IRS Cost Segregation ATG

Key Takeaways

• The IRS published its own guide for auditing cost segregation studies — the Cost Segregation Audit Techniques Guide (Publication 5653)
• A defensible study addresses all 13 principal elements outlined in the ATG
• Engineering-based cost analysis is the gold standard; the IRS explicitly warns against flat-percentage methods
• Reconciliation to total cost basis and component-level documentation are non-negotiable

Not all cost segregation studies are created equal. The IRS published its own guide for auditing them — the Cost Segregation Audit Techniques Guide (Publication 5653). A defensible study follows that guide. Here’s what that means in practice.

The ATG isn’t a suggestion. It’s the document IRS examiners use when they review a cost segregation study during an audit. It tells them what to look for, what questions to ask, and what a quality study should contain. Any provider producing studies that don’t hold up against these criteria is creating liability for their clients. For a broader look at how cost segregation works and whether it actually increases audit risk, see our separate analysis.

The 13 Principal Elements

The ATG identifies 13 principal elements that a quality cost segregation study should contain. IRS examiners use this list as their evaluation framework. A study that addresses all 13 is far more likely to withstand scrutiny than one that skips half of them.

• Preparation by a qualified individual. The study should be prepared or supervised by someone with expertise in engineering, construction, and tax law related to property classification. The ATG doesn’t require a specific credential, but it expects demonstrable competence.
• Detailed description of the methodology. The report must explain how the study was conducted — what approach was used, what data sources were consulted, and how costs were estimated. Vague descriptions like “industry standard methods” aren’t sufficient.
• Use of appropriate documentation. This includes blueprints, appraisals, property records, photographs, cost data, and any other materials used to support the analysis. The documentation trail should be clear enough for an examiner to follow.
• Discussion of relevant law and regulations. The study should cite the specific IRC sections, Treasury Regulations, and Revenue Procedures that govern asset classification — primarily IRC §168 and Rev. Proc. 87-56.
• Analysis of relevant court cases and rulings. Key case law (Hospital Corporation of America v. Commissioner, Whiteco Industries, among others) established precedents for how specific building components are classified. A defensible study acknowledges these.
• Identification of all property components. Every building system, fixture, and site improvement should be identified individually. Lumping components into broad categories (“electrical” or “plumbing”) without further breakdown is a red flag.
• Determination of asset classifications. Each identified component must be assigned to a specific MACRS asset class — 5-year, 7-year, 15-year, or remaining in the default 27.5-year or 39-year class — with a stated basis for the classification.
• Explanation of cost estimating techniques. How were dollar values assigned to each component? The study should explain whether it used actual construction invoices, engineering estimates from cost databases, or other quantifiable methods.
• Legal analysis supporting classifications. Beyond citing law, the study should explain why each component qualifies for the assigned class. This is the analytical backbone that connects the engineering work to the tax treatment.
• Reconciliation to total cost. The sum of all classified components must reconcile to the property’s total depreciable basis. If the pieces don’t add up to the whole, the study has a structural problem.
• Identification and treatment of indirect costs. Construction overhead, contractor fees, architectural costs, and other indirect costs must be allocated across components — not ignored or lumped into a single category.
• Statement of unit costs. The study should include unit-level cost data (cost per square foot, cost per fixture, etc.) that an examiner can verify against published cost data or construction records.
• Inclusion of supporting documentation. The full package — photographs, property records, cost schedules, depreciation tables, and any supplemental materials — should be organized and accessible for review.

Not every study on the market addresses all 13. Some providers produce a depreciation schedule and call it a cost segregation study. That’s not the same thing. A depreciation schedule tells you the numbers. A study explains and documents why those numbers are correct.

Which of the 13 an automated study actually covers

You will read, on comparison sites and on the marketing pages of firms that sell the alternative, that “software tools typically satisfy only some” of these 13 elements. That claim is doing a lot of work in very few words, and it is worth taking apart, because the honest answer is neither “all of them” nor “only some.”

The 13 elements are not equally sensitive to how a study is produced. Sorted by that sensitivity, they fall into three groups.

Group A — seven elements where automation is at least as strong as manual work. These are rule-governed and repetitive, which is exactly what a machine does not get tired of: the description of methodology, the discussion of relevant law and regulations, the analysis of court cases and rulings, the determination of asset classifications under Rev. Proc. 87-56, the reconciliation to total cost, the statement of unit costs, and the assembly of supporting documentation. An engine applies the same citations, the same classification logic, and the same reconciliation arithmetic to every property. A human under deadline pressure on a Friday afternoon does not. Reconciliation in particular is a structural property of a well-built engine rather than a step someone can forget.

