The Debt Stack Writes the Brief

By the time an architect gets the call, most of the building has already been decided. A land price, a rent assumption, a cap rate and a loan have written the brief, and the sketch is a downstream document.

I learned this the slow way. In my first job I was handed trace paper with twenty flair-pen strokes on it and told to validate the assumptions. Those strokes weren't the assumptions. The assumptions were in a spreadsheet I never saw. On one Arizona deal I ran nearly 20 iterations to solve it, and the lesson I wrote down years later still holds: iterations matter, but coordinated iterations with the finance and construction team matter a lot more.

So here's the thesis, stated plainly. Capital sets the terms. Design tests them. The architect who understands which financial assumption is carrying a requested change gets to argue with the brief. The one who doesn't gets value engineering as a surprise.

The return requirement writes the budget

Start with the arithmetic every developer runs and most architects never see. Required yield on cost is annual net operating income divided by total cost. Fix the income and the yield, and the total cost is fixed too. Everything you draw has to fit inside that number.

The illustrative study I built for this used a 120,000 GSF project, hard cost on every gross foot, and soft costs at 20% of hard cost. No debt, no exit, no taxes. Even that stripped-down model makes the point: a deeper unit, another stair or a fancier window system changes area, cost or both, and every one of those moves lands back in the pro forma.

Developers usually run it in reverse. Once the site is bought, the project has to produce enough income to support what's already been spent. The building becomes the residual: sale value minus land, minus hard costs, minus a soft-cost allowance, minus financing. In my residual study that was 180,000 SF of gross area with $1.8M of financing cost. What's left over helps decide what the land was worth. It is not a guaranteed profit. Unit count, mix, parking and amenities stop being taste. They're terms in an equation.

That's not a reason for architects to give up design. It's a reason to ask what the program is supposed to accomplish before you defend it.

The same building, worth less

Here's the part that frustrates designers most. You can lose budget without changing a wall.

A cap rate turns net operating income into an estimate of value: value equals NOI divided by the cap rate. Hold the income steady, raise the cap rate, and the implied value drops. A lower value means a smaller loan, and a smaller loan means less money to build with. Nothing about the plan moved. The market's view of risk did.

Interest rates hit twice. They raise the cost of borrowing during the project, and they change what buyers will pay for the income at the end. My rate study fixed NOI at $3M a year and let the rate move. The debt service coverage ratio, NOI divided by annual debt service, slides as the rate rises, and the margin narrows before anyone opens a drawing.

The responses are spatial: a simpler structure, less area for parking and amenities, a smaller scope, or a decision to wait. None of them follows automatically from a rate change, because zoning, bids and the local market still get a vote. But if you're an architect wondering why the client suddenly wants to cut the roof deck, check the rate sheet before you check your drawings.

A cap-rate example is a model, not a forecast. It isolates one variable. Real valuation also depends on the income estimate and the market used to pick the rate. That's exactly why the assumption should be named in the room.

The land starts the clock

The price paid for land enters the project before the building is designed, and it starts charging rent on day one.

My carry study used the simplest possible version: original land basis times an annual rate times months over 12. It's not a loan schedule. It just shows the shape. Approvals, design, construction and lease-up all keep capital tied up, and every month on that path costs money.

Two similar sites can face completely different constraints because they were bought at different prices. That doesn't mean the expensive one has to max out density or cut amenities. It means the team should know that the acquisition basis is pushing on the trade-offs, and that no design move erases a bad assumption about rents, timing or financing.

A cheaper permit can cost more

Permit and impact fees look like line items. Time is the bigger line item.

The permit study modeled a 200-home project with $8M of predevelopment exposure carrying at 7% a year, simple interest. Compare two paths: one with a lower fee and a longer review, one with a $300k permit fee, the same impact fees and three months to review. At that exposure, a few extra months of waiting can eat more than the fee you saved. A cheap permit you wait for is not a cheap permit.

Try it: drag the base fee, the impact fee per home and the review duration, and watch waiting carry overtake the fee you saved.

This is why I want fees and review timing in the model at concept, with the basis and the date recorded. Otherwise nobody can tell whether a cost change came from the city or from the project.

Lease-up is a design constraint

A finished building still isn't a performing asset. It has to fill.

My lease-up study used 180 homes, a 93% stabilization target, fixed operating costs of $580 per available home per month and debt service of $165k a month. Each cohort of new residents gets its own free-rent clock. Slow the absorption and the project carries that debt and those operating costs longer before income catches up. The cash shortfall is real money someone has to fund.

