All Insights
Development Strategy49 min read

The Math of Building Types: Garden, Wrap, Podium, and How to Model Density on Any Lot

A developer who can sketch the math of a site on the back of an envelope in ninety seconds is a developer who never wastes a quarter on a deal that was never going to work. The math is not complicated. It is one identity with five inputs, layered on top of a construction-type decision that fixes the ceiling. The whole exercise can be done in your head once you have done it on paper enough times.

This is the playbook. Garden, Wrap, Podium, and the high-rise that sits above them. The bottom-up equation that builds from the ground floor up. The top-down FAR shortcut that any zoning code will hand you. The height caps that the building code imposes on each type. The levers that move the answer and the order in which they bind.

The article is arranged in the order the math itself runs. Start with the equation. Walk through each input in the sequence it appears (setback, coverage, stories, efficiency, unit size). Then layer the construction types on top. Then the shortcuts, the constraints, and the cross-checks. By the end you should be able to look at a parcel, know the zoning, and price the rough development envelope for any product type in a few minutes. That is the foundational skill of development underwriting. Everything else, the rents, the cap rate, the debt yield, the IRR, sits downstream of getting this right.

1. The Master Identity

Every multifamily site reduces to the same equation. Memorize it, because every other model you build will be a special case of it.

gross lot            = acreage × 43,560 SF/acre
buildable lot        = gross lot × (1 − setback factor)
building footprint   = buildable lot × site coverage
gross residential SF = footprint × rentable stories
rentable SF          = gross residential SF × efficiency
units                = rentable SF ÷ avg unit SF

Six lines. Five inputs (setback, coverage, rentable stories, efficiency, unit size). One output (units). The construction type does not appear explicitly in the equation because it is sitting inside the third and fourth lines. Coverage and rentable stories are both functions of the construction type, the zoning envelope, and the parking solution. Change the type, and those two inputs change.

The bottom-up framing is the more intuitive one because it follows the way the building actually gets built. Start with the dirt. Take out the setbacks. Lay down the footprint. Stack the floors. Strip out the corridors and mechanical. Divide by unit size. Out come the units.

For one acre at the defaults used in most early feasibility work (15 percent setback, 65 percent coverage, 80 percent efficiency, 1,000 SF average unit), the arithmetic looks like this:

StepCalculationResult
Gross lot1 × 43,56043,560 SF
Buildable lot43,560 × 0.8537,026 SF
Footprint37,026 × 0.6524,067 SF
Gross residential (5 floors)24,067 × 5120,335 SF
Rentable (80 percent)120,335 × 0.8096,268 SF
Units (at 1,000 SF)96,268 ÷ 1,000~96 units

That 24,067 SF footprint applies to a podium or a wrap, where the parking sits inside the building. A garden is different. Surface parking eats most of the buildable lot, which collapses the building footprint to roughly half that size (closer to 12,000 to 14,000 SF on the same one-acre site). The construction type does not just decide what happens above the slab. It decides how much of the lot the slab can cover in the first place, because the parking has to live somewhere, and only podium and wrap put it inside the building envelope.

The rest of the article walks through each input in this identity in the order it appears, then applies the assembled inputs to each construction type.

2. Setbacks: What the Lot Lines Take

The first input in the master identity is the setback factor. It comes first because it operates on the raw lot, before any other constraint applies. Setbacks are the strip of land along each property line that the zoning code reserves as unbuilt buffer, for light, air, fire separation, and street character. They are the first cut taken out of the gross lot and the first place a developer loses density to the code.

Setbacks are specified in linear feet from each lot line, not as a percentage. A typical urban infill multifamily zone might require 10 ft front, 5 ft side (each side), and 15 ft rear. A suburban garden zone might require 25 ft front, 15 ft side, and 25 ft rear. A dense urban transect zone might require zero on all sides (lot-line-to-lot-line construction). The four numbers always read in the same order (front, side, side, rear) and the side setback is usually the same on both sides, sometimes with an aggregate minimum.

The percentage equivalent depends on the lot geometry. The same setback table eats a different share of two lots with the same acreage but different shapes.

Lot shape (1 acre)Setbacks (ft)Buildable areaPercent lost
Square (208 × 208 ft)10 / 5 / 5 / 10188 × 198 = 37,224 SF15 percent
Square (208 × 208 ft)25 / 15 / 15 / 25158 × 178 = 28,124 SF35 percent
Square (208 × 208 ft)0 / 0 / 0 / 0208 × 208 = 43,264 SF0 percent
Rectangular (100 × 436 ft)10 / 5 / 5 / 1080 × 426 = 34,080 SF22 percent
Rectangular (100 × 436 ft)25 / 15 / 15 / 2550 × 416 = 20,800 SF52 percent
Long narrow (50 × 871 ft)10 / 5 / 5 / 1030 × 861 = 25,830 SF41 percent

Notice the rectangular and long-narrow lots lose far more to setbacks than the square lot of the same acreage. A 50-foot-wide lot with 5-ft side setbacks on each side loses 20 percent of its width before any other rule applies, and the building gets squeezed to a narrow band down the middle. This is why frontage matters. A 100-ft-wide site is buildable. A 50-ft-wide site, even on the same acreage, often is not.

Setbacks come in several flavors that the zoning code spells out separately:

Setback typeWhat it doesWhen it binds
Front yard setbackLinear feet from the front property line (street)Always present, usually 10 to 30 ft in residential zones
Side yard setbackLinear feet from each side property lineAlways present, usually 5 to 15 ft, sometimes zero in attached zones
Rear yard setbackLinear feet from the rear property lineAlways present, usually 15 to 30 ft
Corner side setbackLinear feet from the side that fronts a second street on a corner lotLarger than interior side, often equal to front setback
Aggregate side setbackSum of both side setbacks, often with a minimum on eachCommon in narrow-lot suburbs
Build-to lineThe building face must sit at a specific distance from the front lot lineCommon in form-based codes (Miami 21, Nashville)
Step-back or bulk planeUpper floors must step back from the street wallCommon above 35 to 45 ft in dense urban zones
Floodplain setbackBuilding must sit X ft from a regulated waterway or floodplain edgeTriggered by FEMA mapping
EasementUtility, drainage, or access easements run inside the lot but cannot be built onSurvey-driven, site-specific

The step-back is the one that most often surprises a model. Many urban zones permit a building to extend to the lot line at street level (zero front setback) but require the upper floors to step back 10 to 20 ft above a certain height (often 35, 45, or 55 ft). This effectively gives you two different setbacks at two different heights, and the upper-floor footprint is smaller than the ground-floor footprint by the step-back area. A 5-over-1 in a zone with a 12-ft upper-floor step-back loses roughly 10 to 15 percent of the footprint on every wood-frame floor, which moves the unit count by the same percentage.

The 15 percent default setback used elsewhere in this article is the rough percentage equivalent of a typical urban-infill multifamily zone (10 ft front, 5 ft side, 15 ft rear, on a roughly square lot of 1 acre or larger). It is a reasonable screening assumption but always wrong in detail. The right way to model a real site is to pull the actual setback table from the zoning code, draw the buildable envelope on a survey, and measure the area. The percentage equivalent is a screening tool. The actual envelope is the underwrite.

