How Do You Extract Financial Data From an OM? Separate Actuals From the Pro Forma

An offering memorandum often puts historical results and a future business plan a few pages apart. An extractor can copy every number correctly and still build the wrong model if it treats the broker's projected NOI as current performance. The first job is to preserve what each number means.

Offering memorandum data extraction turns a commercial property brochure into financial data an analyst can review. For multifamily acquisitions, the important handoff is from source values to explained model inputs: unit counts, rents, operating expenses, and the NOI basis. An underwriting platform should keep those facts traceable as the investment analysis develops.

Build an input sheet that keeps the context

For every material figure, capture the source label, value, unit, financial period, page, and evidence type. Use explicit evidence types such as historical actual, broker projection, third-party reference, and analyst assumption. The team can then inspect an extracted figure before deciding whether it belongs in the model.

Do not treat a blank price as zero. Do not infer that a property has no debt simply because financing is absent from the OM. The input sheet should distinguish an omitted figure from a number the broker explicitly reports as zero.

Keep actuals, the broker case, and your case separate

Suppose an OM lists historical NOI of $720,000 and a stabilized NOI of $900,000. The projected improvement is $180,000. An illustrative bridge might attribute $120,000 to higher rents and $60,000 to lower expenses. Those are hypotheses to test. They are not additional revenue or savings merely because the OM prints them.

Ask when the rent increases begin, which units need renovation, what the work costs, how long units will be offline, and which expense reductions are supported by contracts or quotes. Keep the seller's case intact while constructing your own. That makes disagreement explainable and prevents an analyst edit from rewriting the original evidence.

Cross-check the OM against the deal package

Compare the OM's unit count and income table with the current rent roll and T-12. Check that they describe the same property and periods. A more recent rent roll can legitimately differ from the OM, but the difference should be explained. If a subtotal cannot be reproduced from its detail, keep both values and flag the mismatch.

The OCC's Commercial Real Estate Lending handbook discusses assessing property cash flow and testing underwriting assumptions. For an investment team, the practical takeaway is to test the business plan against evidence and downside cases rather than accept a marketing projection as the conclusion. This article's input sheet is our suggested workflow, not an OCC form.

• Recalculate table totals independently.
• Check whether expenses are per unit, per square foot, monthly, or annual.
• Separate operating expenses from capital spending and financing.
• Identify stale financials and request the missing months.
• Record which market assumptions have supporting comps and which do not.

Turn missing evidence into specific requests

A vague request for more information slows the deal. A useful follow-up names the number and the document needed to check it: the latest tax bill to review a tax assumption, a renewal quote to test insurance, or a current rent roll to confirm the occupancy described in an older OM. Assign an owner and keep the item open until it is resolved.

Avoid filling every missing field merely to complete the spreadsheet. A screen with an explicit unknown can be useful. A polished report built on an invented expense is difficult to defend.

Follow the data into the underwriting report

Cactus uses OMs, rent rolls, and operating statements together in one automated underwriting workflow. The output connects the income and cost story to valuation, buyer returns, and follow-up questions. Document extraction is a step in that process, not a separate product a buyer has to assemble.

For a demo, choose a package with a meaningful difference between historical and projected performance. Ask where the reported value came from, what assumption changed it, and what the team still needs to verify. That is a stronger test than asking whether the software can summarize the OM's executive summary.