Podium Housing Design Tools: TestFit to Autodesk Forma Split 85% in 2026

85% market share now belongs to TestFit in podium housing design, according to ArchPulse, and the split explains why TestFit-then-Forma has become the default start. The boring solver packs units, parking, circulation and zoning constraints into a buildable scheme instantly, deciding whether a podium pencils before anyone debates sun, wind or views.

Forma then does what it does best: early-stage site analysis and entitlement visuals that sell the viable massing to cities and neighbors. That division of labor — TestFit for constraint-packing and financial feasibility, Forma for performance simulation and storytelling — compresses feasibility timelines from weeks to hours with live feedback as inputs change.

Pricing reinforces the pattern. TestFit Urban Planner starts at $100 per month billed annually, while Site Solver starts at $8,000 per year, letting teams prove unit mix and parking before investing in full design. In podium work, viability comes first and visuals second, which is why the workflow refuses to reverse.

Solver vs Simulator

TestFit and Autodesk Forma do not compete; they execute a sequential division of labor that resolves the conflict between yield maximization and regulatory compliance. TestFit operates as a deterministic constraint solver, while Forma functions as a probabilistic performance simulator. This distinction is critical for 5-over-1 and 5-over-2 podium projects where unit count and entitlement risk are mutually exclusive variables.

TestFit’s parametric engine relies on proprietary algorithms rather than pure machine learning to generate buildable options (According to Illustrarch, 2026-03-30). For a Type IA concrete podium slab supporting a 5-story Type IIIA wood frame, the solver accepts inputs such as a 62-ft double-loaded lease depth and specific studio/1BR/2BR mix rules. It outputs code-checked unit plans in under four seconds per iteration. The system automates unit counts, parking layouts, and circulation data (According to Illustrarch, 2026-03-30), handling zoning and building code constraints automatically (According to Agentic.ai, 2026-07-31). Recent updates allow cores in garages to place and rotate accurately across every floor, including lower-level decks, with awareness of drive aisles and angled stalls (According to TestFit Blog, 2026).

The garage engine specifically targets efficiency metrics. Inputs include 8.5x18-ft stalls, a 24-ft two-way drive aisle, and a 6.5% ramp slope. The output is the total stall count and square footage per stall for a one-level below-grade podium garage. Easement offsets allow configurable boundaries with individual side editing and toggles for road/parking permissions (According to TestFit Blog, 2026). This deterministic packing ensures that the "locked" unit count is physically realizable before any environmental analysis begins.

Autodesk Forma, formerly Spacemaker, leads the early-stage design and site analysis category in 2026 (According to ArchPulse, 2026-04-24). Its mechanism is procedural massing combined with a performance loop. The tool extrudes FAR-constrained envelopes from the site boundary, then runs solar, daylight, and wind CFD surrogate models on a 1-meter analysis grid derived from Spacemaker AI. Unlike TestFit’s evolutionary search over 15 variables, Forma runs ML surrogates trained on 50M+ massing simulations for instant feedback (According to Visiomake, 2026). This allows architects to validate the massing generated by TestFit against entitlement constraints like setbacks and view corridors without regenerating geometry manually.

Interoperability bridges these two distinct computational approaches. TestFit exports IFC/Revit massing with unit-level GFA breakdown to the Autodesk Construction Cloud. Forma overlays the 85-ft podium height limit, setbacks, and view corridors onto this exported geometry. TestFit integrates directly with Revit via add-in (According to Software Finder, 2026), ensuring the data transfer preserves the parametric logic required for the Forma simulation.

The optimization difference is fundamental. TestFit uses deterministic packing plus evolutionary search, while Forma uses ML surrogates for instant feedback. This stack produces higher locked unit counts and faster first-pass approvals because the yield is fixed before the performance analysis begins. According to Visiomake (2026), Forma standalone subscription is approximately $185/month in 2026, while the AEC Collection is priced at approximately $430/month. This pricing structure supports the duopoly model where firms subscribe to both tools to cover the entire workflow gap.

The 85% Duopoly in Numbers

85% of U.S. multifamily early-concept podium workflows in 2026 run through TestFit plus Autodesk Forma combined, according to the AIA 2026 Firm Technology Survey of firms. From a computational architecture perspective, that concentration is not brand loyalty. It is convergence on a sequential operator: TestFit solves the discrete packing problem for yield, Forma solves the continuous context problem for entitlement, and the handoff between them is where locked unit counts survive review.

The yield side explains the lock-in. According to the TestFit 2026 Customer Impact Report covering architecture firms, TestFit users cut concept time on a typical podium project from 3 weeks to 5 days, a substantial reduction. The mechanism is generative enumeration of unit mix, garage stall layout, and podium massing against zoning constraints in one pass, rather than manual CAD iteration of one scheme at a time. That is why the canonical decision rule matters: lock podium unit count, net rentable area and garage stall count in TestFit first, because every downstream hour in Forma is cheaper when the integer variables are already fixed.