A few rules of thumb for picking the setback input:

  • Lot-line urban (Miami T6, Manhattan, dense downtown): 0 to 5 percent
  • Urban infill multifamily zones (RM-65, MU-3, transect T5): 10 to 18 percent
  • Urban-edge and walk-up (RM-40, RM-50): 15 to 22 percent
  • Suburban garden (R-3, R-4, RM-15): 25 to 35 percent
  • Estate or low-density suburban: 35 to 50 percent

The setback is the input that scales worst with poor lot geometry. A square 2-acre lot loses 15 percent to setbacks. A long narrow 2-acre lot with the same setback table loses 35 to 45 percent. The geometry of the lot has to be in the screening model, not just the acreage, or the early-stage answer will systematically overstate density on poorly shaped lots and understate it on well-shaped ones.

3. Coverage: The 35 Percent You Cannot Build On

The second input in the master identity is the site coverage. Coverage takes the buildable lot (what is left after setbacks) and tells you what share of it the ground-floor slab can cover. It is the input that most people misread on their first pass through the math.

Coverage is the share of the buildable lot that the ground-floor slab actually covers. If the buildable lot (after setbacks) is 37,026 SF and the footprint is 24,067 SF, the remaining 12,959 SF is unbuilt at ground level. That unbuilt portion is doing work, just not building work.

Use of the unbuilt ground areaWhy it eats coverage
Surface parking, drop-off, fire laneEven with a structured garage, you need a porte cochère, leasing-office parking, ADA accessible spaces, and fire-truck access lanes around the building
Courtyard, pool deck, amenityResident amenity competes for tenants, and student or family product especially competes on outdoor amenity
Landscape bufferMost zoning codes require perimeter green space or planted setbacks
Stormwater managementDetention ponds, bioswales, and pervious-area requirements vary by jurisdiction but consume real square footage
Walkways, plaza, hardscapeSidewalks, ADA accessible paths from public right of way to entries, trash enclosure access
Building separationIf the project has multiple buildings, code requires fire-rated separation distance between them

The coverage cap also tends to be set in two layers. Zoning sets one ceiling (typically 50 to 75 percent for urban multifamily, sometimes lower in suburban zones). Practical design constraints set the other. You can almost never push coverage above 75 percent on an actual lot, because at that point there is no room for amenity, ADA paths, fire access, or stormwater.

A few rules of thumb for picking the coverage input:

  • Suburban garden, low density: 0.30 to 0.45
  • Urban-edge wrap or podium: 0.55 to 0.70
  • Dense urban infill: 0.70 to 0.85, sometimes pushed higher with rooftop amenity and below-grade stormwater
  • Tower on a small podium: 0.40 to 0.55 (the tower footprint is small but the podium covers the rest)

Coverage and the setback work as a pair. The product of (1 − setback) × coverage is the actual ground-floor share of the gross lot, and the right way to sanity-check the math is to back into that combined number rather than treating the two inputs separately. At the urban defaults of 15 percent setback and 65 percent coverage, the ground-floor slab covers 0.85 × 0.65 = 0.553, or 55 percent of the gross lot. At the suburban garden defaults of 25 percent setback and 35 percent coverage, the slab covers 0.75 × 0.35 = 0.263, or 26 percent. The combined "ground-floor share of gross lot" is the more honest screening number, because it bundles both inputs into the one ratio that actually matters.

4. Stories and Height Caps: The Construction Type Sets the Ceiling

The third input in the master identity is the number of rentable stories. Stories sit downstream of the construction-type decision, which is set by two ceilings: the IBC story and height cap for the construction type, and the zoning height cap. The binding constraint is the lower of the two.

The building code sets the absolute story and height caps for each construction type. Zoning may impose a lower cap on top of that, and the binding constraint is the lower of the two. Anyone modeling a site for the first time should know both numbers cold for the R-2 residential occupancy.

Construction typeMaterialMax stories (R-2, sprinklered)Max height above grade plane (R-2)Practical role
Type V-BWood frame, unprotected340 ftSuburban garden with surface parking
Type V-AWood frame, 1-hour rated460 ftSuburban garden, slightly taller
Type III-BWood frame interior, non-combustible exterior465 ftLow-rise mixed use
Type III-AWood frame interior, non-combustible exterior, 1-hour585 ft (with NFPA 13)Workhorse of urban podium product
Type II-ASteel or light-gauge framing, 1-hour465 ftLess common in residential
Type IV-HTHeavy timber (legacy)665 ftAdaptive reuse
Type IV-CMass timber, exposed with charring layer985 ftMid-rise mass timber
Type IV-BMass timber, mostly encapsulated12180 ftMid-to-high-rise mass timber
Type IV-AMass timber, fully encapsulated18270 ftHigh-rise mass timber
Type I-BNon-combustible (concrete, steel)Generally 11 to 13 by table, more with modifications180 ft (table value, often exceeded)Mid-to-high-rise concrete
Type I-ANon-combustible, 3-hour ratedUnlimitedUnlimitedSkyscrapers

The numbers above come from IBC Tables 504.3 and 504.4 for R-2 occupancy, and they assume sprinklered construction (which is mandatory for almost all multifamily). Each jurisdiction amends the IBC in its own way, so the actual cap in your market may sit a story above or below the table.

The 75 ft line above grade plane is the single most consequential height threshold in the code. Crossing 75 ft triggers Chapter 4 of the IBC, the high-rise provisions, which add standpipes, fire pumps with backup power, smoke control, elevator recall, areas of refuge, and emergency responder radio coverage. The cost delta from a 74-ft building to an 80-ft building is typically 20 to 40 dollars per gross square foot on the entire building, not just the marginal floor. This is why so many podium projects target exactly 74'-11" at the top of the parapet. The line is a cliff, not a slope.

Zoning height caps sit on top of the code caps and are usually expressed in feet rather than stories. A T4 transect zone in a form-based code might cap at 55 ft. An RM-65 zone caps at 65 ft. The MUR-85 in Seattle caps at 85 ft. The binding height is the lower of the code cap and the zoning cap, and either one is binding before you start counting parking.

The conversion between feet and stories matters because the two caps are expressed in different units. Residential floor-to-floor heights are typically 10 to 11 ft in wood-frame product and 9 to 10 ft in concrete tower product, with a 12 to 14 ft podium and a 3 to 4 ft parapet at the roof. A 5-over-1 lands at roughly 60 to 70 ft total. A 5-over-2 lands at 75 to 85 ft. A 6-over-1 lands at 70 to 80 ft. A 25-story Type I tower lands at 250 to 280 ft. Knowing the rough heights lets you cross-check the story count against the height cap and catch the mismatch before it lands in the model.

5. Efficiency: Common Area, Amenity, and Retail

The fourth input in the master identity is the efficiency factor. Efficiency is the single most abused input in the equation. People plug in 0.80 because they have seen it in a textbook, without knowing what the other 0.20 represents or how it scales with product positioning. The whole 20 percent of the gross residential building that does not produce rent has a name and a structure, and a feasibility model that treats it as a single black-box percentage misses the trade-offs hiding inside it.

The total gross square footage of a multifamily building is the sum of every measured part of the structure. Walk the building from top to bottom and tally up what is in it.

