What is the AI stack cost per deal underwritten for CRE investors? The AI stack cost per deal underwritten is the fully loaded monthly run-rate of every artificial intelligence subscription and usage charge you carry, divided by the number of commercial real estate deals you actually push through those tools in the same period. Most investors track their AI spend as a vague line of monthly subscriptions and never connect it to output. The investors who get the most from AI in 2026 do the opposite: they treat the stack as a cost of production and measure it per deal underwritten and per deal closed. For the full toolkit behind the stack, see our pillar guide on AI tools for real estate investors.
Key Takeaways
- Cost per deal underwritten, not monthly subscription total, is the metric that tells you whether your AI stack is cheap or expensive, because it ties spend directly to output.
- A typical solo to small-team CRE AI stack runs roughly $80 to $250 per month in fixed subscriptions plus variable token or usage costs, which is the number to amortize.
- At eight deals underwritten per month, a $200 monthly stack costs about $25 per deal underwritten; at two deals per month, the same stack costs about $100 per deal.
- Cost per closed acquisition is far higher than cost per deal underwritten, because most underwritten deals never close, so always carry both numbers.
- The fastest way to lower cost per deal is to underwrite more deals on the same fixed stack, not to cancel the one tool doing the heaviest analytical work.
Why Cost Per Deal, Not the Monthly Bill, Is the Right Metric
A monthly subscription total is a vanity number. It tells you what you spend, but not what you get. Two investors can both pay $200 a month for nearly identical tools, yet one underwrites forty deals a quarter and the other underwrites six. The first investor is buying analytical leverage at a rounding error per deal; the second is paying a premium for capacity that sits idle. Cost per deal underwritten exposes that difference instantly. It reframes your AI stack from an expense you tolerate into a cost of production you can manage, the same way an operator thinks about cost per unit turned or a lender thinks about cost per loan originated. Once you see the per-deal number, the decision about whether a tool earns its place becomes obvious, and you stop debating subscriptions in the abstract. This is the discipline that separates a deliberately chosen stack from the accidental sprawl we warn against in our guide to the ideal AI tech stack for CRE investors.
The Three Cost Layers Hiding in Your AI Stack
Before you can compute a clean per-deal number, you have to capture every cost, and most investors only count the first layer. A complete CRE AI run-rate has three.
- Fixed subscriptions: The flat monthly fees you pay regardless of usage. A frontier assistant such as Claude or ChatGPT runs $20 to $200 per month depending on tier, a research tool like Perplexity runs $0 to $20, and a no-code automation connector runs $0 to $30. This is the layer everyone tracks.
- Variable usage: Pay-as-you-go charges that scale with volume, primarily API token costs if you run any tool or custom workflow through a model API rather than a flat-rate chat subscription. For most small investors this is $0 to $60 per month, but it rises quickly if you automate document processing at scale.
- Hidden costs: The integration time, onboarding, and prompt-library upkeep that never show on an invoice but are real. A few hours a month of your own time setting up and maintaining the stack has an opportunity cost, even if it never hits a credit card.
Add all three and you have your true monthly run-rate. For investors deciding which of these layers is worth paying for at all, our analysis of free versus paid AI tools for real estate investors breaks down where the paid dollars actually earn their keep.
How to Calculate Your True Cost Per Deal Underwritten
The formula is deliberately simple, because the value is in tracking it consistently, not in precision to the penny. Take your total monthly run-rate, the sum of all three cost layers, and divide it by the number of deals you ran through the stack that month. A deal underwritten means any opportunity you took far enough to produce a real analysis: a screened offering memorandum, a reconstructed trailing twelve month statement, a populated underwriting model, or a written deal memo. A property you glanced at and discarded in two minutes does not count. The result is your cost per deal underwritten. To get cost per closed acquisition, divide the same monthly run-rate, or better, the annualized run-rate, by the number of deals you actually closed in that period. Carry both numbers, because they answer different questions: cost per deal underwritten measures the efficiency of your analytical pipeline, while cost per closed acquisition measures what the stack costs against revenue-producing outcomes.