Design can't control the market. It can influence the leasing proposition through unit layouts, shared spaces and the mix. What it shouldn't do is present one absorption timeline as certain. Lease-up assumptions should move every time competing supply, pricing or the unit offering moves.

The budget is set before the architect shows up

This is the thesis the original slug was reaching for, and the evidence supports it in a narrower form than the slug claimed. The debt stack doesn't pick the architect. It picks the budget the architect inherits.

Developers and lenders set an initial cost target from comparable projects, early contractor input and their own experience, often before a design exists. Design then tests that target. Structure, floor-to-floor, facade, MEP and unit mix all move the cost, and the early estimate changes as details show up. Treat the opening number as settled and every later correction feels like an ambush.

Two habits fix most of this:

• Keep cost and design in the same conversation. When a choice changes, say which estimate moved and why.
• Date every estimate. State its scope and its exclusions. Otherwise a changed project and a changed market look identical on paper.

A contractor's price is evidence about one market and one scope at one moment. It's not a universal cost rule. Early ranges should stay ranges until the project is actually defined.

Architecture models too late

Here's the cost of all this showing up late. The pro forma sets the KPIs before anyone draws, and then the architect is held to them.

My study set targets of 120 units, 96,000 net rentable square feet, 82% efficiency and a 6.5% yield on cost ($36M of cost, $2.34M of income, a 5.5% cap rate). The first design starts 20% short. Each design change closes a share of what's left, and the share shrinks: 45% of the gap, then 36%, then 29%. A change before drawings issue costs $60k in fees and delay. The first change after issue costs three times that, and every change after that costs 25% more than the last.

Now watch the start date:

• Model from iteration 0. Four changes take the gap from 20 points to 3.9, for $240k. You land at 115 units and a 6.12% yield on cost.
• Start at iteration 2. Six changes, all before issue, leave 2.7 points for $360k.
• Start at iteration 4. The same six changes recover the same 2.7 points, but two land after issue. The bill is $645k.
• Start as drawings issue. Twelve changes cost $9.76M, and the gap is still 1.7 points.

Try it: drag the cyan marker to start the model later and watch the cost bars jump the moment changes land after drawings issue.

No number of changes closes it to zero, because each one closes a smaller share of a smaller gap. The architect ends up chasing KPIs that were set before the model existed. That's the whole argument for spatial feasibility: a schematic plan can test the assumptions before they harden into a brief that nobody can revisit.

Feasibility is the real design phase

Feasibility decisions decide whether a project happens, how big it is, and what money it gets. Those are design decisions, even when nobody has drawn a plan yet.

Try it: step through four schemes on one site, garden, wrap, podium and townhomes, and compare homes, parking, cost, yield on cost and land residual in one table.

Site price, program, unit mix, budget and financing all interact. If the target program shifts, area, cost and income have to be revisited together, or the original feasibility case quietly stops being true. And feasibility isn't a phase you finish. Estimates and market assumptions keep changing, so the test has to keep running.

The pro forma is a forecast. Treat it like one.

Every pro forma is wrong. The useful questions are how wrong, in which direction, and whether the project survives the miss.

My forecast study used illustrative numbers: $10M of initial cost, $600k of first-year NOI, 3% annual NOI growth and a 6% exit cap. The downside case combines lower growth, a higher exit cap and a higher entry cost. The upside reverses them. A project that only works in the upside has a different risk profile from one that still pencils in the downside, and a waterfall shows who eats the loss first, lenders or equity.

That's a decision tool, not a promise. Its assumptions need to stay visible as the project changes, and design is one of the things that changes them.

What I'd do on Monday

If you're an architect: ask for the five numbers that are writing your brief. Land basis, target yield on cost, cap rate assumption, debt terms and absorption. Then show which design moves touch which numbers.

If you're a developer: get a real test fit in front of the pro forma at iteration 0, not iteration 8. The study above says starting four iterations late can cost you hundreds of thousands of dollars for the same result, and starting at issue costs millions for a worse one.

The drawing doesn't settle the economics. But a drawing that arrives early enough can still argue with them.

Capital writes the first draft. Make sure design gets to edit it.

Run your own land price, yield target and unit mix through a test fit before the brief hardens. Book a TestFit demo and run this on your own site.