CategoryWhat it includesTypical share of gross residential SF
Net rentable unit interiorThe inside of the units, counted to the centerline of the demising walls78 to 84 percent
Horizontal common circulationDouble-loaded corridors, single-loaded corridors, mailroom-to-elevator paths8 to 12 percent
Vertical coreElevator shafts, stair towers, smoke shafts, fire-rated lobbies3 to 5 percent
MEP shafts and mechanical roomsPlumbing risers, electrical closets, mechanical rooms, transformer vaults, fan rooms2 to 3 percent
Trash and recyclingChute rooms on each floor, central trash room at grade0.5 to 1 percent
Ground-floor lobby and leasingEntry vestibule, leasing office, mail and package room1 to 2 percent
Amenity (interior)Fitness, club room, business center, lounge, co-working, pet spa2 to 5 percent

These add up to roughly 100 percent of the gross residential building. The "net rentable" share at the top of the table is what the 80 percent efficiency input is trying to capture. Push it up to 84 percent by tightening corridors and shrinking amenity, you gain density but lose product positioning. Push it down to 76 percent by adding amenity and oversizing the lobby, you lose density but improve the rent comp set.

The trade-off is not abstract. A 96-unit 5-over-1 at 80 percent efficiency holds 96 units at 1,000 SF average. At 76 percent efficiency (more amenity), the same gross residential building holds 91 units. At 84 percent (lean amenity), it holds 101 units. Five to ten units swing in either direction on the same site, just from the amenity decision.

Amenity benchmarks by product class

Amenity competes for tenants in markets where the residential product is otherwise commoditized. The right amenity package is a function of the rent positioning, the resident profile, and the local comp set. The wrong amenity package, oversized for the rent or mistargeted to the wrong resident, is dead weight on the rentable square footage.

Product classInterior amenity (SF per unit)Outdoor amenity (SF per unit)Notes
Garden, Class B suburban20 to 40100 to 300 (pool, courtyard)Pool and outdoor grilling carry most of the weight
Garden, Class A suburban40 to 80200 to 400Adds clubhouse, business center, fitness
Wrap, Class A urban-edge50 to 10050 to 150 (rooftop or courtyard)Fitness, lounge, co-working, sometimes a pool deck
Podium, Class A urban75 to 150100 to 250 (podium top, rooftop)Full amenity stack with co-working, pet spa, multiple lounges
Podium, Class A luxury150 to 250200 to 400Concierge, screening room, golf simulator, multiple kitchens
High-rise, Class A luxury200 to 400100 to 300 (terraces)Two amenity floors common, often spread across podium and rooftop
Student housing75 to 17550 to 150Heavy on study lounges, game rooms, gaming bays
Senior independent living200 to 400100 to 250Includes dining, library, activity rooms
Affordable LIHTC30 to 7550 to 200Resident services space mandated, amenity restrained

A typical 200-unit Class A podium in 2026 carries 15,000 to 30,000 SF of interior amenity, which is one to two full floors of the wood-frame structure dedicated to amenity rather than units. The decision of whether to put amenity on the second floor (above the podium, taking a full residential floor) versus on the rooftop (taking the roof but not a residential floor) versus distributed across the building (taking partial floors) is a layout decision with a 10 to 15 unit consequence.

The right way to model amenity in early feasibility is to subtract it from gross residential SF explicitly rather than burying it in the efficiency factor. Pull out 100 SF per unit for amenity as a separate line, apply the efficiency factor to the remaining residential gross SF, and the math is cleaner and more defensible.

Retail and ground-floor commercial

Retail at grade is a separate building, financially and physically, sitting inside the same structure as the residential. The math treats it as such.

In a mixed-use 5-over-1, the podium typically holds parking and amenity, but the ground floor street frontage of the podium is often retail. The retail eats coverage at the podium level and counts toward FAR, but produces no residential units. A 24,000 SF podium footprint might allocate 4,000 to 8,000 SF to ground-floor retail (typically 15 to 30 percent of the podium ground floor), with the rest going to parking, lobby, and back-of-house.

Retail enters the math through three channels:

EffectHow it interacts with the math
Coverage and FARRetail GSF counts toward FAR in most jurisdictions. It does not reduce the residential FAR allowance unless the zoning has a separate residential FAR cap.
Rent streamRetail commands its own rent, typically 18 to 45 dollars per SF per year in urban-edge markets, which can exceed or trail residential per-SF rent depending on submarket.
CostRetail shell costs 40 to 80 dollars per SF less than residential to build out (warm shell, no plumbing fixtures, no interior partitions), but the tenant improvement allowance (TI) typically runs 60 to 150 dollars per SF, partially offsetting.
ScheduleRetail tenants take longer to sign and longer to fit out than residential units, typically 12 to 18 months from CO to stabilization versus 6 to 12 months for residential.
Capital stackRetail NOI gets capitalized at a higher cap rate (50 to 150 basis points wider) than the residential NOI in most submarkets, which is why developers sometimes condo the retail and sell it separately.

The decision to include retail is largely market-driven. In street-grid urban contexts with pedestrian frontage, ground-floor retail is often mandatory under mixed-use zoning overlays. In suburban garden contexts, retail is rare and usually a mistake. In urban-edge podium contexts, retail is optional and economic. The question is whether the achievable retail rent, net of TI amortization and longer lease-up, beats the alternative of additional amenity, lobby expansion, or below-grade parking on that same square footage.

The retail share of total project value in a mixed-use podium is typically 5 to 15 percent. Anything above 20 percent and the project is functionally a retail building with apartments on top, which is a different underwriting exercise (the cap rate is wider, the lender is different, and the resident profile shifts).

How amenity and retail change the master identity

For a mixed-use 5-over-1 with retail at grade, the math expands by two lines:

total GSF              = parking GSF + retail GSF + residential GSF + amenity GSF
residential GSF        = footprint × wood-frame stories − amenity floors
rentable residential SF = residential GSF × efficiency
units                  = rentable residential SF ÷ avg unit SF
retail rent            = retail GSF × retail rent PSF × (1 − retail vacancy)
residential rent       = units × avg unit SF × residential rent PSF × (1 − residential vacancy)
total NOI              = residential rent + retail rent − operating expenses

The residential unit count is unchanged by the retail presence (the retail sits in the podium, not in the residential stack). What changes is the project NOI, the cost stack, the financing structure, and the exit cap rate. A model that ignores retail on a mixed-use project is missing 10 to 20 percent of NOI on one side and 10 to 15 percent of cost on the other.

Measurement standards

Multifamily measurement is sloppier than it should be. The industry has not converged on a single standard the way office (BOMA Z65.1) has. The same building can be marketed with three different square footage numbers depending on who is doing the measuring.

MeasureWhat it counts
Gross Building Area (GBA)Everything inside the exterior walls, including parking, retail, amenity, common area, and unit interiors
Gross Residential AreaThe residential portion only, excluding parking and retail
Net Rentable Area (NRA)Unit interiors measured to the centerline of demising walls, sometimes with a load factor applied for common area
Net Usable AreaUnit interiors measured to the inside face of demising walls, no load factor
Saleable AreaUsed in for-sale condo product, sometimes includes proportional share of common area

Pro formas should specify which measure they are using on the gross-to-net line. A model that quotes a unit at 1,050 SF without specifying whether that is centerline or inside-face is missing 30 to 50 SF per unit on a 200-unit building, which is 6,000 to 10,000 SF of either real or imagined building. At a $310 hard cost per GSF, that is 1.8 to 3.1 million dollars of cost on a $40 million project, hiding in a measurement convention.

