Depreciation Diaries · Wine country · reverse-engineered study

What a top-down estimate missed:
$120K+ in a gym and a sauna

A client came to us worn down by a third-party partner, three months of back and forth and $3,000 into a study they did not trust after a string of mistakes. They asked us to reverse engineer it, and we found more than $120,000 in missed deductions in how the gym and the sauna were treated.

The edge, in four numbers same property · two studies · 100% bonus, 2026
The reclassified ADUs
$120,425
Gym plus sauna basis, lifted out of the 39-year shell into 5-year §1245.
Into the fastest bucket
+$82,925
More basis moved to the 5-year class, from $169,213 to $252,138.
Basis the estimate missed
+$97,319
Depreciable basis the single top-down number never captured.
The payoff · year one
$533,023
First-year deduction at 100% bonus for a 2026 placed-in-service asset.
Fully sourced · a deduction, not cash
Unlevered vs. top-down estimate: basis allocationTHE DIVERGENCESame property. Two studies. Two answers.One 2,759 sq ft property, analyzed top-down and bottoms-up. Where each method lands the basis is the whole story.5-yr (fastest)15-yr39-yr shell$951,27827.5-yr shell$284,273$169,213$993,47339-yr shell$256,472$252,138basis $1,404,764basis $1,502,083+$82,925into the 5-yr classSe████top-down · estimated · no line sourcesUnleveredbottoms-up · sourced · audit-readyWHERE THE EDGE IS+$120,425gym + sauna, reclassifiedto 5-year §1245 personalproperty.The top-down study buriedboth units in the 39-yearshell. We booked them wherethey belong.
Same 2,759 sq ft property, two methods · Se████ top-down estimate vs. Unlevered bottoms-up reconstruction

A client came to us worn down. They had spent three months going back and forth with a third-party partner on a cost segregation study, one revision after another, one mistake after another. By the end they had paid $3,000 for a study they did not trust, and they were sure a lot had been missed. So they asked us to do something simple. Reverse engineer it, and tell them what the real answer was.

We found more than $120,000 in missed deductions, almost all of it based on how the gym and sauna were classified and treated. A gym and a sauna. The original study had folded them into the main house, where they sat in the long building shell and gave back almost nothing in the first year. Booked where they belong, as 5-year property fully eligible for 100% bonus in 2026, their $120,425 of basis turns into a first-year deduction the original study left on the table. One classification call, made correctly.

Isometric view of the wine country property: two cabins, a pool, and detached studio units on a wooded parcel
The property: a main cabin, a second cabin, a pool, and the two detached studio units at the center of this study. The gym and sauna are the small structures set on pads, not attached to either house.

An ADU sauna and gym highlight the impact of 3D data mapping

The owner had invested in two accessory dwelling units. One was a prefab gym studio, a small building in its own right, delivered and set on a pad. The second was a prefab cedar sauna. Neither sits on a permanent foundation, both are relocatable, and both came with the owner’s own paid vendor quotes, $92,500 for the studio and $27,925 for the sauna. That is the $120,425, and every dollar of it traces to a document the owner already had.

Picture what these actually are. A prefab studio set on a pad, not tied into either house, and a sauna you could lift onto a trailer. You could truck either one to your next Airbnb.

That is not a stretch. The Whiteco six-factor test lands both units in the 5-year class as §1245 property, fully bonus eligible for a 2026 placed-in-service asset. The original study had buried them in the 39-year shell instead.

Two studies, two different approaches. The traditional firm, stuck within the confines of old macros and Excel. The other, building frontier AI verification to extract and map every depreciable component.

Why a top-down estimate cannot see an ADU

A top-down study starts from one basis number and spreads it with industry medians like RSMeans and Marshall. Those medians are useful when actual construction costs are gone, but they describe an average house. Faced with two extra structures on the same parcel, the macro has two moves, fold them into the house or rebuild each at 39 years, and neither reaches personal property, because the method never had the fact that these units are movable. Estimators fall back to the same buckets for almost every house, a rough 12% kitchen and so on, and anything genuinely unusual gets averaged into something ordinary.

Hosting is exactly where this bites. The ways people actually build and operate short-term rentals, detached wellness studios, movable units, furnishings brought in after the fact, are the cases a median was never designed to price.

Step through both studies

Below is the same property analyzed two ways. Step through each, click any component to trace it to its receipt, and vote on the two questions the study turns on.

InteractiveTop-down estimate vs. sourced reconstructionSwipe ‹ › · click · vote

What reconstruction actually reads

The ADUs are the headline, but they are one case of a broader capability. Rebuilding from the actual asset, we read the floor plan, the photos, and the video, separate the furnishings that were carried in from the walls that were built, read the year built off the images, and tell drywall from paneling or a Bosch range from a Whirlpool. Each of those calls changes how a component is classified, which lets us work at the level of the specific unit rather than the average.

That specificity is what moved the numbers. The reconstruction carried a $1,502,083 basis built from actual invoices, against the estimate’s $1,404,764, a difference of $97,319 the single number never captured, and it moved $82,925 more into the 5-year class.

Every dollar is verifiable

What I care about most is that the number is defensible. Every component links to a source document, and the CPA gets the classification, the citation, the Whiteco analysis, and the formulas behind the rollup, so any line traces from the deduction back to the receipt. Across the sourced studies we run, we see roughly 10% to 15% more in identified deductions than comparable top-down work, and we can stand behind each one. We publish them too, because there is no real public repository of cost segregation studies anywhere, and showing the work is the point.

The firm that ran the first study did solid work inside its method. Top-down estimating and bottoms-up reconstruction are simply two different specialties. We built ours on verification: a 3D data map lets us verify, down to the last square foot, the actual allocation of every component of the home, which is what makes this one of the most defensible studies on the market, if not the most.

Wondering if your study left money on the table?

If you have already paid for a cost segregation study, we can reverse engineer it and show you, line by line, what a sourced reconstruction finds. If you are earlier than that, we can tell you whether this is even the right fit for your property.

Illustrative case study built from a real reverse-engineered study; identifying client and vendor details removed and the property referred to only as wine country. Figures reflect two studies of the same 2,759 sq ft multi-structure property. Comparison firm name masked. Unlevered reconstruction: 5-yr $252,138 · 15-yr $256,472 · 39-yr shell $993,473 · basis $1,502,083. Nothing here is tax advice; treatment depends on your facts and your preparer’s judgment.