Group B — four elements that are only as good as the input data. Identification of all property components, use of appropriate documentation, explanation of cost estimating techniques, and the treatment of indirect costs. A remote study sees what public records, aerial imagery, permit data, and your own photographs show it. It does not see custom process equipment, unusual mechanical density behind a finished wall, or a tenant build-out that never reached a permit record. It also works from modeled component costs rather than actual invoices unless you supply the invoices.

Group C — two elements where a self-service tool is genuinely weakest. Preparation by a qualified individual, and the legal analysis supporting each classification. A DIY tool’s honest answer to “who prepared this study” is “you did.” That is not automatically a bad answer — a CPA who understands the classification rules is a qualified individual within the ATG’s meaning, and the guide deliberately does not require a specific credential. But it is a different answer from one that names a reviewer, and buyers comparing on price rarely notice which answer they bought. The same applies to legal analysis: reciting the correct authorities is Group A work, while reasoning about why this component in this building takes that class is not.

What this means in practice

The gap between an automated study and an on-site engineering study is concentrated almost entirely in Group B, and Group B’s size scales with property complexity. On a $450,000 single-family rental, what the data cannot see is close to nothing: the components that drive the 5- and 15-year buckets are the site improvements, land improvements, and interior finishes that public records and photographs show plainly. On a $10 million manufacturing plant, what the data cannot see may be the whole study.

That is why the market has split into tiers, and it is a property question rather than a technology question. The useful version of “does software satisfy the 13 elements” is therefore: for this property, how much of the answer lives in Group B?

The question to ask any provider

Ask for a complete sample report — not an excerpt — and check whether it walks all 13 elements and shows its work on each. Every study we deliver includes a table mapping each of the 13 principal elements to the section of the report that addresses it, and our full sample library is public at /sample-report/. Any provider in any tier should be able to hand you the same thing. A provider who cannot show you a finished study before you buy is telling you something about the finished study.

For how the three delivery models in the automated tier differ from each other, see the cost segregation software comparison.

Engineering-Based vs. Rules-of-Thumb

The methodology question is where most studies diverge in quality. There are two broad approaches, and the IRS has a clear position on which one holds up.

The engineering approach uses construction cost databases (industry-standard databases) and calculates component costs from the ground up. It starts with the building’s physical characteristics — square footage, construction type, age, quality grade, geographic location — and builds a cost model for each component: foundation, framing, roofing, electrical, plumbing, HVAC, finishes, fixtures, site work. Each component gets a dollar value based on published unit costs, adjusted for local construction indices (derived from BLS Producer Price Index data). The final study shows how every dollar of the depreciable basis is accounted for across specific MACRS asset classes.

The percentage method (rules-of-thumb) skips the engineering work entirely. It applies flat percentages to the total cost basis — “25% goes to 5-year, 10% goes to 15-year” — regardless of the actual property. A luxury mountain cabin with a hot tub, outdoor kitchen, and extensive landscaping gets the same percentages as a basic tract home. The numbers aren’t derived from the property; they’re derived from a template.

The IRS explicitly warns against this approach. The ATG states that the percentage method “generally does not produce a reliable result” because it doesn’t account for property-specific characteristics. An examiner reviewing a percentage-based study will immediately question whether the allocations reflect the actual property or are simply generic assumptions. For more on how to evaluate cost segregation providers on methodology, see our separate guide.

This matters practically, not just theoretically. If the IRS challenges a reclassification and the study’s basis is “we applied 22% to 5-year property because that’s our standard allocation,” there’s nothing to defend. If the basis is “we calculated $18,400 in cabinetry using industry-standard unit costs for custom-grade cabinets in 320 SF of kitchen and bath space, classified as IRS Asset Class 57.0 per Rev. Proc. 87-56,” that’s a position with substance behind it.

What Documentation a Defensible Study Should Include

When your CPA files the depreciation schedule from a cost segregation study — typically via Form 3115 for existing properties or on the initial return for newly acquired properties — the study itself becomes the supporting documentation. If the IRS asks questions, this is what they’ll review. A complete study package should contain the following. Methodology narrative A written explanation of how the study was conducted: the cost approach used, data sources consulted, observation methods, and how components were identified and classified. Cost approach calculations Component-level cost estimates showing unit costs, quantities, and total allocated cost for each building element. This is the engineering backbone of the study. IRS asset class citations per component Each reclassified component should reference the specific MACRS asset class and the legal authority (Rev. Proc. 87-56, IRC §168, relevant case law) supporting the classification. Property imagery documentation Photographs or satellite imagery documenting the property’s physical characteristics, site improvements, and observable building components. Reconciliation to total cost basis A reconciliation schedule showing that all classified components sum to the total depreciable basis (purchase price minus land). No gaps, no unexplained residuals. MACRS classification tables Complete depreciation schedules organized by recovery period (5-year, 7-year, 15-year, 27.5-year or 39-year) with component detail, ready for CPA filing.