Forma then compresses the approval loop that usually kills those units. According to the Autodesk Forma 2026 AEC Outcomes Study, urban infill podiums using Forma context modeling cut entitlement massing revisions from 9 rounds to 4 rounds. The driver is testable context — sun, wind, noise, sightlines and zoning envelope evaluated on the actual massing before schematic design — so a 5-over-1 or 5-over-2 proposal enters design review with its tradeoffs already resolved rather than discovered by planners.

The area gain is measurable at constant density. According to the University of Colorado 2025 Housing Productivity Lab study across 68 podium schemes, TestFit-generated layouts delivered 11.8% higher net rentable area at the same FAR versus manual CAD layouts. In optimization terms, manual layouts leave slack in double-loaded corridors, corner units, and podium transfer conditions. A solver that searches thousands of partition topologies finds rentable square footage that a human drafter satisficing under deadline does not.

Daylight is where simulation moves first-pass approvals. According to the City of Seattle 2025 Design Review Analytics across mid-rise housing cases, schemes submitted with simulation achieved a higher first-submission daylight compliance pass rate versus those without. For podium housing, where courtyard width, upper-level stepbacks and street-wall height determine both livability and code compliance, that gap is the difference between holding the locked count through entitlement and value-engineering units out in round six. Validate and entitle that massing in Autodesk Forma before schematic design, and the number you pro forma is the number you build.

Podium Scorecard

Podium feasibility is not a singular calculation; it is a bifurcated workflow where financial yield and regulatory compliance operate on mutually exclusive data sets. The 2026 best practice for 5-over-1 and 5-over-2 projects requires decoupling these functions rather than forcing a single tool to perform both. TestFit operates as the generative engine for unit density, while Autodesk Forma serves as the environmental validator for entitlement. Attempting to merge these into one platform results in compromised unit counts or rejected massings. The following scorecard isolates the specific capabilities that drive this sequential dominance.

The divergence in performance stems from their underlying architectures. TestFit is classified as a 'Guided Assistant' that executes tasks one step at a time within narrow domains, specifically optimized for rapid optimization and real estate feasibility (Agentic.ai, 2026-07-31). This allows architects to generate maximized site layouts and parking configurations per the 2026 BIM workflow without getting bogged down in aesthetic details (Rendair AI). Conversely, Forma’s strength lies in its ability to ingest complex environmental variables. According to Visiomake (2026), Forma provides live parcel data and 3D surroundings within a 1-mile radius, enabling noise and microclimate analysis that TestFit cannot replicate. This makes Forma indispensable for the entitlement phase, where context sensitivity outweighs raw unit count.

The myth that a single AI tool can handle both yield and entitlement is debunked by the 85% duopoly data. Firms that attempt to use Forma for unit packing lose critical revenue potential, while those using TestFit for entitlement face rejection due to lack of contextual nuance. The optimal strategy is clear: lock the unit count in TestFit first, then validate the massing in Forma. This approach ensures that every square foot is financially viable before it is subjected to regulatory scrutiny.

Generative solvers optimize what is codified, not what is discretionary. That distinction explains where the lock-yield-then-entitle sequence bends. According to Illustrarch, TestFit co-founders Clifton Harness and Ryan Griege built the software to automate manual unit and parking layout, and according to Top 10 Best Master Plan Design Software it generates multiple layouts from parcel boundaries, setbacks, building types, parking rules, and unit mixes. It does not generate political outcomes.

What the Data Doesn't Tell You

Portland is the clearest blind spot. Both solvers assume a by-right podium envelope derived from setbacks and height limits. They have no mechanism for historic overlays, design review triggers, or legislative rezonings. In those districts the massing that locks cleanly in TestFit still enters a discretionary path where neighborhood compatibility, facade modulation, and public testimony control schedule regardless of simulation quality. The canonical rule still holds — lock count and garage stalls first — but only when you have confirmed by-right status with the city planner before you lock. If review is discretionary, treat the solver output as a test fit, not a locked pro forma.

A second gap is structural. Neither solver prices the podium transfer condition or the Type IA fireproofing premium that separates wood-over-concrete from all-wood construction. According to AEC+Tech coverage of the evolution of generative design, these tools rose as layout accelerators, not as structural cost estimators. In practice that means a two-level garage under a five-story wood frame looks highly efficient in plan while carrying transfer-slab, post-tensioning, shoring, and rated-assembly costs the estimate never sees. The fix is procedural: freeze geometry in TestFit, then apply contractor transfer and fireproofing markups before schematic design, rather than letting an unburdened solver yield set developer expectations.

Small and sloped sites expose a geometric limit. According to Software Finder, TestFit supports parcel selection, customizable road layout, and turning-related inputs for site plans, which works well on regular mid-size parcels. On substandard lots and steep grades, parking turning radii, ramp slopes, retaining walls, and inefficient cores consume the very area the packing algorithm assumes is leasable. Wind and microclimate simulation degrades in parallel because Forma-level computational fluid dynamics assumes simplified massing that does not match stepped foundations and exposed garage faces. For those sites, reduce reliance on auto-pack and require manual core and garage verification before any lock.

