rent roll can focus their judgment on the anomalies that matter, not on whether the column headers match.
The extraction quality check matters as much as the extraction itself: any automated output needs a human review pass on occupancy status flags, date formats, and the rent figure unit (monthly vs. annual vs. annualized-rent-PSF are all common in broker rent rolls and all mean different things). A structured output with clear field definitions is only as good as the normalization logic applied to the source document.
Diligence speed matters — but diligence accuracy matters more. The rent roll is the foundation of the income model. Getting it wrong at that layer compounds through every calculation downstream.