The cleanest convention for early feasibility is to use the centerline measurement for unit SF and apply efficiency as a separate line. The cleanest convention for marketing is to use NRA with a clearly stated load factor. Mixing the two is what produces the 10 to 15 percent discrepancies between the pro forma and the appraisal.

6. Garden: Type V Walk-Up

With the inputs defined (setback, coverage, stories, efficiency), the construction types are just specific configurations of those inputs. Garden is the suburban workhorse and the simplest configuration.

The garden product is two to four stories of dimensional lumber on a slab-on-grade foundation, with surface parking at 1.5 to 1.8 spaces per unit.

The parking is on the dirt, not in the building, which means the building footprint is much smaller than the buildable lot. Surface parking on a flat site consumes roughly 320 to 350 SF of land per stall (the stall itself plus drive aisle and circulation). At 1.6 spaces per unit, parking eats roughly 530 SF of land per unit, which is more than half the typical land cost of a garden site.

The garden math for one acre, with the parking pulled out explicitly:

StepGarden (Type V-A, 4 stories)Calculation
Gross lot43,560 SFone acre
Setback factor15 percenttypical suburban setback aggregate
Buildable lot37,026 SF43,560 × 0.85
Coverage35 percentparking takes most of the rest
Footprint12,959 SF37,026 × 0.35
Stories4Type V-A cap
Gross residential SF51,836 SF12,959 × 4
Efficiency85 percentcorridors short, no elevator core in true walk-up
Rentable SF44,061 SF51,836 × 0.85
Average unit SF1,000 SFsuburban units run larger
Units~4444,061 ÷ 1,000
Density44 units per acrewithin Type V garden range

Notice what changes. Coverage drops from 65 to 35 percent because surface parking is taking the other 30 to 35 percent of the lot. Efficiency rises to 85 percent because the building has no elevator and no structured parking circulation eating gross-to-net. The net effect is roughly 44 units per acre, in the textbook range for garden product (18 to 40 units per acre in walk-up form, occasionally up to 50 on tight sites).

The advantages of garden product are speed, cost, and simplicity. A 4-story Type V-A garden builds in 12 to 14 months for a 200-unit project, costs 180 to 220 dollars per gross square foot in hard costs in most U.S. markets in 2026, and finances cleanly with any agency or bank lender. The constraint is parking. Once the land cost per buildable square foot crosses roughly 30 dollars, the math on surface parking stops working, and the deal needs to escalate to wrap or podium.

7. Wrap: The Texas Donut

The wrap product solves a specific problem. The site is too expensive for surface parking but not expensive enough to justify a podium. The solution is to put a parking garage in the center of the building and wrap the residential units around it on every floor. The garage and the residential are both wood-frame Type V or Type III construction, sitting on the same slab-on-grade.

The garage core takes up 30 to 50 percent of each floor's interior, depending on the floor plate dimensions and the parking module. The residential ring around it is typically 30 to 35 ft deep from the corridor to the exterior window line, which is the depth a wood-frame double-loaded corridor wants to be.

The wrap math for one acre:

StepWrap (Type III-A, 5 stories with internal garage)Calculation
Gross lot43,560 SFone acre
Setback factor12 percenttighter than suburban garden
Buildable lot38,333 SF43,560 × 0.88
Coverage65 percentgarage and residential under the same roof
Footprint24,917 SF38,333 × 0.65
Stories5Type III-A cap
Gross residential SF (before garage deduction)124,583 SF24,917 × 5
Garage share of interior35 percentcentral parking module
Net residential GSF80,979 SF124,583 × (1 − 0.35)
Efficiency on residential portion80 percentdouble-loaded corridor
Rentable SF64,783 SF80,979 × 0.80
Average unit SF1,000 SFurban-edge product
Units~6564,783 ÷ 1,000
Density65 units per acrewithin wrap range

The wrap gets you 50 to 80 units per acre, sitting between garden and podium on the density curve. The cost premium over garden is real but bounded, typically 15 to 25 dollars per gross square foot, because the structure is still all wood frame. The garage is the same wood frame as the residential, with a fire-rated assembly between the parking module and the units, which is a much cheaper detail than the concrete podium it replaces.

The wrap has two structural quirks that distinguish it from garden. The garage column grid has to align with the residential column grid above and below, which constrains the parking layout (typically 60 to 65 ft of garage width allows two parking bays back-to-back with a drive aisle in the middle, plus a ramp). And the wood-frame parking deck has to be sprinklered to a higher standard than the residential, with proper drainage and vehicle-impact protection.

A wrap is what you build when the parcel is at least 1.5 acres (anything smaller cannot fit the garage module and the residential ring around it) and the land basis is too expensive for surface parking but the zoning does not allow you to go past four or five stories.

8. Podium: 5-Over-1, 5-Over-2, 6-Over-1

The podium is the dominant urban-infill product in the United States. The configuration is straightforward. A concrete podium (Type I-A) at the base handles parking, retail, and amenity. Wood-frame residential (Type III-A) sits above. The two are treated as separate buildings under IBC 510.2 horizontal building separation, which is what makes the entire stack possible.

The naming convention reads vertically. A 5-over-1 is five stories of Type III-A wood frame over a single-story Type I-A podium. A 5-over-2 is five wood-frame floors over a two-story podium. A 6-over-1 is six wood-frame floors over one podium level, possible in jurisdictions that have amended the code to allow it (California, Washington State, and a growing list of others).

For the bottom-up math, the critical move is to count only the rentable wood-frame floors. The podium is parking and amenity, not residential, and it does not produce units. Mixing those floors into the gross residential calculation inflates the unit count by exactly the proportion of podium to total floors, which on a 5-over-2 is 28 percent.

The podium math for one acre, 5-over-1:

StepPodium (5-over-1)Calculation
Gross lot43,560 SFone acre
Setback factor15 percenttypical urban infill
Buildable lot37,026 SF43,560 × 0.85
Coverage65 percentsame building, podium and residential aligned
Footprint24,067 SF37,026 × 0.65
Rentable stories5wood-frame only, podium excluded
Gross residential SF120,335 SF24,067 × 5
Efficiency80 percentcentral corridor, elevator core, MEP shafts
Rentable SF96,268 SF120,335 × 0.80
Average unit SF1,000 SFurban product
Units~9696,268 ÷ 1,000
Density96 units per acretextbook 5-over-1 range

The same site, modeled as a 5-over-2, holds the same residential calculation. The additional podium story adds parking capacity (roughly 70 to 90 more cars at one full podium level of 24,000 SF at 320 SF per stall) but adds zero residential units, because the count of wood-frame floors is unchanged.

Where the model breaks for most first-time underwriters is here. People look at a 7-story building, see 7 floors, multiply 24,067 × 7 × 0.80 ÷ 1,000, and arrive at 135 units. That is 40 percent too high. The bottom two floors are parking. They produce no rentable square footage. The correct unit count for that 7-story building is the 5-over-2 version of the calculation, which holds at 96 units, not 135.