A Worked Example: Solo Investor Versus Small Team
Consider a solo investor running a lean stack: a paid frontier assistant at $100 per month, a research tool at $20, and an automation connector at $20, plus about $40 in monthly API usage for document extraction. That is a $180 monthly run-rate, or $2,160 per year. If this investor underwrites eight deals a month, the cost per deal underwritten is about $22.50. Spread across a year of 96 underwritten deals that produce, say, four closings, the cost per closed acquisition is about $540. Against a single acquisition fee or the first month of improved net operating income (NOI) on a closed deal, $540 is trivial. Now consider a three-person team paying $300 a month in subscriptions and $120 in usage, a $420 run-rate. If the team underwrites 30 deals a month, their cost per deal underwritten is just $14, lower than the solo investor despite a bigger bill, because they amortize the fixed cost across more output. The lesson is direct: scale of use, not size of bill, drives the per-deal economics. These figures are illustrative, but the structure holds for any stack you plug in.
How Deal Velocity Changes the Math
The single biggest lever on cost per deal is your deal velocity, because the fixed-subscription layer does not move whether you underwrite two deals or twenty. A $200 monthly stack costs $100 per deal at two deals a month and $10 per deal at twenty. That convexity has a practical implication: in a slow month, your AI stack looks expensive on a per-deal basis, and in a busy acquisition window it looks nearly free. Do not cancel tools in a slow month based on a spiking per-deal number, because the moment volume returns the math inverts. Instead, judge the stack on a trailing twelve-month average that smooths out the seasonality of deal flow. Investors operating below a certain volume should also right-size the stack itself, which is exactly the calculus we walk through in our guide for AI for CRE investors with under $1M AUM, where one paid assistant plus free tools often delivers the best per-deal economics.
Cutting Cost Per Deal Without Cutting Capability
When the per-deal number runs high, the instinct is to cancel a subscription. That is usually the wrong move, because the tool doing the heaviest analytical lifting is also the one that lets you underwrite faster, which raises velocity and lowers cost per deal. The better levers are these: consolidate overlapping tools so you are not paying twice for the same capability, move high-volume document processing to whichever model offers the best price per token, build a reusable prompt library so you spend less of your own time per deal, and increase throughput by running more opportunities through the stack you already pay for. The goal is a lower cost per deal driven by more output, not a lower monthly bill driven by less capability. The AI Consulting Network helps CRE investors instrument their stack this way, so the per-deal number is something they manage on purpose rather than discover on a credit card statement. Investors who want a stack costed and tuned around their actual deal volume can connect with Avi Hacker, J.D. at The AI Consulting Network. For broader market context, research from firms like CBRE continues to track how AI adoption is reshaping operating costs across commercial real estate.
Frequently Asked Questions
Q: What is a reasonable AI stack cost per deal underwritten for a CRE investor?
A: For a solo or small-team investor with a typical $80 to $250 monthly stack, a healthy range is roughly $10 to $40 per deal underwritten, assuming you run several deals through the tools each month. If your number is far higher, the issue is usually low deal velocity against fixed subscriptions, not overspending on tools.
Q: Should I measure cost per deal underwritten or cost per closed deal?
A: Both, because they answer different questions. Cost per deal underwritten measures the efficiency of your analytical pipeline, while cost per closed acquisition ties the spend to revenue outcomes. Cost per closed deal will always be much higher, since most underwritten deals never close, so do not be alarmed by the gap.
Q: Do API token costs really matter for a small investor?
A: For most small investors using flat-rate chat subscriptions, token costs are minimal or zero. They become significant only when you automate high-volume document processing through a model API, where costs scale with the number of pages or tokens processed. Track usage monthly so a scaled automation does not quietly inflate your run-rate.
Q: How often should I recalculate my cost per deal?
A: Review it monthly to catch usage spikes, but make decisions on a trailing twelve-month average. Deal flow is seasonal, and a single slow month can make a perfectly efficient stack look expensive. The smoothed number is the one that should drive whether you add, cut, or consolidate tools.