Cost segregation reclassifies portions of a building's depreciable basis from the default 27.5-year (residential) or 39-year (commercial) recovery period into 5-, 7-, and 15-year buckets — fully deductible Year-1 under 100% bonus depreciation. Done correctly, the analysis is engineered: every component is identified, costed using a defensible cost basis, and classified per IRS rules.
1. Component identification
We start with the building's component library — appliances, flooring, cabinetry, HVAC, plumbing fixtures, electrical, structural shell, site work — built up from RSMeans 2024 base $/SF cost data. The library is property-type specific (residential vs. commercial vs. STR vs. industrial) so the starting cost mix reflects what's actually present.
2. Geographic + quality adjustment
RSMeans regional cost multipliers calibrate the costs to your specific market — Phoenix construction differs from Boston construction. Quality and finish-level multipliers further adjust for premium vs. economy build. We use a 6-tier geographic index: pinned metros (5), calibrated metros (JSON-driven), manual cities (20), state-level (50), regional (4), and national fallback.
3. MACRS classification
Each component is classified to its IRS-prescribed recovery period: 5-yr (personal property like appliances and removable fixtures), 7-yr (some specialty equipment), 15-yr (land improvements like fencing, landscaping, paving), or 27.5/39-yr (the building shell). The classification follows the IRS Audit Techniques Guide for cost segregation studies.
4. Reconciliation
The unadjusted component costs rarely sum to exactly your purchase price — quality and intensity vary. We reconcile by computing a scaling factor S = adjusted_basis / total_unadjusted, applied uniformly so the final bucket sums match your basis to the penny.
5. Quality control
Sixteen QC checks gate every report: PASS (ship), REVIEW (engineer review before ship), or FAIL (block). We classify flags by reason family — hard invariant violations, market regime outliers, calibration outliers, input quality, and narrative safety — and downgrade single soft flags when other metrics look healthy.
6. Data sources & remote observation
No site visit is required. We pull data from county assessor records (for land allocation), satellite imagery (for site work), property APIs (RentCast for property characteristics), OSM (for building type and landuse), and BLS PPI (for cost-index time adjustment). Each data source has documented reliability gates.