The cost segregation study, start to finish

A cost segregation study moves through six stages: a feasibility screen you can run yourself in ten minutes, provider selection and scoping, engineering fieldwork, a written report built to the IRS's own quality specification, filing (on the current return or via Form 3115 for older buildings), and then years of use — depreciation schedules, partial dispositions, and recapture tracking at exit. From engagement letter to final report typically takes four to eight weeks and costs $5,000 to $20,000 for a single property.

This guide walks the whole arc from the owner's chair. What a study is and where it comes from legally is covered in what a cost segregation study is; this piece is about commissioning one well and using it correctly. Throughout, we carry a single worked example: a $3.2 million office/flex building purchased in March 2026.

Stage 1: Is a study even worth pricing?

Before calling providers, run the arithmetic. Four inputs decide the question.

Depreciable basis. Purchase price minus land. The benefit scales with basis; the fee mostly does not. Below roughly $500,000 of depreciable basis, fees eat the present value; between $500,000 and $1 million it depends on building type; above $1 million a study on any equipment- or site-intensive property usually pencils comfortably.

Marginal tax rate. A deduction is worth its face amount times the rate. A 37 percent individual owner gets nearly half again the benefit of a 21 percent C corporation, and an owner in a low-income year may prefer to wait.

Holding horizon. The value of acceleration is time value. Sell in two years and most of the deferral reverses at sale — with the short-life portion recaptured at ordinary rates under Section 1245 rather than the 25 percent ceiling that applies to straight-line building depreciation. Three to five years is a reasonable floor; a decade is where the numbers get comfortable.

Passive-loss posture. This is the screen most often skipped. Rental losses are passive under Section 469 unless the owner is a real estate professional or has passive income to absorb them. A study that generates a $700,000 first-year loss for a passive W-2 investor produces a suspended carryforward, not a refund. The deduction is not lost, but its present value collapses toward zero. This failure mode and several others are cataloged in when cost segregation doesn't make sense.

A quick go/no-go screen:

Our worked example clears every gate: $3.2 million purchase price, $600,000 allocated to land (supported by the assessor's ratio and the appraisal), $2.6 million of depreciable basis, a 37 percent owner who qualifies as a real estate professional, and a ten-year hold. Acquired March 2026 — after January 19, 2025 — so 100 percent bonus depreciation applies to whatever the study finds.

Stage 2: Choosing a provider and scoping the work

Cost segregation is unregulated: no license, no credential requirement, no standard fee schedule. Quality varies accordingly, and the owner's leverage is all in the scoping conversation.

What separates providers

The Cost Segregation Audit Techniques Guide — the IRS's manual for examining these studies — ranks methodologies. The detailed engineering approach from actual cost records sits at the top; survey and residual estimation approaches sit lower; rule-of-thumb percentage allocations sit at the bottom and, per the ATG, merit little weight. Ask which approach the provider uses and whether the people doing takeoffs have engineering or construction-estimating backgrounds. Ask how many of their studies have been examined and what the outcomes were. A provider who cannot answer either question crisply is telling you something.

Engagement letter terms that matter

• Fixed fee, not contingent. A fee contingent on the deductions found gives the analyst a direct financial stake in aggressive classification — a fact examiners notice. Fixed fees are the professional norm.
• Deliverable specification. The letter should commit to a report containing the ATG quality elements (next two sections), not merely "a schedule of reclassified assets."
• Site visit included. Confirm it is in scope, not an upsell.
• Audit support. What happens if the study is examined in year three? Standard terms include defending the work at no charge or at stated hourly rates; silence here is a red flag.
• Form 3115 responsibility. For look-back studies, establish who prepares the Section 481(a) computation and the form itself — the study firm, the CPA, or both.

For our building, the owner signs a fixed-fee $8,500 engagement for a full engineering-based study with site visit, ATG-conforming report, and examination support, with the CPA to review classifications before filing.

Stage 3: What the engineering team actually does

The owner's active work is front-loaded into document production and one site walk. The rest is the provider's.

