CRE Underwriting Software: Extraction, Modeling, Screening and Deal Management Are Different Tools

Why “underwriting software” means 4 different things

Search the term and most results are ranked lists of 10 or 15 products. Those lists rarely mention that the products do different jobs. A tool that reads a rent roll and a tool that builds a 10-year cash flow are both sold as “underwriting software,” and neither can do what the other does.

Once a broker’s email arrives, underwriting a deal involves four steps:

1. Getting the numbers out of the documents.
2. Putting them into a model to produce returns.
3. Deciding whether the deal deserved a model in the first place.
4. Keeping track of the deal, its documents, the people involved and what was decided.

Software has grown up around each step separately. A few vendors cover 2 or 3, but the core jobs stay distinct.

Document extraction tools read the offering memorandum, the rent roll and the T12 and turn them into structured fields: unit, tenant, square footage, rent, expiry, line-item income and expense. The output is a table or a populated template.

Modeling tools take those fields and produce a cash flow and returns. At most firms the modeling tool is Excel, usually an in-house template refined over years. ARGUS handles lease-by-lease modeling for office, retail and industrial. A newer group of products offers structured models that run in a browser.

Screening tools score a deal against the firm’s investment criteria, such as asset type, market, size, going-in yield and price per unit or per square foot. The output is a pass, a closer look, or a ranking against the rest of the pipeline.

Deal management platforms hold the pipeline, the record for each deal (documents, emails, counterparties and notes), its current stage, and who is responsible for the next step. Some also handle approvals, IC materials and investor reporting.

What each category does, and what it won’t do

Two patterns in this table matter most.

Extraction and screening depend on each other. Screening only works once the fields are filled in. Extraction without screening gives you an organized pile of deals that you still have to read one by one.

Modeling stays separate. The model captures the firm’s judgment, and firms are right to resist moving it into a vendor’s format. The useful question for every other tool is whether it can send its numbers into your model and bring the outputs back.

The questions to ask any vendor

he demo will show the product reading a rent roll in 20 seconds. These questions get at what happens next.

• Where does the extracted data go? A spreadsheet you download, the vendor’s database, your model or a deal record? Each answer means a different amount of re-entry.
• Do we keep our data if we leave? Ask to see the export: its format, its scope, and whether it includes documents and emails or only fields.
• Can it export to our model? Directly into the cells your template expects, or into a generic sheet that someone has to map by hand?
• What happens to a deal we pass on? If the answer is “it’s archived” or “you can delete it,” the tool doesn’t treat passed deals as comps. For many firms, that record of passed deals is one of the most valuable things underwriting produces, and it’s usually the first thing lost.
• Does it connect to our inbox, or do we have to upload files? If you have to upload, your analyst is still doing the filing.
• Who checks the numbers? Any tool that reads documents will occasionally get something wrong. Ask how errors are caught, who catches them, and what the tool shows you so you can check its work yourself.
• How does pricing change as we grow? Per-document pricing penalizes a busy quarter. Per-seat pricing penalizes adding a junior analyst. An enterprise contract penalizes being small.

How the 4 fit together in one deal team's week

Take a 3-person acquisitions team buying multifamily and industrial properties in two markets. Between Monday and Friday, 18 deals arrive by email from 11 brokers.

Extraction reads all 18 OMs, along with any rent rolls and T12s attached, and saves the data. Doing this by hand at 20 to 40 minutes a deal, an analyst with other work to do might get through 8 of the 18 by Friday.

Screening scores the 18 deals against the buy box: two markets, two asset types, $15 million to $60 million, and a minimum going-in yield. Thirteen fall out. Five are worth a closer look.

Modeling takes those five. The analyst pulls the extracted data into the firm’s Excel template and runs the returns. Two look strong enough to bid on.

Deal management holds all 18 deals, including the 13 passes and the 3 that were modeled and dropped. Each record has its documents, broker emails, screening score and model outputs. When the same building comes back nine months later from a different broker, the record is already there. When the investment committee asks what else the firm saw in that submarket this year, the answer is a filter, not a search through email.

The team used all four categories of tool, and the model was the only one it built itself.

Where Altrio sits

Altrio covers extraction, screening and the deal record, and leaves modeling to your own model.

It connects to Gmail or Outlook and captures every deal from whitelisted broker senders, pulling in the prior 2 months on connect. It reads the OM, rent roll and T12 and files the numbers into a deal record, scores each deal against the firm's buy box, and merges the same deal from several brokers into one record with each email as a tab. Passed deals stay as comps and deal-flow counts, searchable by market. It exports to Excel, into your model. It is not a modeling tool.

The self-serve edition is $49 per user per month, with a 30-day trial, no card, and setup in minutes. Altrio Enterprise adds firm-wide workflows, approvals, IC dashboards, portfolio and CRM; PGIM Real Estate, AEW and Hines run on it. See pricing and data extraction.

Which do you need?

One analyst drowning in email. Extraction and screening, connected to the inbox. The problem is volume at the front of the funnel, and a model (you already have one) does nothing about volume. A deal record matters less today, though if the tool gives you one, you'll want it within a year.

A 3-person deal team with a good model and no record of passes. Deal management with extraction built in. Keep the model. What you're missing is the record: what came in, what you passed on and why, and the comps that record produces. Check that the platform exports to your template and that passed deals stay searchable.

A 20-person firm with an IC and LP reporting. All 4, and the deal management layer needs workflow: stages, required fields, approvals, IC materials, reporting. This is where enterprise contracts earn their price, because 20 people filing deals 20 different ways costs more than the licence. Keep the model in-house here too; the platform stores its outputs and routes them to the committee.

A firm that buys the 20-person stack at 3 people pays for workflow no one runs. A firm that stays on the 1-analyst stack at 20 people has 20 inboxes and no shared record.

For how the underwriting itself works once the tools are in place, see how to underwrite a commercial real estate deal, and for the front of the funnel, deal sourcing and screening.