A deal team's capacity is the number of full underwritings it can finish in a week, not the number of people on the org chart. Count it once, in hours, and compare it with the deals that reach you. The gap is the part of your pipeline that gets a no without a close read.
Updated 26 October 2026. A brief for principals, CIOs and heads of acquisitions or credit: the people who sign.
Key takeaways
- Capacity = analyst hours left for underwriting each week ÷ hours one full underwriting takes. Headcount is only one input.
- Most of an analyst's week goes to calls, decks and lender questions, so the underwriting share is the number to measure, not assume.
- The deals you pass on without a full read are the cost you never see on a report.
- On our engine, an underwriting took a median of about 58 minutes over the 30 days to 18 October 2026 (44 runs); one in ten took longer than about 95 minutes.
- The limit then moves from building models to deciding, which is the part that should stay with your team.
Why headcount is the wrong number
Ask a principal how much their team can underwrite and the usual answer is a headcount: "two associates and an analyst." Headcount tells you cost. It does not tell you throughput, because an analyst's week is split between the model and everything around it: broker calls, the lender's questions on last month's deal, the committee deck, the rent roll that arrived as a scanned PDF.
The useful number is narrower: how many full underwritings does the team finish in a week, where full means a model you would put in front of your committee. Everything else in this brief follows from that one count.
The capacity formula, with stated inputs
The formula has three inputs, and all three are yours to measure. The example below uses stated inputs, not market statistics: replace them with your own week.
| Input | Example value | How to get yours |
|---|---|---|
| Analysts who underwrite | 2 | The people who build the model, not the people who review it |
| Hours each spends on underwriting in a week | 20 | One ordinary week of calendars, minus calls, decks and lender questions |
| Hours for one full underwriting | 8 | The last five deals your team took to committee, start to model |
| Capacity, full underwritings a week | 5 | 2 × 20 ÷ 8 |
| Deals that reached the team that week | 25 | The inbox and the CRM for the same week |
| Share read in full | 20% | 5 ÷ 25 |
On these inputs, four deals in five get a decision from the teaser, the asking price or a gut read. Some of those are right to drop. The trouble is that nobody knows which ones, because nobody read them.
What the deals you do not read cost
The cost of capacity is not the analyst's salary. It is the deal you would have bought and never opened. Our brief on screening days and missed deals works one stated example: a $10,000,000 purchase with $4,000,000 of equity and an assumed 1.8x equity multiple carries $3,200,000 of profit. Missing one such deal a year because it sat in the queue costs more than most analyst teams.
The same arithmetic runs the other way on risk. A deal the team did read, but in a hurry, is where a figure that disagrees across the files gets through. In one deal review of a six-file package in October 2026, our engine ended with 6 red flags and 46 more items marked to review. What a deal review catches shows how a rent gap of $124 per unit, monthly, becomes a seven-figure value question.
What we measured on our own engine
The run times below come from one clock: every run started in the 30 days to 18 October 2026 that finished and was charged, ours and our clients' alike. They are run times of our engine, not of your team.
| Product | Median, minutes | p90 (one run in ten is slower), minutes | Runs |
|---|---|---|---|
| Underwriting: memo and live Excel model | about 58 | about 95 | 44 |
| Value Opinion | about 45 | about 81 | 66 |
| Comps | about 19 | about 49 | 76 |
| Market rent | about 16 | about 41 | 90 |
What changes is not that a model appears faster. It is that the formula's third input stops being the limit. If one full underwriting no longer takes a working day of an analyst's time, the team's capacity is set by how many deals it can discuss and decide, which is the work a principal wants the team doing.
What capacity does not replace
An AI analyst reads the offering memo, the rent roll and the T-12, gathers the property record and comps from 2,100+ sources and returns a memo and a model with a source for each figure. Three things stay with your team:
- The call. When the rent roll and the T-12 disagree and the answer depends on a phone call to the broker, a person makes it.
- The judgment. Whether a 0.97x coverage on trailing income is a reason to walk or a reason to restructure is your committee's view.
- The signature. The analyst signs nothing, writes to no one outside your firm without your ok, and spends nothing. What an AI analyst never does alone explains how those lines are held in the product.
How to count your own capacity in one week
- Pick an ordinary week, not a quarter end.
- Count every deal that reached the team: inbox, CRM, broker calls.
- Count the full underwritings finished that week, as defined above.
- Ask each analyst for the hours that went to underwriting, from the calendar, not from memory.
- Divide, and write the share read in full on one line for your next partners' meeting.
If the share is low, there are two ways to raise it: more analyst hours, or fewer hours per underwriting. Titleman vs outsourced CRE analysts compares the options side by side, and from probation to a team of analysts shows how firms start with one deal a day and widen from there.
To see the count on your own pipeline, book a demo and forward the deals from last week.
Updated 26 October 2026. Written by the Titleman team with AI assistance. Run times: one query against production, 30 days to 18 October 2026. Worked examples use stated inputs, not market statistics.
FAQ
How do I measure my deal team's capacity?
Take one ordinary week. Count the analyst hours that actually went to underwriting, not to meetings, broker calls and lender questions, and divide by the hours one full underwriting takes your team. That is your capacity in deals a week. Then count the deals that reached you in the same week. The gap between the two numbers is the share of your pipeline nobody reads closely.
Why is headcount a poor measure of capacity?
Because most of an analyst's week is not underwriting. A team of four that spends half its week on calls, decks and lender questions has the underwriting capacity of two, and adding a fifth person adds less than a fifth if the new hire also needs review time from a senior.
How long does an AI underwriting take?
On Titleman's engine, the median underwriting took about 58 minutes from request to delivered files over the 30 days to 18 October 2026, across 44 runs that finished and were charged. One run in ten took longer than about 95 minutes. The output is a memo and a live Excel model with sources for each figure.
Does more capacity mean fewer analysts?
Not by itself. More capacity means more of the pipeline gets a full read before someone says no. Whether that becomes more deals reviewed by the same team, or the same deals reviewed earlier, is a choice for the person who runs the team.
What part of underwriting still needs a person?
The judgment and the calls. When the rent roll and the T-12 disagree and the answer depends on a phone call to the broker, a person makes it. An AI analyst reads, ties the documents together and builds the model; it signs nothing and writes to no one outside your firm without your ok.