A study that includes all of these gives your CPA what they need to file confidently and gives you documentation to withstand review. A study that only delivers a summary depreciation table leaves gaps an examiner can walk through.

Red Flags the IRS Looks For

The ATG doesn’t just tell examiners what a good study looks like. It also tells them what to flag. These are the patterns that invite additional scrutiny.

Common red flags identified in the IRS Audit Techniques Guide:

• Unsupported allocations — Dollar amounts assigned to asset classes without component-level cost calculations or published cost data to back them up
• Missing reconciliation — The classified components don’t sum to the total depreciable basis, leaving unexplained gaps in the cost allocation
• No engineering basis — Allocations derived from flat percentages or “industry standards” rather than property-specific analysis
• Aggressive land improvement classification — Placing items in the 5-year class that belong in 15-year land improvements, or classifying structural components as personal property
• Copy-paste studies not specific to the property — Reports that use identical language, identical percentages, and identical component lists across different properties with different characteristics

The last one deserves emphasis. If an examiner pulls two studies from the same provider and they read identically except for the address and purchase price, that’s strong evidence the analysis wasn’t property-specific. A defensible study should reflect the actual property: its age, construction type, site improvements, quality grade, and location. Two properties that are genuinely different should produce studies that are genuinely different.

How to Evaluate a Provider

Most investors rely on their CPA’s recommendation when choosing a cost segregation provider. That’s reasonable, but it’s worth knowing what to look for whether you’re choosing yourself or evaluating a recommendation.

Ask about methodology. “How do you calculate component costs?” is the single most important question. The answer should involve construction cost data, component-level analysis, and property-specific inputs. If the answer is vague — “we use proprietary models” or “industry-standard allocations” — push further. The methodology should be explainable in plain terms.

cost segregation methodology →

Ask for a sample report. Read it. Does it include a methodology narrative? Component-level cost breakdowns? IRS asset class citations for each reclassified component? A reconciliation schedule? Or is it just a depreciation table with a cover page? The sample tells you exactly what you’ll receive. If the provider won’t share one, that’s informative too. See a sample report to understand what’s included, or browse our example reports page for what a complete study looks like.

Verify they address all 13 elements. You don’t need to audit the study yourself. But you can check whether the report addresses the principal elements listed above. Does it cite relevant law? Does it explain the cost approach? Does it reconcile to total basis? These aren’t obscure requirements — they’re the IRS’s own checklist.

Check if CPAs accept their reports without modification. A CPA-ready study is one that your tax professional can file from directly. If a CPA regularly asks for supplemental documentation, corrections, or explanations before they’ll use a provider’s study, that’s a signal the study isn’t complete. Our reports are designed to be CPA-ready — here’s what you receive and how it works with your tax professional. For our full audit-defense framework — including the case-law foundation and what’s covered if a study is examined — see /audit-defense/.

Frequently Asked Questions What makes a cost segregation study IRS-defensible?

An IRS-defensible cost segregation study follows the methodology outlined in the IRS Cost Segregation Audit Techniques Guide (Publication 5653). It must be prepared by a qualified individual, use an engineering-based cost approach rather than flat-percentage allocations, classify components per Rev. Proc. 87-56 asset class guidelines, reconcile to total cost basis, and document all 13 principal elements the IRS expects to see. Does the IRS require an engineering-based approach for cost segregation?

The IRS does not mandate a single methodology. However, the ATG explicitly warns against the “percentage method” — flat-percentage allocations not tied to property-specific analysis — stating it “generally does not produce a reliable result.” The ATG identifies six acceptable methodologies, all of which involve detailed cost analysis. Engineering-based studies using national construction cost databases and component-level classification are the most widely accepted and most defensible approach. What are the 13 principal elements of a cost segregation study?

The 13 principal elements are: (1) preparation by a qualified individual, (2) detailed description of methodology, (3) use of appropriate documentation, (4) discussion of relevant law and regulations, (5) analysis of relevant court cases and rulings, (6) identification of property components, (7) determination of asset classifications, (8) explanation of cost estimating techniques, (9) legal analysis supporting classifications, (10) reconciliation to total cost, (11) identification and treatment of indirect costs, (12) statement of unit costs, and (13) inclusion of supporting documentation. These are the IRS’s own evaluation criteria, published in the Cost Segregation Audit Techniques Guide.

Related Reading

• Cost Segregation vs. Standard Depreciation: The $50,000 Difference
• Does a Cost Segregation Study Require a Site Visit?
• Cost Segregation for Renovations and Property Improvements
• Is Cost Seg Smart Legit? Reviews, Methodology & What You Get

See What an IRS-Defensible Study Looks Like