Document requests. For an acquired building like ours: closing statement, purchase agreement, appraisal, property tax assessment (for the land allocation), and any drawings the seller or the county can supply. For new construction, the request is richer: the contractor's final application for payment (the AIA G702/G703 forms, which break the job into cost divisions), change orders, and soft-cost invoices. The gap matters — on an acquisition there is usually no cost record for the carpet, so the engineer must estimate it.

Site visit. An engineer walks the property, photographs components, verifies drawings against as-built conditions, and finds what documents never show: dedicated circuits serving equipment, supplemental HVAC over a server room, process plumbing, the extent of the parking field. The ATG lists property inspection among its principal elements of a quality study; a report with no photographs of the actual building is conspicuously weaker.

Takeoffs and costing. Each candidate short-life component is quantified — square feet of carpet, linear feet of millwork, count of decorative fixtures, area of paving — and costed. Where actual cost detail exists it is used; where it does not, the engineer estimates from published construction cost data (RSMeans is the standard source), adjusted for location and the building's age, then depreciates the estimate to the acquisition date. Critically, everything must reconcile: component costs, short-life and long-life together, must sum exactly to depreciable basis. A study that only prices the 5-year property and calls the rest "residual" is using the weakest method in the ATG's hierarchy.

Allocation of indirects. Soft costs — architect and engineering fees, permits, general conditions, contractor overhead and profit — attach to the components they benefit, generally pro rata over direct costs. On a $2.6 million basis, indirects can be several hundred thousand dollars; allocating them only to long-life property understates the benefit, and allocating them only to short-life property is an audit flag.

Classification. Each costed component is assigned a class life under Rev. Proc. 87-56 and a recovery period under Section 168, applying the Whiteco/Hospital Corporation of America factors to close calls. Recovery periods and conventions are laid out in Publication 946; the conceptual groundwork is in our depreciation primer.

Stage 4: The report — what a defensible deliverable contains

The ATG devotes a chapter to the principal elements of a quality study, and a good report reads like it was written against that checklist, because it was:

Two pages of percentages with a cover letter is not a study; it is an invitation. When an examiner opens a file, the ATG hands them this same checklist, and every missing element shifts the burden of proof in practice against the taxpayer.

Our building's result. The study allocates the $2.6 million of depreciable basis as follows: $416,000 (16 percent) to 5-year property (carpet and resilient flooring, millwork and cabinetry, dedicated electrical and data serving workstations and flex-space equipment, window treatments, decorative lighting); $52,000 (2 percent) to 7-year property (built-in telecom equipment); $286,000 (11 percent) to 15-year land improvements (the parking field, curbing, site lighting, landscaping, monument signage); and $1,846,000 (71 percent) remains 39-year nonresidential real property.

5-year property$416,000

7-year property$52,000

15-year land improvements$286,000

39-year building$1,846,000

Worked example used throughout this article; a real building's allocation depends on its actual components and documentation.

Twenty-nine percent reclassified is toward the high end of the office range, supported here by an unusually large parking field and equipment-heavy flex space. The engineering detail, not the percentage, is what defends it.

Stage 5: Filing — current-year return or Form 3115 look-back?

How the study reaches the IRS depends entirely on timing.

Placed-in-service-year study. If the study is done for the year the building is placed in service, the classifications simply go on that year's return: the assets are entered on the depreciation schedule and reported on Form 4562. No method change is involved, because no method has been established yet. This is the clean path, and it is why the best time to commission a study is at or shortly after closing.

Look-back study. If the building has already been depreciated as a single 39-year asset on two or more filed returns, that treatment is an established (impermissible) accounting method. The fix is not amended returns. It is Form 3115, filed under the automatic change procedures, with the entire catch-up — the difference between depreciation actually taken and depreciation that would have been taken under the study — deducted in the year of change as a negative Section 481(a) adjustment. A building depreciated wrongly for six years produces six years of catch-up in a single year, on top of that year's normal depreciation. Automatic consent means no user fee and no advance IRS ruling; the form is attached to a timely filed return with a copy sent to the IRS separately. The mechanics, eligibility rules, and traps are covered in look-back studies and Form 3115; the general framework for method changes is in Publication 538.

