A 42-Sale Dataset,
Rejected

Minnesota's sales-comparison approach relies on one unstated assumption: the adjustment rates used to reconcile differences between comparable sales and the subject property are themselves derived from transactions that are actually comparable to the subject. Violate that assumption and every number downstream is contaminated. Diamond Lake Road Industrial Owner, LLC v. Hennepin County is the first Minnesota Tax Court decision to write that requirement down in plain terms, and to apply it strictly enough to cut a 22.5% reduction out of a large distribution warehouse.

What the Dataset Contained

The subject property is a 386,724-square-foot distribution warehouse at 19850 Diamond Lake Road in Rogers, with 4% office buildout and 32-foot clear heights — a modern bulk-distribution asset on a 968,788-square-foot site. The county's appraisal used a single 42-sale dataset to derive all of its sales-comparison adjustments: adjustments for clear height, office buildout, dock-door ratio, age, parking, and land-to-building ratio. The same 42 sales drove every adjustment rate in the county's report.

The problem was what those 42 sales contained. The record showed the dataset blended bulk warehouses with manufacturing plants, engineering research and development labs, office showrooms, and industrial flex facilities — five distinct property subtypes, treated as a single population for regression purposes. The county's appraiser testified that the dataset was mixed intentionally, because a single broad industrial dataset was useful for "bracketing" the subject across the range of adjustment variables. The court accepted the testimony about intent. It rejected the methodology built on top of it.

The petitioner's appraisal used a dataset restricted to the same property subtype as the subject — large distribution warehouses. Its maximum gross adjustment across any comparable was 42%. The county's dataset, with five property subtypes thrown in together, required gross adjustments as high as 127.9% and net adjustments as high as 64.6% to make the data fit. The court did not need to resolve whether any single adjustment was computed correctly. It held that the underlying data could not support reliable adjustment rates at all, regardless of how the math was run on top of them.

Market Delineation First

The rule the court wrote is narrow on its face and broad in its implications. Market delineation and segmentation — identifying which property subtype and submarket the subject actually operates in — must come before adjustment rates are derived from comparable sales, not after. The court quoted the principle directly from the record and held that it applies with equal force to regression-based adjustments, scatter-diagram trend lines, and paired-sales analysis. It is a foundational requirement, not a stylistic preference.

Our principal concern is that [the county's] data analyses do not produce meaningful trend lines to underwrite reliable adjustments. Market delineation and segmentation should be the first step in valuation modeling, including regression, as it is a fundamental requirement for validity. — Hon. Bradford S. Delapena, Diamond Lake v. Hennepin County (April 22, 2025)

The court did not stop at the doctrinal holding. It worked through the specifics of the 42-sale dataset on the record. Only nine of the 42 sales were large warehouses (more than 280,000 square feet) with clear heights over 21 feet — the structural features that define a modern bulk-distribution asset. Ten of the 42 had clear heights of 16 feet or less, which the record established is below the practical minimum for modern distribution. Nineteen of the 42 had office buildout greater than 20% — four to five times the subject's 4% ratio. One of the 42 was a 114,667-square-foot building with 72% office buildout and 14-foot clear heights. None of that was contested. The court simply found, sale by sale, that the dataset was populated with properties that are not competitive alternatives for a purchaser of a 386,724-square-foot distribution warehouse.

What Diamond Lake Establishes

Three rules emerge from the decision that we expect to cite in every Minnesota industrial appeal going forward.

First, adjustment rates must be derived from same-type comparables. A distribution warehouse adjustment rate cannot come from a regression that includes manufacturing, R&D, or showroom sales. This applies even where mixed data "brackets" the subject on specific features like clear height or dock-door ratio. Bracketing is a property-feature test; delineation is a property-type test. They are not substitutes.

Second, gross adjustment magnitudes are a diagnostic. When an appraiser's sales-comparison approach requires adjustments in the 75%-to-128% range to reconcile comparables to the subject, something upstream in the data is wrong. The court did not adopt a specific threshold, but the contrast on the record — 42% maximum gross for the same-type dataset versus 128% for the mixed dataset — was treated as probative evidence of dataset unreliability.

Third, a single-tenant sale of the subject does not require a "property rights" adjustment for vacancy. The county's appraisal added a 26.8% upward adjustment to the subject's own September 2021 sale price on the theory that the building was vacant at sale. The court rejected this entirely. Single-tenant industrial properties are naturally 0% or 100% occupied. The price paid by an investor for a vacant single-tenant warehouse already reflects the fee-simple interest and the long-term vacancy characteristics of the asset class. Grossing it up for a "property rights" adjustment double-counts the occupancy risk.

The Rule Travels

The immediate audience for Diamond Lake is owners of large distribution and bulk-warehouse assets — the subtype most vulnerable to bulk-industrial regressions that sweep in manufacturing and flex comparables. Every Minnesota county has been assessing distribution warehouses off of datasets built this way for years. Each of those assessments is now challengeable on the same grounds the Tax Court accepted here.

The longer-range implication is broader. The doctrinal move in Diamond Lake — that market delineation is an evidentiary prerequisite to adjustment, not an afterthought — is not a warehouse-specific rule. It applies with equal force to any commercial subtype whose assessor relies on a pooled regression across related-but-distinct property types. Class A office valued against a dataset that includes Class B and medical. Senior housing valued against a dataset that includes multifamily. Suburban retail valued against a dataset that includes strip centers and big-box. All of these are vulnerable to the same objection, and the appraisal-theory framework the court cited applies cleanly to each one.

The petition deadline for the 25pay26 tax year is April 30, 2026. Any commercial owner whose assessment was built on a pooled-subtype regression should have the dataset behind that assessment reviewed before that date.