In Toronto, a building called 1 St. Clair West received a conversion score of 81% from Gensler's Conversions+ algorithm, an automated feasibility tool that has now assessed more than 1,300 office buildings across 130 North American cities. Eighty-one percent sounds definitive, like the building is ready. What the score actually measured was structural grid spacing, floor-to-floor height, floor plate geometry, location desirability, and zoning compatibility. It did not measure whether someone lying in bed at 7 a.m. on a Saturday would see morning light moving across the wall, or whether the kitchen would feel like a kitchen instead of a corridor with appliances.
That distinction matters more than the algorithm acknowledges.
National office vacancy hit 18.2% in late 2025, up 190 basis points year-over-year, according to CommercialEdge. Cities from Calgary to San Francisco are offering unprecedented incentive packages to push conversion projects forward: tax abatements lasting up to 25 years in Boston, transfer tax waivers in San Francisco, zoning modifications that waive dwelling unit exposure requirements and parking minimums. San Francisco's program even allows a 33% expansion of building volume beyond the existing envelope. All of which assumes the fundamental question has been answered. Can an office become a home?
What the Algorithm Measures, and What It Cannot
Gensler developed Conversions+ in 2020, originally road-testing it by scoring 6 million square feet of downtown Calgary. Steven Paynter, who led the research team, described the tool's value in terms of speed: what once required months of architectural feasibility study could be completed in minutes. Since then, the tool has supported assessments for over 130 cities and hundreds of building owners. Its core finding is that roughly 25% of assessed buildings score as suitable conversion candidates, a figure that aligns with a separate Capital Innovations analysis estimating 15% suitability across the 105 largest U.S. metro areas, potentially yielding 171,470 housing units.
Paynter's most quotable insight is that "all the features that result in an unpleasant office make for an excellent multifamily product." A Class C office with 12-foot floor-to-floor height produces oppressive 8-foot ceilings once ducts and suspended lighting are installed. Strip those systems out and you get 11-foot clear residential ceilings, a dimension that luxury developers would ordinarily pay dearly to achieve in new construction. Gensler claims conversion can cost 30% less than building new.
Compelling economics, but economics is not architecture.
Sixty Feet from the Window Wall to the Core
A typical office building constructed between 1960 and 1990 has a floor plate depth of 50 to 70 feet, measured from the exterior window wall to the central service core containing elevators, stairs, and mechanical shafts. In office use, this depth is unremarkable. Fluorescent fixtures provide uniform illumination at roughly 50 foot-candles across the entire plate, cubicle partitions rarely exceed five feet, and nobody expects to see the sky from their desk. An office worker tolerates artificial light for eight hours because the compensation is a paycheck, not a life.
Residential occupants have different expectations, and building codes enforce them. IRC Section R303.1 requires that every habitable room in a dwelling unit contain glazed area equal to at least 8% of the room's floor area, with that glazing opening directly to the outdoors. For a modest 150-square-foot bedroom, the code demands 12 square feet of window. Not borrowed light through a transom or a light tube, but window glass that faces outside.
Now consider a residential unit carved from a 60-foot-deep floor plate. Apartments lining the perimeter get windows. Rooms at 30 feet from the facade do not, unless the developer carves a light well or atrium through the building, sacrificing rentable floor area to create an interior void that admits daylight from above. Every square foot of light well is revenue permanently removed from the pro forma, not just construction cost but perpetual lost rent across the building's economic life.
I ran a rough calculation using typical Manhattan conversion economics. Assume a 25,000-square-foot floor plate with a 60-foot depth. If you cut a 1,500-square-foot atrium to bring light to interior units, you lose approximately $75,000 per year in rental income at $50 per square foot, compounding across every floor the atrium penetrates. A 20-story building with a full-height atrium sacrifices $1.5 million in annual rent. Against Gensler's claim of 30% construction savings over new build, the math tightens considerably once you account for the floor area you had to destroy to make the remaining space habitable.
55 Broad and the Buildings That Convert Well
Not every conversion produces dim apartments. 55 Broad Street in lower Manhattan, a 36-story former financial district tower, opened in 2025 as 571 luxury rental units and became the first office-to-residential adaptation to earn LEED certification. CetraRuddy, the project architect, designed most units with direct views of the river and the Statue of Liberty. Metro Loft and Silverstein Properties delivered a 25,000-square-foot amenity collection anchored by a rooftop pool.
55 Broad worked because its floor plate cooperated. Narrower plates, with shallower depths from facade to core, allow perimeter units to wrap the building with each room touching an exterior wall. Not every office tower has this geometry, and most don't. And the buildings that cities most desperately want to convert, the Class B and C towers from the postwar boom with deep, boxy plates and minimal fenestration, are precisely the ones where the daylighting problem is most severe.
Calgary's conversion program illustrates the selection bias. Five projects are under construction and ten more are in development, collectively adding 24% to downtown's residential unit count. But those 15 projects came from an initial assessment of 6 million square feet, and Gensler's own data says only a quarter of assessed buildings pass. Calgary is converting its best candidates first, which makes the program look successful, which makes other cities assume all their vacant offices are candidates too. They are not.