The total building height for a 5-over-1 is roughly 60 to 70 ft (a 12 to 14 ft podium plus five 10 to 11 ft wood-frame floors). A 5-over-2 lands at 75 to 85 ft, right at the high-rise threshold, which is why so many projects in this configuration target exactly 74'-11" at the parapet. A 6-over-1 lands at 70 to 80 ft. A 6-over-2 crosses 85 ft cleanly and is no longer feasible under Type III-A without modifications.

9. High-Rise: Type I Tower

Above 85 ft, the building has to be Type I concrete or Type IV mass timber. The math identity does not change, but the inputs do.

The tower coverage drops because the tower footprint is small relative to the buildable lot. A typical residential tower in 2026 has a floor plate of 8,000 to 14,000 SF, sitting on a podium that covers the rest of the lot. Counted against a one-acre site, the tower footprint might be 25 to 35 percent of the buildable lot. The podium below it covers 70 to 85 percent. The two-tier coverage requires two-tier math.

The tower math for one acre, 25-story Type I-B concrete on a 3-story podium:

StepHigh-rise (25 over 3 podium)Calculation
Gross lot43,560 SFone acre
Setback factor10 percentdense urban, tight setbacks
Buildable lot39,204 SF43,560 × 0.90
Tower footprint11,761 SF30 percent of buildable lot
Rentable tower stories25residential
Gross residential SF (tower)294,025 SF11,761 × 25
Efficiency76 percentlarger core, more elevators, MEP risers
Rentable SF (tower)223,459 SF294,025 × 0.76
Average unit SF1,000 SFurban luxury skews larger
Units~223223,459 ÷ 1,000
Density223 units per acretypical urban high-rise

Towers can run higher (Type I-A is unlimited by code), and the density per acre scales roughly linearly with rentable stories. A 40-story tower on the same footprint produces around 360 units per acre. Beyond about 50 stories the efficiency drops further because the core consumes more floor plate (more elevators, more egress, more MEP shafts), and the marginal density per added floor decays.

Two costs grow non-linearly with height. Construction time stretches from 22 to 30 months for a 15-story tower to 32 to 48 months for a 50-story tower, which compounds the interest carry. And the lateral-force resisting system gets exponentially more expensive above 30 stories, particularly in seismic zones.

10. The Top-Down FAR Shortcut

Zoning codes do not usually express the building envelope in the bottom-up terms above. They give you a Floor Area Ratio, a height limit, a setback table, a coverage cap, and a parking minimum, and they let you figure out which constraint is binding.

FAR is the total gross building area divided by the gross lot area. A FAR of 3.5 means the total building can be 3.5 times the lot size. On a one-acre lot (43,560 SF), a FAR of 3.5 allows 152,460 SF of total building.

The top-down equation is much shorter than the bottom-up:

total GSF      = gross lot × FAR
residential SF = total GSF × residential share
rentable SF    = residential SF × efficiency
units          = rentable SF ÷ avg unit SF

The residential share is the portion of the total building that is residential rather than parking, retail, amenity, or back-of-house. For a 5-over-1 podium, the residential share is typically 70 to 80 percent (the podium counts in FAR in most jurisdictions, and the podium is parking and amenity, not residential). For a wrap, it is closer to 55 to 65 percent (the garage is internal and eats more of the total). For a garden, it can be 90 percent or higher if surface parking is excluded from FAR, which it usually is.

The FAR shortcut for the same one-acre podium:

StepCalculationResult
Gross lot1 × 43,56043,560 SF
FAR3.5zoning
Total GSF43,560 × 3.5152,460 SF
Residential share75 percentpodium and amenity excluded
Residential SF152,460 × 0.75114,345 SF
Efficiency80 percentgross-to-net
Rentable SF114,345 × 0.8091,476 SF
Units (at 1,000 SF)91,476 ÷ 1,000~91 units

The bottom-up math came in at 96 units on the same site. The top-down math came in at 91 units. The five-unit gap is within rounding on the inputs, and the answers reconcile.

The two approaches should always reconcile within a few percent if the inputs are internally consistent. Where they diverge, one of two things is wrong. Either the FAR cap is non-binding (the height cap or the parking cap is binding before the FAR is exhausted), or the bottom-up coverage is too aggressive for what the FAR actually allows. In either case, the lower of the two answers is the correct one, because both represent constraint envelopes.

The cross-check between bottom-up and top-down is the single most useful sanity test in early feasibility. Build both, see where they agree, find the binding constraint, and report that as the unit cap. If a developer or a broker tells you a site holds 150 units when your bottom-up and top-down both come in at 105, the broker is the one who has the math wrong.

11. Reading the Zoning Code

Every input in the bottom-up identity has a corresponding line in the zoning code. Pulling those lines correctly is what separates a defensible feasibility model from a marketing pro forma. The exercise is mechanical once you know where to look.

American zoning codes fall into two structural families. Euclidean codes (the older form, named for the 1926 Supreme Court case Village of Euclid v. Ambler Realty) organize the city into use districts (R-1, R-2, R-3, C-1, C-2, M-1, and so on), with each district carrying its own table of dimensional standards. Form-based codes organize the city into transect zones (T1 through T6, sometimes with suffixes like T5-Open or T6-Urban Core), with each transect specifying building form (height, setbacks, frontage type) rather than narrowly specifying use. Most American cities run Euclidean. A growing list (Miami, Nashville, parts of Cincinnati, downtown Buffalo, the Hudson NY downtown) have adopted form-based or hybrid codes. The math identity does not change between the two. The line items just get named differently.

The minimum set of zoning inputs to extract for any multifamily feasibility model:

Zoning inputWhat it capsWhere it shows up in the math
Permitted useWhether multifamily is allowed at all, and at what unit count thresholdGates the entire exercise
FAR (Floor Area Ratio)Total gross building area as a multiple of lot areaTop-down envelope
Maximum height (feet)Total building height above grade planeCaps the rentable stories
Maximum storiesStory count, sometimes separate from height in feetCaps the rentable stories
Front, side, rear setbacksLinear feet from each lot lineSets the buildable lot
Maximum lot coverageGround-floor footprint as percent of lotSets the footprint per floor
Minimum parking ratioStalls per unit, sometimes per bedroomDrives the parking solution and podium count
Maximum density (units per acre or per lot)Direct unit capOften binding in older suburban codes
Minimum open spaceRequired pervious or amenity areaReduces effective coverage
Minimum unit sizeFloor area per unitAffects average unit SF
Frontage and build-to linesWhere the building face must sit relative to the streetConstrains the site plan, sometimes adds coverage
Bulk plane or step-backRequired upper-floor setback from street wallReduces footprint on upper floors
Affordable set-aside requirementInclusionary zoning percentageAffects mix and pro forma
Bonus FAR or height for affordability, sustainability, or TODDensity bonusExpands the envelope above the base zoning

The single most useful trick when reading a new code is to find the use table first and confirm that multifamily at the target unit count is by-right in the target zone. If it is not by-right (it requires a special use permit, a conditional use, or a Planning Commission discretionary approval), the entitlement risk dominates everything else in the model, and the project is no longer a feasibility question but a political one. By-right multifamily is the only product that finances cleanly with a construction lender.

