How to standardize your CRE underwriting workflow across analysts

Define one assumptions set for every analyst to use

Two analysts on the same team, underwriting two multifamily deals in the same submarket in the same month, should not land on a 75 basis point spread in going-in cap rate because one used last quarter's assumptions and the other used a number the broker mentioned on a call. A spread like that is noise, and it is the surest way to make a committee stop trusting the model instead of the deal.

The fix is a firm-wide assumptions table, reviewed on a set cadence, monthly is common and quarterly is the minimum, broken out by asset class because a core multifamily acquisition and a value-add industrial deal do not share market inputs. Analysts do not set cap rates, rent growth, or expense ratios from memory. They pull from the table, and any deviation gets written down as a specific, documented reason tied to the asset, not a feeling about the market.

These ranges are a starting point, and every firm should replace them with its own house view. But that view needs to be grounded in something more current than what a partner remembers from the last deal they closed. The national average multifamily cap rate sat at 5.04% across the 30 most active U.S. markets as of late 2025, up from a cycle low of 4.1% in 2021, and CBRE's most recent cap rate survey found rates held steady through the second half of 2025, with most respondents saying cap rates have already peaked for this cycle. On the operating side, a well-managed multifamily property typically runs an expense ratio between 35% and 50%, and industrial assets run leaner because single-tenant NNN leases push most operating costs to the tenant.

Rent growth assumptions need the same grounding. CBRE's first-quarter 2026 multifamily figures show national vacancy at 4.8% and average rent up just 0.2% year over year, which argues against a first-year rent growth assumption above 1% on a core deal today. On the industrial side, NAIOP's first-quarter 2026 forecast expects rents to hold flat across markets in the near term, with national vacancy at 6.9% after two years of deliveries running ahead of absorption. The raw material underwriting starts from, the seller's offering memorandum, reflects the seller's assumptions, not the firm's, so the table is what converts a broker's pitch into a number the firm actually stands behind.

Run the going-in and exit numbers through a cap rate calculator before they get written into the firm table, so the standard itself is not carrying a spreadsheet error into every deal that references it. Then put the table somewhere analysts cannot quietly override: as required fields inside the deal workspace for every deal, not a PDF in a shared drive nobody opens after the first week.

Build one model template the whole firm runs on

A shared assumptions table does not help if every analyst still builds the model from a personal copy with a different tab order, different formula logic, and a different named range for the same input. Comparability breaks at the model level before it ever reaches the assumptions.

Build one Excel or Google Sheets template for the whole firm, locked so analysts input into designated cells and cannot restructure the calculation engine. The template should force, not suggest, the following: the same order of tabs (assumptions, sources and uses, pro forma, debt schedule, returns, sensitivity), the same formula for the same calculation on every deal, and the same output summary page regardless of who built it or how complicated the deal got.

Version the template itself. When someone finds a formula error or a better way to model a specific lease structure, the fix goes into the master template through one person, typically the head of acquisitions or a senior analyst, and every open model gets rebuilt from the corrected version rather than patched individually. A model template that thirty deals have quietly forked into thirty variants no longer produces comparable numbers.

Debt sizing deserves its own locked section of the template rather than a free-form line, because debt assumptions are where value-add deals get overstated most often. Size debt against both a maximum LTV and a minimum debt service coverage ratio and use the more conservative of the two. Track the lender quotes that back up the assumed rate and terms rather than a number an analyst estimated, so the committee sees the actual financing market behind the model instead of a guess.

Standardize the underwriting checklist so nothing gets skipped

A locked template controls the math. A checklist controls the inputs going into the math, and it matters just as much, because a perfectly built model fed a wrong rent roll number produces a confident, precise, wrong answer. Underwriting starts after a deal clears initial screening, and every deal that clears screening should run through the same sequence of checks before a number goes into the model.

1. Rent roll and lease audit. Verify unit mix, in-place rents, lease expiration dates, concessions, and delinquency against the actual rent roll and lease abstracts, not the broker's summary page.

1. Trailing operating statement reconciliation. Reconcile trailing twelve-month actuals against the seller's pro forma line by line and flag any expense category running more than 10% below the trailing average.

1. Comps pull. Pull three to five closed sale comps and three to five rent comps from the trailing twelve months in the same submarket and confirm they support the proposed basis.

1. Cap rate assumption check. Set the going-in and exit cap rate against the firm's current assumptions table, and document any deviation with an asset-specific reason.

