Comparable Sales Analysis in Real Estate

Introduction

Sales comparison is the most intuitive of the four real estate valuation methods and the cleanest when the conditions are right: a deep set of recent comparable transactions in the same submarket gives a direct market read on per-unit pricing that does not require any input-assumption calibration. The conditions are right less often than candidates assume. In active core submarkets with regular trading (Class A multifamily in major metros, top industrial markets, prime retail nodes), comp data is deep and sales comparison is the strongest method. In thinner or specialty submarkets (tertiary cities, specialty asset classes, dislocated markets), comp data is sparse and sales comparison gives way to the income approach as the primary method.

The mechanical work in sales comparison breaks into three judgment-heavy steps: choosing the right unit of comparison for the property type, selecting a defensible comp set, and applying adjustments that translate the comp prices into a defensible per-unit price for the subject. Each step has standard conventions and well-known failure modes. The output of a careless run looks indistinguishable from the output of a careful run on the spreadsheet, which is the structural reason why sales-comparison work is the part of an underwriting where bad analysts hide longest.

Each sub-sector uses a standard unit of comparison that reflects how its assets are bought and sold:

The choice of unit matters because it isolates the property's economic value in the way buyers actually price it. Multifamily buyers think in per-door terms because the rent and unit-level economics drive value; hotel buyers think in per-key terms because the room is the revenue unit; data center buyers think in per-kilowatt terms because power capacity (not floor space) is the binding constraint on the asset's earning power. Mismatching the unit to the sub-sector produces nonsense comparisons.

Whatever the sub-sector, the per-unit metric itself is the same simple ratio: the sale price divided by the relevant count of revenue-bearing units. The variable in the denominator is what changes across property types.

The count is square footage for office and industrial, doors for multifamily, keys for hotels, beds for senior housing, pads for manufactured housing, and kilowatts of IT load for data centers. Computing this ratio across the comp set produces the raw per-unit prices that the adjustment process then refines.

Unit of Comparison

The standard per-asset measure used in real estate sales comparison: price per square foot for office, industrial, and most retail; price per door (per unit) for multifamily; price per key (per room) for hotels; price per bed for senior housing; price per kilowatt (per megawatt) of IT load for data centers; price per pad for manufactured housing. The unit is chosen to reflect how buyers in each sub-sector underwrite economic value.

Cross-Checks Across Units

For some property types, multiple units of comparison are checked simultaneously. A multifamily comp set might be cross-checked using both price per door (primary) and price per square foot (secondary), because the two metrics flag different attributes: per door isolates unit-level economics; per square foot adjusts for unit-size differences across comps. An office comp might be checked using price per square foot alongside implied cap rate (price divided by NOI), because the two together reveal whether the comp is trading on physical attributes or on income attributes. The same dual-lens discipline appears in public-markets comparable company analysis, where a single peer is cross-checked across several multiples to see which one the market is actually pricing on.

Once the comp set is identified, each comp is adjusted to the subject property. The underlying logic is to take the per-unit price the market paid for a comparable asset, nudge it for the ways the subject differs, and scale it by the subject's own unit count to land on an indicated value.

Here the adjusted comp price per unit is the comp's raw per-unit price after the size, quality, location, lease, time, and special-feature adjustments below, and "units" is whichever count the sub-sector trades on (SF, doors, keys, beds). The standard adjustment categories:

The adjustment magnitudes are calibrated either through paired sales analysis (comparing two near-identical comps that differ in a single attribute to isolate the per-attribute adjustment) or through market-derived rules of thumb at experienced brokerages and appraisal firms. Different practitioners arrive at different adjustment magnitudes; defensible work documents each adjustment explicitly with the supporting rationale.

How Many Adjustments Is Too Many

A widely accepted convention is that net adjustments over 15% of the comp price or gross adjustments over 25% indicate the comp is not really comparable enough to be informative. A comp requiring 30% in total adjustments is essentially a different property than the subject, and its informational value is low even after adjustment. Strong analysts replace heavily adjusted comps with closer comps where possible, even if the closer comp is slightly older or further from the subject submarket.

The comp set is the foundation. Standard practice identifies 3 to 5 comp sales that meet the following criteria:

In active submarkets, finding 5 qualifying comps within the criteria is usually possible. In thinner submarkets, the comp set may shrink to 2 or 3, which weakens the method's reliability and pushes the analyst toward direct capitalization as the primary method instead.

Arms-Length Transaction

A sale between unrelated parties acting independently in their own self-interest, with full market exposure of the asset and no special financing or non-cash consideration that would distort the apparent price. The standard requirement for any comp included in a sales-comparison analysis. Forced sales, related-party transactions, and partial-interest trades fall outside the arms-length standard and should be excluded from the comp set or flagged explicitly as non-comparable.

Sales comparison fails or weakens in three specific situations:

Three errors recur often enough on actual deals that they are worth naming together. First, including non-arms-length comps in the set without flagging them: a forced sale, a related-party transaction, or a partial-interest trade looks like a normal comp in CoStar or Real Capital Analytics but reflects a price the open market would not have produced. Second, leaving out the time adjustment when the comp is more than 6 months old: a 9-month-old comp at a 5.0% cap rate is a 5.5% comp today if cap rates have widened by 50 bps in the interim, and ignoring that adjustment systematically biases the valuation. Third, applying the method at all when the submarket does not have the trading depth to support it: 1 or 2 comps is not a comp set, and forcing a sales-comparison conclusion on a 2-comp universe is what the income approach is designed to replace. Each of these errors looks defensible on a comp sheet at first glance, which is why senior reviewers always check the comp list itself before they look at the adjusted per-unit prices.