A typical zoning string compresses a lot of these inputs into a short code. Decoding the string is a learned skill specific to each city, but a few common patterns repeat. Numeric suffixes on residential zones (RM-40, RM-65, RM-100) usually indicate either density (units per acre) or height in feet. Letter suffixes (T5-O, MUR-85-TOD, DM-OZ) usually indicate an overlay or modifier. The overlay layer is where most of the density and height bonuses sit, and reading the overlay is what reveals whether a site actually has 65 ft of allowed height or 85 ft of allowed height under a transit-oriented overlay.

Overlay districts and modifiers worth checking on every site:

OverlayWhat it usually grantsWhere it shows up
Transit-Oriented Development (TOD)Reduced parking, increased height, increased FAR within a defined radius of a transit stopCommon near rail in West Coast metros, increasingly common in Sunbelt cities
Historic districtRestrictions on demolition, design controls, height caps, sometimes tax creditsMost pre-war neighborhoods in older cities
Opportunity ZoneNo zoning change, but federal tax benefits for OZ fundsDesignated 2017, currently being renewed for 2026 to 2033
Inclusionary zoningRequired percentage of affordable units, often offset by density bonusMost major metros
Affordable density bonusAdditional FAR, height, or units for income-restricted unitsCalifornia State Density Bonus Law, Washington State HB 1110, Florida Live Local Act
Form-based code transectSpecifies building form by transect zoneMiami 21, Nashville, downtown Buffalo
Planned Unit Development (PUD)Allows departure from base zoning with negotiated commitmentsProject-specific, requires entitlement work
Mixed-use overlayGround-floor commercial requirement, residential aboveMost urban infill corridors
Environmental overlayStormwater, tree protection, slope, wetland restrictionsCommon in jurisdictions with environmental regulations
Airport approach overlayHeight restrictions based on FAA Part 77 surfacesWithin 5 miles of most commercial airports

The Affordable Density Bonus deserves a special mention because it has restructured the math in California, Washington, Oregon, and a growing list of states. California's Density Bonus Law (Government Code 65915, with significant amendments through 2026) allows up to 50 percent additional density and up to three additional stories above the base zoning in exchange for setting aside 5 to 24 percent of units as affordable. Florida's Live Local Act (2023, expanded 2024) preempts local zoning for multifamily with affordable components, allowing the highest density and height in the jurisdiction by-right. Washington's HB 1110 (2023) requires cities to allow up to four units on any single-family-zoned lot.

These bonus regimes have changed the question. The default zoning envelope is now often the floor, not the ceiling. The actual feasibility envelope is the base zoning plus the density bonus that the project can credibly capture by including affordable units. A 1-acre site zoned for 65 units per acre might support 100 units per acre under a 35 percent state density bonus, which moves the building type from a 4-story garden into a 5-over-1 podium. The density bonus is often the difference between a deal that pencils and one that does not.

The variance and PUD path matters where the base zoning will not accommodate the project. A variance is a hardship-based departure from a specific dimensional standard, granted by a Board of Adjustment or similar body, typically on a narrow showing that the strict application of the code produces an unnecessary hardship. A PUD is a comprehensive rezoning of a single parcel or assemblage to a custom standard negotiated with the planning authority. Both are slow (6 to 24 months for a variance, 12 to 36 months for a PUD), both are political (requiring neighborhood consent in most jurisdictions), and both are entitlement risk that does not finance.

A site that pencils only under a variance is not a feasibility candidate. A site that pencils under base zoning and has additional upside under a PUD or a density bonus is the one to pursue. The order of operations is to underwrite the base zoning first, confirm the deal works there, then layer the bonus and the upside as optionality, not as the central case.

12. The Pocket Hack: Three Constants

The full master identity has six lines and five inputs. That is too many to do in your head at a meeting. Collapse all of it into three constants, one per construction type, and the math becomes a one-line answer.

Start from the master identity:

units = 43,560 × acres × (1 − setback) × coverage × stories × efficiency ÷ avg_SF

Plug in the typical inputs for each product type and collapse everything that is constant for that type into a single coefficient. What is left is a one-line formula that gives you units per acre at a glance.

For each product type, calculate units per acre per rentable floor:

units per acre per floor = 43,560 × (1 − setback) × coverage × efficiency ÷ avg_SF

Run the math once for each product type at typical inputs (1,000 SF average unit), and the constants fall out:

ProductSetbackCoverageEfficiencyGarage deductionUnits / acre / floor
Garden (Type V-A)0.150.350.85none (surface parking)11
Wrap (Type III-A)0.120.650.800.65 (internal garage)13
Podium (5-over-1)0.150.650.80none (podium is separate)19
High-rise (Type I tower)0.100.300.76none (podium below)9

The hack formula reduces to one line per product:

Garden units per acre   = 11 × stories × acres × (1,000 ÷ avg_SF)
Wrap units per acre     = 13 × stories × acres × (1,000 ÷ avg_SF)
Podium units per acre   = 19 × wood-frame stories × acres × (1,000 ÷ avg_SF)
High-rise units / acre  = 9  × tower stories × acres × (1,000 ÷ avg_SF)

The last term, (1,000 ÷ avg_SF), is the unit-size correction. Smaller units pack denser inversely. At a 900 SF average, multiply by 1.11. At 1,100 SF, multiply by 0.91. At 800 SF (student or workforce), multiply by 1.25.

Worked examples for a 1-acre site at 1,000 SF average unit:

ProductStoriesHack calculationUnits
Garden (Type V-B, 3 stories)311 × 3 × 1 × 1.0033
Garden (Type V-A, 4 stories)411 × 4 × 1 × 1.0044
Wrap (Type V-A, 4 stories)413 × 4 × 1 × 1.0052
Wrap (Type III-A, 5 stories)513 × 5 × 1 × 1.0065
5-over-1 podium5 wood19 × 5 × 1 × 1.0095
5-over-2 podium5 wood19 × 5 × 1 × 1.0095
6-over-1 podium6 wood19 × 6 × 1 × 1.00114
10-story Type IV-B mass timber1019 × 10 × 1 × 1.00 (same form factor as podium)190
25-story Type I tower259 × 25 × 1 × 1.00225

Verify against the bottom-up math earlier in the article. The 5-over-1 hack returns 95 units. The bottom-up master identity returned 96 units. The 1-unit gap is rounding error. The hack and the long-form math reconcile to within one or two percent.

For a 2.5-acre site of student housing (smaller average unit, 800 SF) modeled as 5-over-1:

units = 19 × 5 × 2.5 × (1,000 / 800) = 297 units

That is the answer in your head. Sanity-check it against the master identity:

units = 43,560 × 2.5 × 0.85 × 0.65 × 5 × 0.80 / 800 = 300 units

297 versus 300. The hack works.

The hack has limits. It assumes typical setback, coverage, and efficiency for the product type, which is fine for the screening pass but wrong if the site has unusual geometry, deep zoning setbacks, or unusual coverage caps. It assumes the building can be built to the rentable-story cap, which fails if the zoning height cap binds below the construction-type cap. And it assumes no special FAR cap binds before the volumetric envelope is exhausted, which fails on sites with restrictive FAR.