1. Rent and expense growth assumptions. Apply the firm's standard growth curve for the asset class rather than a number that happens to make the deal pencil.

1. Capital expenditure and reserve budget. Underwrite a reserve for replacement and, on value-add deals, an itemized renovation budget with a contingency line of at least 10%.

1. Debt sizing. Size against both maximum LTV and minimum DSCR, log the assumed rate, amortization, and interest-only period, and attach the lender quote it is based on.

1. Sensitivity analysis. Run exit cap rate and rent growth sensitivities at minimum, stressing the deal at a cap rate 50 to 100 basis points wide of the base case.

1. Return summary. Report unlevered and levered IRR, equity multiple, and cash-on-cash by year on a single page a committee member can read without opening the model.

1. Second-analyst review. A second analyst checks every input against its source document before the model goes to the deal lead.

The categories map to who owns each step and when it needs to happen, which is worth writing down once so nobody has to ask.

Ten items feels like a lot until a deal reaches committee missing one of them. A rent roll audit skipped under deadline pressure is how a unit mix error becomes a wrong NOI, which becomes a wrong valuation, which the committee approves because nobody flagged the gap.

Assign consistent review before a deal reaches committee

The second-analyst review on the checklist above catches transposed numbers, stale comps, and assumptions quietly nudged to clear the firm's minimum return threshold. Assign it by rotation so review does not fall to whichever analyst has the lightest week, and require the reviewer to check inputs against source documents rather than only confirm the model runs without errors.

Set a second checkpoint with the deal lead before anything reaches the investment committee. The deal lead is not re-underwriting the deal. They are confirming the assumptions match the firm table, the checklist is complete, and the risks section names the two or three things most likely to be wrong. A committee that receives a model with an unverified assumption is being asked to catch an analyst's error live in the meeting, which is the most expensive place in the whole process to find one.

Write the review sequence down: analyst builds, second analyst checks inputs, deal lead checks assumptions and framing, then committee. Four sets of eyes before capital moves, and each one checking something different rather than repeating the same read.

Version and log every model change

Deals do not stay static between LOI and closing. Rent rolls update, expense actuals arrive, financing terms firm up, and a renovation budget gets revised after a contractor walk. Every one of those changes should update the model through a logged revision, not a silent overwrite of the version the committee already approved.

Keep every model version, dated, with a short note on what changed and why, attached to the deal record rather than buried in a file name like "model_v7_final_FINAL.xlsx." When a committee approved a deal at a 5.10% going-in cap rate and the deal lead later widens it to 5.35% after a rent roll update, that change needs a paper trail showing what moved and who signed off, the same way a pipeline stage approval gets logged when a deal advances. A firm that cannot reconstruct why a model changed between approval and closing cannot tell the difference between a legitimate update and someone quietly moving the numbers to protect a deal they are attached to.

This is also where a shared file system earns its place. Every version, every rent roll, every lender quote lives in one location tied to the deal rather than scattered across individual inboxes and desktops, so the second-analyst review and the deal lead checkpoint above are checking the same source documents instead of five different copies with five different file names.

Benchmark actuals against underwritten assumptions after close

Standardization does not end at closing. The only way to know whether the firm's assumptions table holds up is to check it against what actually happened, and most firms skip this step entirely because the deal has moved to asset management and nobody owns the comparison.

Assign someone, usually the analyst who underwrote the deal or the acquisitions team on a quarterly cycle, to pull actual year-one rent growth, actual expense ratio, and actual NOI against what the model assumed at approval. A deal that underwrote 3% rent growth and delivered 1% happens on its own. A pattern across six deals in the same market means the firm's rent growth assumption for that submarket needs to move before the next six deals get underwritten on the same number.

This step turns the assumptions table into a living document instead of a number someone picked once and never revisited. It also closes the loop between acquisitions and asset management, since the actuals used for the comparison are the same numbers the asset management team is already tracking post-close. Build the habit into the calendar, not into hope: a standing quarterly review where the acquisitions team pulls every deal that closed twelve months earlier and checks the model against reality.

Do the math on what this catches. A firm running 15 deals a year that underwrites a half-point too aggressive on exit cap rate across the book is overstating IRR on every deal it takes to committee, beyond the one where someone happened to notice. Catching that in month fourteen through a benchmarking review costs an afternoon. Catching it in year seven when the fund cannot hit its return target costs the fund's next raise.