Two providers quote the same $1,900,000 retail center. One says 24% reclassified for $1,500 in 48 hours; the other says 28% for $8,500 in four weeks. The numbers look close enough that the choice seems obvious. They are not the same product, and the difference shows up when you cannot fix it: when an examiner asks how you arrived at your numbers.
The short answer
- An engineering-based cost segregation study measures your building. It performs a takeoff, prices actual assets, classifies them one by one, and reconciles the total back to what you paid.
- A software estimate models your building. It applies a statistical allocation from comparable properties to your purchase price. No asset is measured, priced, or individually classified.
- The IRS Audit Techniques Guide asks for methodology, documentation, unit costs, engineering rationale, and a reconciliation to actual cost. A model output has none of those.
- A modeled estimate is legitimate for feasibility screening and reasonable on simple, low-basis property where the fee gap would eat the benefit.
- The dividing line is roughly $500,000 of depreciable basis, plus any real complexity in finishes, process systems, or site work.
What an Engineering-Based Study Actually Does
An engineering-based study starts from the building, not the price. After collecting the closing statement, appraisal, construction contracts, drawings, and your depreciation schedule, someone inspects the property and performs a takeoff: linear feet of millwork, square feet of specialty flooring, fixture counts, dedicated electrical runs, square yards of paving, site utility runs.
Each quantity gets a unit cost, from your actual cost records where they exist or from recognized estimating data adjusted for location and vintage, and each asset is classified individually against the permanence and function tests with the reasoning written down.
Then the step nobody asks about: the total allocated cost is reconciled back to your total actual cost. Every dollar of basis lands somewhere and the sum matches the closing statement. That is what proves the study describes your building rather than a category of buildings.
What a Software Estimate Actually Does
A software estimate does something genuinely different, and it deserves an accurate description rather than a dismissive one. The tool takes your property type, size, region, vintage, and price, then applies allocation percentages drawn from a database of studies on comparable buildings.
The math behind it is often reasonable. The percentages come from real engineering work, just not on your building. The output cannot tell you whether your center has vinyl composition tile or polished concrete, whether the fit-out included dedicated 60-amp runs to equipment, or whether the lot is asphalt or a striped concrete apron. Nor can it tell an examiner why any specific number is what it is, because there was no takeoff, no unit cost, no legal analysis, and no reconciliation.
Why Two Similar Numbers Defend Completely Differently
The IRS Cost Segregation Audit Techniques Guide, Publication 5653, revised February 2025, lists 13 elements of a quality study: a preparer with construction and tax expertise, a description of methodology, appropriate documentation, the legal analysis, unit costs and asset groupings, an engineering rationale for cost allocations, an asset-by-asset listing of §1245 property, and a reconciliation of allocated costs to actual costs. A model output satisfies almost none of that, not because it is dishonest but because those elements describe a process it never performed.
The ATG also describes six methodologies. It publishes no numbered ranking, but it identifies the Detailed Engineering Approach from Actual Cost Records as the most reliable and rule-of-thumb approaches as the least, with the Detailed Engineering Cost Estimate Approach the standard where cost records are unavailable. Our walkthrough of what the ATG expects covers all six.
Side by Side, Honestly
| Engineering-based study | Software estimate | |
|---|---|---|
| How the number is produced | Takeoff, measured quantities, unit costs | Statistical model of comparable buildings |
| Site inspection | Yes, or documented substitute | No |
| Asset-level detail | Every asset listed and classified | Category totals only |
| Legal analysis | Written per classification decision | None |
| Reconciliation to actual cost | Required and shown | Not performed |
| Typical fee | $2,500–$6,500 residential; $5,000–$15,000 commercial; $15,000–$40,000 complex | Several hundred to ~$2,000 |
| Turnaround | Two to six weeks | Hours to days |
| Audit posture | Defensible on its own documentation | Depends on preparer support you may not have |
| Best fit | Anything above ~$500,000 basis, or with real complexity | Screening, and simple low-basis property |
Our breakdown of what a study costs explains what drives a quote.
The Worked Example: Where the Fee Gap Goes
Example: a $1,900,000 retail strip center Land allocated at $350,000, leaving $1,550,000 of depreciable basis. The software estimate returns 24%. The engineering study, after a site visit finding tenant-specific electrical, decorative lighting, and more site work than the model assumed, lands at 28%. The benefit line is the incremental deduction: reclassified dollars less the 39-year depreciation those dollars would have thrown off anyway, shown as a simplified full year.
| Software estimate | Engineering study | |
|---|---|---|
| Reclassified to 5-, 7-, 15-year | $372,000 | $434,000 |
| Additional reclassification | $62,000 | |
| Less first-year 39-year depreciation forgone | ($9,538) | ($11,128) |
| Incremental first-year deduction | $362,462 | $422,872 |
| Cash value at a 35% combined rate | $126,862 | $148,005 |
| Study fee | $1,500 | $8,500 |
| Net first-year cash after fee | $125,362 | $139,505 |
The engineering study costs $7,000 more and returns roughly $21,100 more in first-year cash, about $14,100 net of the fee difference, before you value the documentation at all. Your result depends on your tax rate and whether you can use passive losses.
When Engineering Rigor Is Not Worth Paying For
Here is the line as we actually draw it.
If your depreciable basis is under roughly $500,000, the property is a plain single-family rental or a simple pad site, and there is no meaningful site work or specialty system, the fee for a full study can consume most of the incremental benefit. A modeled estimate, or no study, may be right. The same holds if you cannot use the deduction this year, since a perfect study producing a suspended loss is an expense with no near-term return.
Modeled estimates are also the right tool for screening. We use one before quoting, because it answers "is there enough here to justify the engineering" in an afternoon, at no cost.
Complexity changes the answer. Restaurants, medical suites, hotels, manufacturing, self-storage, and car washes are what a model handles worst, because their variance from the category average is the whole point. Add a Form 3115 look-back with a large §481(a) catch-up and you have a visible filing supported by a document that cannot explain itself. That is the combination worth avoiding, and the subject of cost segregation audit risk.
Frequently Asked Questions
Is a software-generated cost segregation study legal?
Nothing prohibits estimating. The question is evidentiary, not legal. If your classifications are examined you must support them with methodology, documentation, and unit costs, and a modeled allocation has none of that. On a small, simple property the exposure may be modest. On a $3M medical building it is not.
Does a site visit have to happen?
Not in every case, but its absence has to be explained and replaced. Detailed construction records, complete drawings, dated photography, and documented interviews can substitute on some properties. A provider who never inspects and does not say what stands in its place is telling you the answer came from a model.
How do I tell which one I am being sold?
Ask for a redacted sample report and look for three things: an asset-by-asset detail schedule, stated unit costs, and a reconciliation to total actual cost. If all three are present, it is engineering. The full question list is in our guide to choosing a cost segregation provider, and the study process is described step by step here.
See What Your Property Would Yield
Every building is different, and the only way to know your number is to look at your building. Precision Cost Segregation provides a no-cost feasibility analysis: send us the property address, purchase price, closing date, and any improvements, and we'll model your likely reclassification and first-year benefit before you commit to anything.
Request a free feasibility analysis →
This article is provided by Precision Cost Segregation for general educational purposes and does not constitute tax, legal, or accounting advice. Tax outcomes depend on your specific facts, your tax rate, your passive activity position, and your entity structure. Figures shown are illustrative. Consult your CPA or tax advisor before acting, and engage a qualified professional to perform any cost segregation study. Information is current as of publication.
This article is general information, not tax or legal advice. Depreciation outcomes depend on your facts, elections, and current law — consult your CPA before acting. © 2026 Precision Cost Segregation.