Use the hack to triage. Sites that fail the hack do not need a feasibility model. Sites that pass the hack get the full bottom-up and top-down treatment. The hack is the filter. The master identity is the underwrite.

A second useful version of the hack, for converting any product to any other product on the same site, holds the ratio constant:

Wrap / Garden   = 13 / 11 ≈ 1.18×
Podium / Wrap   = 19 / 13 ≈ 1.46×
Podium / Garden = 19 / 11 ≈ 1.73×
High-rise / Podium (per floor) = 9 / 19 ≈ 0.47×

Per floor, a high-rise produces less density than a podium, because the tower footprint is smaller. The high-rise wins on density only by stacking many more floors. At equal heights (which the code does not actually allow), podium would win. The high-rise math only works because you cannot legally build a 25-story podium.

13. The Seven Levers

Each input in the master identity is a lever. Move it, and the unit count moves in a predictable direction. The sensitivities at the defaults are roughly as follows for a 5-over-1 podium at one acre.

LeverMoveEffect on units per acreWhy
Add one wood-frame story (4 to 5)+1 floor+20 percentSame footprint, one more residential floor
Add one podium story+1 floor0 percentAdds parking, no residential
Coverage (0.55 to 0.65)+10 pp+18 percentBigger footprint per floor
Coverage (0.65 to 0.75)+10 pp+15 percentBigger footprint, diminishing returns from coverage cap
Setback (0.20 to 0.15)−5 pp+6 percentMore buildable lot
Setback (0.15 to 0.10)−5 pp+6 percentSame effect, additive
Efficiency (0.75 to 0.80)+5 pp+7 percentLess waste on corridors and MEP
Efficiency (0.80 to 0.85)+5 pp+6 percentDiminishing, but real
Average unit (1,100 to 1,000)−100 SF+10 percentSmaller units pack denser
Average unit (1,000 to 900)−100 SF+11 percentSame direction, accelerating
Switch garden to wrap+1 story, structured parking+25 to 40 percentCaptures the parking off the dirt
Switch wrap to podium+1 to 2 stories of wood, podium parking+60 to 90 percentCaptures the parking under the building
Switch podium to high-rise+15 to 30 stories+120 to 200 percentCaptures vertical

The single largest lever is the construction type itself, because moving from garden to wrap to podium is changing the rentable stories input by a multiple. Within a given construction type, coverage and unit size are the two largest movable levers, with each producing 11 to 18 percent shifts at common ranges. Setback and efficiency are smaller and tend to be more constrained by code than by design choice.

The lever that produces zero is the extra podium story. People often expect adding podium height to add units, on the intuition that taller buildings have more units. The math is the opposite. The podium is parking. Adding podium height adds parking. The unit count is governed by the rentable wood-frame floors above the podium, which are constrained by the IBC height cap (Type III-A maxes at 5 stories, 6 in some jurisdictions, almost never more). Adding podium height moves the rentable wood-frame floors closer to the high-rise threshold without adding to their count.

14. The Binding Constraint Hierarchy

Every site has several caps imposed on it. The binding one is the lowest. Knowing which cap is binding is the most important judgment call in feasibility.

CapSourceWhen it binds
FARZoningWhen the parcel is small relative to the allowed envelope, or when the zoning is unusually generous
HeightZoning or IBCAlmost always binding for podium and high-rise on dense lots, since the construction type or the zoning caps height before FAR is exhausted
CoverageZoningRarely binding for podium (which wants to cover the lot), often binding for high-rise (where the tower footprint is well below the zoning cap and the podium handles coverage)
SetbackZoningBinding on small or irregular parcels where the setback eats a large share of the lot
Parking ratioZoningOften binding for suburban product, sometimes binding for urban product when the site cannot accommodate enough parking for the unit count
Open space requirementZoningBinding on dense sites where amenity, courtyard, and stormwater eat into footprint
Unit density capZoningBinding when the code expresses density as units per acre directly (rare in form-based codes, common in older suburban codes)
Egress and corridor geometryIBCRarely the binding cap, but constrains efficiency on irregular floor plates

Most urban infill sites have height or FAR binding first, with coverage close behind. Most suburban sites have parking ratio binding first, with density cap close behind. The ratio of binding constraints in any given submarket is a tell about the local political economy of housing. A city where coverage is the binding cap on every podium has effectively been planning for less density than the zoning suggests on paper.

The right way to surface the binding constraint is to run the bottom-up math against the zoning envelope, then run it against the IBC envelope, then run it against the parking minimum, and take the lowest of the three. The cap that produces the lowest unit count is the binding cap, and the entitlement strategy for the site is built around either accepting it or pushing to relax it through a variance, a planned unit development, or a rezoning.

15. Parking Math: The Hidden Constraint

Parking is the input that hides inside the coverage assumption and quietly governs the whole exercise. Every product type has a parking solution, and the parking solution drives the site plan, the structural system, and the cost stack.

ProductTypical parking ratioParking solutionSF per stall
Suburban garden1.6 to 1.8 stalls per unitSurface320 to 350
Urban-edge garden1.2 to 1.5Surface + tuck-under330 to 360
Wrap (Texas donut)1.0 to 1.4Internal Type V or III garage340 to 380
Podium (5-over-1)1.0 to 1.3Type I concrete podium340 to 380
Podium below grade0.8 to 1.2Below-grade concrete380 to 450
Urban high-rise0.5 to 1.0Multi-level podium or below-grade400 to 500
Transit-oriented0.3 to 0.7Reduced podium or shared structure380 to 450

The SF-per-stall number is the gross structured parking area divided by the count of stalls, which includes the drive aisle, ramps, columns, and circulation. The often-quoted 350 SF per stall is the right number for above-grade podium parking in 2026. Below-grade adds 30 to 70 SF per stall because of the ramps, the ventilation, and the structural reinforcement around the perimeter walls.

At 1.0 stall per unit and 350 SF per stall, every unit requires 350 SF of structured parking. A 96-unit 5-over-1 needs roughly 33,600 SF of parking, which is more than one full podium level on a 24,000 SF footprint. That is why 5-over-1 with a single podium level requires either below-grade additional parking or a lower parking ratio.

The parking ratio is increasingly the binding constraint that pushes the building toward 5-over-2 (two podium levels) or 6-over-1 with below-grade parking. The decision is mostly economic. A second above-grade podium level costs 30,000 to 50,000 dollars per stall, including the structural transfer to the wood frame above. Below-grade parking costs 50,000 to 90,000 dollars per stall, including the excavation and waterproofing. The lower of the two, against the rent the site can support, sets the parking strategy.

In urban markets that have reduced or eliminated parking minimums (Minneapolis, Buffalo, San Diego, Cambridge, Berkeley, and a growing list of others), the math frees up a story of building or a level of below-grade for residential or amenity. That single regulatory change can shift the density on a one-acre site by 20 to 30 percent.

16. Three Products, Same Acre, Three Answers

The most useful exercise in early feasibility is to model the same site under all three product types and see where the answers land. The same one-acre lot, the same buildable footprint, three different vertical stacks.