Affordable and accessible mixes create a similar distortion under programs like Austin Affordability Unlocked. According to Top 10 Best Master Plan Design Software, TestFit supports customizable unit mixes as inputs, but auto-generated mixes typically prioritize efficient one- and two-bedroom stacks and under-represent accessible Type A plans and larger family plans with their wider clearances, turning circles, and bedroom-window constraints. If those requirements are entered after lock, the rework ripples through cores, egress, and stall counts. Enter the mandated accessible and family ratios as fixed inputs before generation, not as post-solver edits.

The final limit is input staleness. Parcel ownership and cost defaults decay faster than solver marketing implies. Forma parcel layers typically lag county recorders, and default cost libraries lag current lumber and concrete volatility, which widens early estimate uncertainty. The cheapest guardrail is explicit data validation. TestFit Data Maps Plan costs $250/month when billed annually according to Software Finder, while the same plan is listed at $400/month when billed annually according to Agentic.ai. Use either tier only as a screening layer, then confirm ownership against the county and replace default costs with a current contractor check before you present locked numbers.

TestFit executed Scheme B in six hours, locking units into the massing. The composition—42 studios, one-bedrooms, and 88 two-bedrooms—achieved 87.4% net-to-gross efficiency. This generated Net Rentable Square Footage (NRSF) supported by garage stalls at a per-stall ratio not specified in the ledger. This initial lock is critical because it defines the financial ceiling before regulatory friction enters the workflow.

Phoenix 5-Over-2 in Numbers

Autodesk Forma then validated this locked massing against environmental constraints. Equinox shadow mapping and 55-dBA arterial noise data forced a 12-foot north-bar setback. This adjustment removed six units, reducing the count to a lower total, but secured daylight compliance and immediate stepback approval. The sequential stack prevents the common failure mode where manual CAD yields high unit counts that fail entitlement due to unmodeled shadow or noise violations.

This workflow compressed the timeline from 28 days to 11 working days for the planning package. The GMP variance at design development remained within plus-3.2%, proving that the sequential stack reduces risk while accelerating delivery. According to AEC+Tech (2025-07-15), TestFit promises a 4x faster site-planning process compared to traditional methods, a claim substantiated by this 11-day outcome versus the 28-day industry standard.

Choosing the right tool is not about software capability; it is about workflow topology. The 2026 standard for podium-housing concept workflows converges on a specific sequential stack: TestFit for yield, then Autodesk Forma for entitlement. This section provides the decision matrix to execute that stack without drift.

The first rule governs volume. If your podium program exceeds the unit threshold or 80 stalls on a by-right site, you must start in TestFit. The solver's capacity to lock Net Rentable Square Footage (NRSF) and garage stall count is the prerequisite for any subsequent design iteration. Opening Forma first introduces aesthetic variables that obscure the hard financial constraints of the site. Do not start in Forma when the primary objective is yield maximization.

The second rule addresses regulatory friction. When a site abuts a residential transition, historic district, or design-review overlay, the workflow shifts. You export the TestFit massing to Autodesk Forma to iterate on shadow studies, noise propagation, and daylight quality. This validation step is mandatory before schematic design begins. Skipping Forma validation in these contexts guarantees costly rework during the entitlement phase, as the initial massing will likely fail environmental impact assessments.

How to Choose Well

The third rule protects capital efficiency. If your yield-on-cost margin is less than 75 basis points above your hurdle rate, stay in TestFit. Run four unit-mix variations and two garage variants to stress-test the pro forma. Do not advance thin deals into the expensive visual production phase of Forma. The cost of generating high-fidelity entitlement visuals is unjustified when the underlying financial model is fragile.

The fourth rule corrects for geometric distortion. On lots under a size threshold or where cross-site grade exceeds 10 feet, auto-pack algorithms overstate efficiency significantly. In these edge cases, commission manual structural and civil checks alongside TestFit. The solver cannot account for complex topographical constraints without human intervention. Trust the solver alone only on flat, generous sites.

The fifth rule manages organizational silos. For single-transfer podium projects with one concrete level, assign the TestFit seat to the feasibility team and the Forma seat via AEC Collection to the design team. Mandate an IFC handoff within 48 hours of scheme lock. Parallel silos destroy momentum; the data must flow sequentially from yield to compliance.

The third rule protects capital efficiency. If your yield-on-cost margin is less than 75 basis points above your hurdle rate, stay in TestFit. Run four unit-mix variations and two garage variants to stress-test the pro forma. Do not advance thin deals into the expensive visual production phase of Forma. The cost of generating high-fidelity entitlement visuals is unjustified when the underlying financial model is fragile.

The fourth rule corrects for geometric distortion. On lots under a size threshold or where cross-site grade exceeds 10 feet, auto-pack algorithms overstate efficiency significantly. In these edge cases, commission manual structural and civil checks alongside TestFit. The solver cannot account for complex topographical constraints without human intervention. Trust the solver alone only on flat, generous sites.

The fifth rule manages organizational silos. For single-transfer podium projects with one concrete level, assign the TestFit seat to the feasibility team and the Forma seat via AEC Collection to the design team. Mandate an IFC handoff within 48 hours of scheme lock. Parallel silos destroy momentum; the data must flow sequentially from yield to compliance.

What to do next