How Property Identity Errors Quietly Corrupt Investment Analysis
When two records describe one property, or one record hides two properties, the error rarely announces itself. It slips into per-unit metrics, comparable sets, and portfolio rollups as a number that looks reasonable but is silently wrong. Resolving identity before analysis begins is the cheapest place to stop that corruption, because every metric downstream inherits whatever the identity layer decided.
How do property identity errors affect investment analysis?
Identity errors change the denominator and the comparable set without changing anything that looks obviously broken. Split one asset into two records and its price-per-unit doubles while its size halves; merge two assets into one and the blended figures describe a property that does not exist. Because those outputs are still plausible numbers, they pass a casual sanity check and flow straight into underwriting, comps, and portfolio totals. Resolving identity first prevents the error at its source instead of chasing it through every derived metric.
Analysts are trained to scrutinize assumptions — cap rates, rent growth, exit timing — but they tend to trust the identity of the thing they are analyzing. That trust is the vulnerability. A model can be flawless and still produce a confidently wrong answer if it was pointed at a record that fused two buildings or fragmented one community into pieces.
Two failure shapes and their downstream damage
- One asset, two records: land area and unit count split across duplicates, inflating per-unit price on each fragment and understating the deal's true size.
- Two assets, one record: unrelated buildings blended into a single row, producing an average that matches neither property and poisons any comparable built from it.
- Cross-contaminated comps: a bad identity leaks into a comparable set, so even unrelated deals inherit the distortion.
- Portfolio miscounts: rollups sum or dedupe on the same broken key, quietly over- or under-stating owned units.
The insidious part is how far the error travels. A single fused record can seed a comparable that anchors the valuation of a completely different deal months later. By then the original mistake is invisible, buried under layers of legitimate-looking analysis, and no one thinks to question the property's identity.
How identity resolution protects the numbers
DealMap Intel resolves identity as a distinct step before any metric is computed. Records are keyed to a jurisdiction and parcel, related parcels are grouped into their real asset, and each material field keeps a source and a date. Analysis then runs on assets whose boundaries were decided deliberately, not on whatever the raw records happened to imply. When identity is uncertain, the system flags it rather than guessing, so a human can settle the boundary before a model relies on it.
| Metric | One asset split in two | Two assets merged |
|---|---|---|
| Price per unit | Overstated | Averaged, misleading |
| Total units | Understated per record | Overstated |
| Comparable set | Skewed high | Skewed toward a phantom |
| Portfolio count | Inflated by duplicates | Deflated by fusion |
Where resolution sits in the pipeline
Identity resolution runs upstream of valuation and comparable selection. Each candidate is reduced to a jurisdiction-plus-parcel key, grouped into its economic asset, and stamped with per-field provenance. Only then is it eligible to enter a model. Uncertain boundaries are surfaced for review instead of being resolved silently, so an analyst never inherits a guess disguised as a fact.
- —It cannot correct an underlying source that is itself wrong; it can only make the disagreement visible.
- —A confidently resolved identity can still be revised when new records arrive, so metrics may shift.
- —It does not judge whether a deal is good — it only ensures the deal describes one real asset.
- —Human review is still required where the evidence for a boundary is genuinely ambiguous.
Frequently asked questions
Isn't this just deduplication?
Deduplication is one half of it. Identity resolution also handles the opposite error — a single record that actually contains two separate assets — which simple dedup never catches because there is only one row to begin with.
Why resolve identity before analysis rather than after?
Because every downstream number inherits the identity decision. Fixing it first stops one error at the source; fixing it later means hunting the same distortion through comps, models, and portfolio rollups it has already contaminated.
See identity resolution running on your pipeline.