VariableGarden (Type V-A, 4 stories)Wrap (Type III-A, 5 stories with internal garage)Podium (5-over-1)
Buildable lot SF37,02638,33337,026
Coverage35 percent65 percent65 percent
Footprint SF12,95924,91724,067
Rentable stories45 (less internal garage)5
Gross residential SF51,83680,980120,335
Efficiency85 percent80 percent80 percent
Rentable SF44,06164,78496,268
Average unit SF1,0001,0001,000
Units446596
Density (units per acre)446596
Parking (stalls at 1.3 per unit)57 surface85 internal125 podium
Hard cost per GSF (2026)$200$260$310
Approx total hard cost$10.4M$21.1M$37.3M
Approx cost per unit$236K$324K$389K

Three different products. Three different densities. Three different cost basis levels. The choice between them is a question of land cost and rent. If the land basis is below 25 dollars per buildable square foot and rent supports 1.80 dollars per SF per month, garden pencils. Between 25 and 60 dollars per buildable SF with rent at 2.20 to 2.80 dollars per SF per month, wrap pencils. Above 60 dollars per buildable SF and rent above 2.80 dollars per SF per month, podium pencils.

The exercise of running the three side by side is what reveals which product the site can actually support. A site that can support a podium but is modeled as a garden is leaving 60 units per acre on the table. A site that can only support a garden but is modeled as a podium is overpaying for the land. The first error shows up as foregone density. The second shows up as a failed deal.

17. Lot Size Sensitivity

The same identity scales linearly with acreage, but the underlying inputs do not. Coverage is roughly constant within a product type, but parking efficiency, common amenity, and structural module size are not.

Site sizeGarden densityWrap densityPodium densityWhy
0.5 acre30 to 40 units per acreNot feasible85 to 105 units per acreWrap requires a minimum 1.2 acre footprint for the garage module
1.0 acre40 to 5055 to 70100 to 115Default model assumption
2.0 acres38 to 4860 to 75105 to 120Slight scale efficiency on shared amenity
5.0 acres30 to 4255 to 7095 to 110Larger sites lose density to circulation and amenity scale
10.0+ acres25 to 3850 to 6580 to 100Master-planned, internal roads and parks consume coverage

The non-linearity comes from two effects. Below about 1.2 acres, the structural module of structured parking does not fit efficiently, and wrap or podium suffers. Above about 3 acres, the site starts to need internal circulation (private drives, fire lanes, looped access), which eats into coverage at the site-plan level. The sweet spot for both wrap and podium is 1.2 to 2.5 acres, where the building is large enough to amortize the structured parking efficiently but small enough that the site does not need internal road infrastructure.

For garden product, the density peaks at smaller sites (0.7 to 1.5 acres) and erodes on larger sites because of the surface parking footprint and the perimeter setback requirements that scale with site size.

18. Common Modeling Mistakes

The same handful of errors show up in early feasibility models across thousands of deals. Knowing which they are saves a quarter of work on bad sites.

Counting podium floors as rentable. The most common error. A 7-story 5-over-2 has 5 rentable floors, not 7. The arithmetic produces 40 percent too many units if all 7 are counted. The fix is to label the wood-frame stories explicitly in the model and treat them as the rentable input.

Setbacks expressed as a single percentage on the wrong lot shape. Setbacks are a linear-foot specification, not a percentage. The percentage equivalent depends on the lot geometry, and the same setback table produces vastly different losses on a square lot versus a long narrow lot of the same acreage. A 1-acre square lot loses 15 to 23 percent to typical urban setbacks. A 1-acre lot at 50 ft wide loses 40 percent or more. The percentage approximation is fine for screening, but only if the inputs reflect the actual lot dimensions.

Ignoring the high-rise threshold. Going from 74 ft to 80 ft of total building height triggers Chapter 4 of the IBC and adds 20 to 40 dollars per gross square foot in life-safety scope. The cost increment falls on the entire building, not just the marginal floor. A model that adds a sixth wood-frame story without flagging this is missing 3 to 6 million dollars of cost on a 100-unit project.

Treating FAR as the binding constraint when height is. FAR caps look generous on paper but rarely bind in urban infill. The height cap, the construction-type story limit, and the parking ratio all bind before the FAR exhausts. A model that solves for the FAR envelope and assumes it is achievable will overstate density by 15 to 30 percent.

Pulling 1,000 SF per unit without modeling the unit mix. The 1,000 SF average is a useful default but only valid if the actual unit mix produces it. A mix that runs 25 percent studios at 525 SF, 50 percent one-bedrooms at 750 SF, 20 percent two-bedrooms at 1,100 SF, and 5 percent three-bedrooms at 1,400 SF produces a weighted average of 838 SF, not 1,000. A mix that runs 10 percent studios, 35 percent one-bedrooms, 45 percent two-bedrooms, and 10 percent three-bedrooms produces 982 SF, close to the default. A mix skewed to family product (5 percent studios, 20 percent one-bedrooms, 50 percent two-bedrooms, 25 percent three-bedrooms) produces 1,128 SF. The unit-count answer moves by 25 to 30 percent on the same site depending on the actual mix. Always derive the average from a real mix.

Ignoring the parking footprint in coverage. Suburban garden models that use 60 to 65 percent coverage are overstating the building by 30 to 50 percent, because the surface parking is not the building. The correct coverage for garden product is 30 to 45 percent. The remainder is parking, drive aisle, and amenity, not building.

Counting amenity floors as rentable. A rooftop amenity, a ground-floor leasing office, a fitness center on level two of the podium, all count as gross building area but produce no rentable square footage. A model that treats every floor as residential is overstating rentable SF by the amenity share, which is typically 3 to 7 percent of total GSF for podium and 5 to 10 percent for high-rise.

Pulling efficiency from the wrong product type. Garden efficiency is 85 percent (no elevator, short corridors). Wrap is 80 percent (corridor wraps an internal garage). Podium is 80 percent (central corridor, elevator core, MEP shafts). High-rise is 76 to 78 percent (more elevators, larger core). Plugging a podium efficiency into a high-rise model overstates units by 3 to 5 percent.

19. The Skill That Compounds

The math above takes about ten minutes to do for any given site once the inputs are at hand. The skill is not the arithmetic. The skill is knowing which inputs to pull for the site you are looking at, and which constraint is going to bind.

A developer who can walk a parcel, pull up the zoning code on their phone, and have a defensible unit-count range in their head before they leave the property is operating at the level where deal flow becomes a function of pattern recognition rather than process. They can sort a list of fifty sites in an afternoon and know which two are worth a feasibility study. The other forty-eight die fast and cheap, which is what good underwriting actually looks like.

Everything downstream of the unit count, the rent stack, the cost stack, the debt service, the cap rate, the yield-on-cost, the IRR, depends on the unit count being right. Get the unit count wrong by 20 percent at the top of the funnel, and every subsequent number is wrong by at least that much, and the error compounds through the model. Get the unit count right, and the model has a shot at telling you the truth.

The construction type is the lever that most affects the unit count, and the lever most often picked badly. Garden where podium would have worked is foregone yield. Podium where garden would have worked is overpriced land. Mass timber or high-rise picked too aggressively for the rent is a deal that does not finance. The math is not what is hard. Knowing the inputs is.

Build the bottom-up model from scratch on a new site every time. Run the FAR top-down as a cross-check. Find the binding constraint. Move on to the next site if it does not pencil. Move into entitlements if it does. The compounding shows up over hundreds of sites, not on any single one. That is the discipline.