Which path applies:

Our example is the clean case: purchased and placed in service in March 2026, study completed in June, everything filed on the 2026 return.

Stage 6: Using the results

The report is not the end. It is the opening balance of an asset ledger the owner will use for the rest of the holding period.

Year-one deductions, with 100% bonus

Because the building was acquired after January 19, 2025, the property with recovery periods of 20 years or less — all $754,000 of 5-, 7-, and 15-year property — qualifies for 100 percent bonus depreciation under Section 168(k), permanently restored by the One Big Beautiful Bill Act. The 39-year remainder depreciates straight-line, mid-month convention, starting in March.

Year-one deduction: study plus bonus versus no study, on $2.6M of depreciable basis:

(The 39-year figures use the mid-month convention for a March placed-in-service date: 9.5/12 of a full year, on $2.6 million and $1.846 million respectively.) The $273,000 is deferral, not a gift — total depreciation over the ownership period is identical either way — but at any positive discount rate it is worth a large multiple of the $8,500 fee. The interaction that produces this result is dissected in cost segregation and bonus depreciation. Had the building been acquired in 2024, the bonus rate would have been 60 percent under the pre-OBBBA phase-down, and the year-one number correspondingly smaller.

Partial dispositions

Because the study itemized components, the owner can later make a partial disposition election when a component is replaced. Repave the parking lot in 2031 and the remaining basis in the old paving can be written off rather than depreciated alongside its replacement — an election that is practically unavailable without component-level cost detail. See partial disposition elections.

Recapture tracking for exit

Every dollar of accelerated depreciation on the Section 1245 property is potential ordinary-income recapture at sale, and the straight-line building depreciation accumulates toward unrecaptured Section 1250 gain taxed at up to 25 percent. The study's schedule is the source document for computing both, and for allocating the sale price among asset classes on Form 8594 in an asset deal. Owners planning a Section 1031 exchange should also know that the personal-property character of study assets matters at exchange time. Keep the study, the fixed-asset schedule, and every subsequent partial disposition in one file; the buyer's diligence team and your own exit-year preparer will both need them.

Timeline and what can go wrong

Timeline. A realistic single-property schedule: week 1–2, engagement and document collection; week 2–3, site visit; week 3–6, takeoffs, costing, and classification; week 6–8, draft report, CPA review, final deliverable. Add time for portfolios, missing records, or a provider's busy season (studies bunch before filing deadlines, like everything else in tax). For a placed-in-service-year filing, work backward from the return's due date with margin — a study delivered April 10 for an April 15 filing helps no one.

The failure modes worth knowing in advance:

• Bad land allocation. Land is not depreciable, and every dollar wrongly shifted from land to building overstates all subsequent depreciation. Examiners check the land number first because it is the easiest to test — against the assessor's ratio and the appraisal. A study built on an indefensible $200,000 land figure for a $3.2 million property is compromised before the first component is classified.
• Aggressive percentages. An office study claiming 45 percent reclassification is not clever; it is a selection risk. The ATG tells examiners what typical ranges look like and which components are commonly misclassified — general-purpose HVAC, ordinary electrical distribution, and structural elements dressed up as personal property are perennial adjustments. When the outlier position loses, the taxpayer also faces the accuracy-related penalty math.
• No site visit. Desktop studies are cheaper and, on examination, worth roughly what they cost. Unverified quantities are the first thread an examiner pulls.
• The unusable deduction. The study performs exactly as promised and generates a loss the owner cannot deduct — suspended under Section 469, or wasted in a low-income year. This is a feasibility failure, not an engineering one, and it is the CPA's to catch at Stage 1.
• The orphaned report. The study is filed and forgotten; five years later, nobody reconciles the fixed-asset system to it, a replaced roof section is depreciated twice, and the exit-year recapture calculation starts from scratch. The report is a living schedule. Treat it like one.

Run honestly, the process is unglamorous: a screen, a scoped engagement, six weeks of engineering, a report written to the government's own specification, and one of two well-marked filing routes. The owners who do badly are almost never the ones who skipped a step in the engineering; they are the ones who skipped Stage 1, or who bought the report and never used it again.