AI Scores the Bones, Not the Dwelling
A 2025 study published in AI EDAM by Tuğçe Çelik at Ostim Technical University tested AI-generated floor plans across five climate zones using daylight simulation software. Of 31 plans generated by ChatGPT, Copilot, and LookX, only eight were architecturally legible enough to reconstruct in AutoCAD and simulate. Her conclusion was blunt: "None of the models consistently integrated solar orientation or seasonal lighting considerations." A gap persists between what AI can generate as a plausible-looking layout and what actually performs as a space where light, air, and human rhythm intersect.
Conversion scoring tools inherit the same blind spot at building scale. Conversions+ evaluates whether an office's structural grid can accommodate residential unit plans, whether the floor-to-floor height permits proper ceiling heights with mechanical distribution, and whether the location supports residential demand. These are necessary conditions, but they are not sufficient ones. A building that scores 81% and has a 65-foot-deep plate will produce apartments where half the bedrooms face interior corridors unless someone intervenes with design imagination that no algorithm currently possesses.
When the Score and the Space Disagree
I've walked converted office apartments in Philadelphia, in Washington, D.C., and in downtown Los Angeles. Several shared the same quality: impressive ceiling heights, often 10 or 11 feet, with the raw concrete and exposed ductwork that signals "authentic" to a certain market segment. Living rooms flooded with light from the original curtain wall fenestration. And then the bedrooms, tucked behind the kitchen, windowless or fitted with a narrow slit that technically satisfied code but admitted roughly the same light as a horizontal mail slot.
People rented them anyway, because housing markets are desperate and a dim bedroom beats no bedroom at all. That is the strongest argument for imperfect conversions, and it deserves its full weight: with a national housing shortage measured in millions of units, and office vacancy approaching one in five buildings in some downtowns, converting mediocre offices into mediocre apartments produces housing that would otherwise not exist. Perfection is not the standard when the alternative is an empty tower generating zero tax revenue and zero housing units while its owner defaults on the mortgage.
But "better than nothing" should not be the design standard for a housing program that cities are spending hundreds of millions in tax incentives to subsidize. San Francisco's adaptive reuse program waives dwelling unit exposure requirements for qualifying projects, meaning converted apartments can legally have habitable rooms without code-compliant natural light, exactly the provision that was supposed to prevent the problem. When the incentive program explicitly relaxes the quality standards that protect residents, the algorithm's high feasibility score becomes self-fulfilling in the worst way: the building converts because you've lowered the bar until it steps over.
What Developers, Buyers, and Cities Should Do
If you're a developer evaluating a conversion: Run the algorithmic feasibility score, then walk the floor plate yourself at noon on a cloudy day with every artificial light turned off. If you can't read a newspaper 25 feet from the nearest window, every residential unit touching that zone will require either a light well (sacrificing $50+ per square foot annually in lost rent per floor) or a code waiver that future tenants will notice every morning. Factor the light well cost into your pro forma before you celebrate the 30% construction savings.
If you're shopping for a converted apartment: Ask which rooms face exterior walls and which face interior corridors, light wells, or other units. Visit at different times of day. A listing photo shot with every light blazing and the exposure cranked up will not tell you what the bedroom looks like at sunrise. IRC R303.1 requires 8% glazed area per habitable room, but code compliance is a floor, not a ceiling, and some municipalities have waived even that floor for conversion projects. Ask whether any dwelling unit exposure waivers apply to your building.
If you're a city official writing adaptive reuse policy: Requiring a minimum conversion score from a tool like Conversions+ is a reasonable first filter. Waiving dwelling unit exposure standards is not a reasonable second step. Daylighting requirements exist because decades of public health research, including a widely cited 2014 study in Building and Environment linking daylit workspaces to 46 minutes more sleep per night, established that people living in perpetually dim spaces experience higher rates of depression, worse sleep quality, and lower satisfaction with their homes. Subsidizing conversions that ignore these findings creates a housing stock that solves a shortage metric while producing units that erode quality of life over decades. Require a daylighting simulation for every conversion project, not just a structural feasibility score, and publish the results alongside the permit application.
What This Analysis Didn't Prove
Gensler's Conversions+ algorithm is proprietary, and I have not reviewed its full scoring criteria. It may include daylighting-adjacent factors that the public-facing descriptions do not emphasize. My criticism is directed at the public framing of these tools and the policy decisions built on top of their scores, not at the tool's internal sophistication, which I cannot evaluate.
Cost estimates for light wells and atriums vary enormously by market, building vintage, and structural system. My $75,000-per-floor lost-rent figure assumes Manhattan Class A rental rates and a specific atrium footprint. In a secondary market at $20 per square foot, the same light well costs $30,000 per floor in forgone rent, making the tradeoff more palatable. Conversion economics are intensely local, and national averages obscure more than they reveal.
No controlled study has compared resident satisfaction in converted offices against purpose-built residential at the same price point, controlling for location. It is possible that ceiling height, open layouts, and urban proximity compensate for reduced natural light in ways that surveys would capture. Without that data, the claim that converted apartments feel worse to live in remains architecturally intuitive but empirically unproven.