Parking Ratios and Demand Models: Picking the Right Number

Every parking count starts as a ratio: spaces per thousand square feet, per unit, per room, per seat. The ratio looks like arithmetic and behaves like judgment, because every term in it is a choice, and the manuscript's consistent instruction is that no development should automatically follow the prescribed zoning ratio: the ordinance is a floor or ceiling set for a generic case, and the actual number belongs to the specific project, its tenants, its hours, and its market.

Get the denominator right first. Ratios attach to square footage, and square footage comes in four flavors that differ by thousands of feet on the same building: Gross Floor Area (all floors, measured to the outside of exterior walls, the usual basis for single-tenant buildings), Gross Leasable Area (GFA minus the shared spaces: corridors, lobbies, restrooms, cores, the usual basis for multi-tenant projects), Net Floor Area (measured inside the exterior walls), and Net Rentable Area (the leasable portion of NFA). A ratio quoted without its denominator type is not yet a number, and mixing bases across a comparison is the cheapest way to be confidently wrong. Residential runs on its own denominators: unit counts by bedroom type and by market-rate versus affordable status, which is where geography enters loudly: the small town mandating one space per one-bedroom coexists with transit-rich cities running 0.25 to 0.75 per one-bedroom, and neither number transfers.

The numerator is people, not floor. Two projects with identical floor area can need wildly different parking because the humans differ. The manuscript's example deserves memorizing: an event center might staff 80 employees in 2,000 square feet while a high-end salon in the same footprint runs 30 stylists, a 50-space swing on identical area. The same logic runs through customers: a medical office building conventionally rates 4.0 to 4.5 spaces per 1,000 square feet, but a specialty tenant seeing one patient per 1,000 square feet needs a fraction of that, and the ratio must follow the business model, not the building type label. Ratio by riders adds the vehicle-occupancy dimension: the downtown restaurant runs 1.5 to 3 passengers per car at lunch (colleagues carpooling from the office) and 1.0 in the evening (everyone wants their own exit), so the same seats generate more evening parking demand than midday. None of these figures is universal; each demands tailoring to venue type, location, local culture, and the real prevalence of alternative modes.

Model by parker type. The full demand model decomposes the population before it multiplies anything: employees (on-site and off, contract and not), visitors (event, medical, retail), residents (by unit type and affordability), plus the categorical layers (accessible spaces, valet, EV charging) and the low-frequency, high-consequence users: delivery vehicles, shuttles, service and emergency vehicles, car-share stations, ride-share stops. Each type carries its own ratio, schedule, and dwell pattern, and the composite of those decomposed demands, not a single blended ratio, is what the shared-parking machinery of Module 3 then overlays in time.

Ratios drift. The most dangerous property of a well-chosen ratio is that its inputs change tenancy by tenancy. The manuscript's medical-office case runs both directions: the complex whose MOB houses the one-patient-per-1,000-feet specialist shares happily with the 65-room hotel peaking evenings and weekends, until the MOB sells to a conventional tenant and the arithmetic collapses; conversely the HMO that densifies services amplifies demand past the original count. A ratio is an assumption with a shelf life, and counts built near the margin need the tenancy-change scenario run before the investor, not after.

Source crosswalk -- Module 4