Deduplicate a CRE Owner List Without Merging Unrelated Owners

Deduplicate a commercial property owner list by identifying what each row represents, comparing jurisdiction-specific parcel and entity identifiers, and preserving the underlying observations. Collapse repeated imports only when they describe the same record. Keep different parcels, ownership periods and unresolved identities visible instead of merging them because their names or addresses look similar.

Decide what a duplicate would mean

A sourcing export can mix tax parcels, buildings, legal entities and proposed contacts. Each has a different identity. Removing duplicate mailing addresses from that export may delete separate owners who use the same professional service. Removing duplicate entity names may discard separate properties held by one company.

Start by naming the output. A parcel inventory needs one current view per identified parcel, with links to older observations. An entity directory needs one row per verified legal entity. A research queue can contain several unresolved candidates for either. Those are separate tables, even if the first version lives in one spreadsheet.

The owner-portfolio research guide explains how to establish the relationships. This exercise deals with the next step: cleaning repeated observations without erasing those relationships.

Preserve the raw fields before standardizing them

Keep the source name, source record identifier, jurisdiction, observed parcel identifier, recorded entity name, record date and retrieval date. Give each imported observation a stable local row ID. Normalized fields belong beside the original values, not in place of them.

Store parcel identifiers as text. Leading zeros and punctuation can be meaningful within a local scheme. A spreadsheet that turns 0007 into 7 has already changed the input. Do not apply a punctuation-removal rule across counties until you have checked how each jurisdiction defines its identifiers.

For an entity, retain its formation or registration jurisdiction and registry document number when available. Florida's corporation-record search guide distinguishes entity-name searches from document-number searches and officer or registered-agent searches. Use the matching record to inspect identity and stated roles; a similar name in the results is not sufficient support for a merge.

Name cleanup can generate candidates: case folding, consistent whitespace, documented abbreviations. It should not make the final ownership decision. Keep a missing identifier missing, and route it to review.

Use separate keys for observations, parcels and entities

An exact imported-row repeat can be removed from the working view while its import history remains in the log. A second observation from a later date is a version, not automatically a duplicate. Two records can agree on the parcel and disagree on the owner because a transfer occurred or a source is stale. Keep the conflict until the documents resolve it.

Avoid a single fuzzy-match score that controls every merge. A name match can help prioritize review; it should not outweigh contradictory registry identifiers. Likewise, an address match should not silently join legal entities.

Worked exercise: eight rows, several different decisions

Everything in this exercise is fictional, including the jurisdictions, identifiers and entity names. It contains no real owner contacts. The packet supplies the source observations and an answer key.

The result is seven retained observations after suppressing one repeated import. Five distinct parcels are established in the current view, linked to four distinct recorded legal entities. The unresolved row is excluded from those confirmed counts. These totals describe the supplied exercise only, not beneficial-owner counts or real-world coverage.

Download the owner-list deduplication exercise. It contains the raw CSV, expected decisions, current parcel view and blank review log. Import identifiers as text when opening the CSV files.

Make every merge reversible

Record the surviving row ID, suppressed row ID, reason, source references, reviewer and review date. Preserve the original observations. Reversibility matters when a later document shows that two apparently identical records represent different entities or different periods.

Keep separate outcomes for an exact repeat, an accepted identity match, a historical version and an unresolved candidate. A rejected match is also useful: it prevents the same weak suggestion from returning in the next refresh.

If a downstream contact list groups several properties under one verified entity, retain the entity-to-parcel table. Deduplicating the contact destination should not erase which properties the research concerns. Contact enrichment and any outreach decision remain separate steps; cleaning a table does not authorize a message.

Test the export, not just the matching screen

Ask the workflow to propose decisions on the practice packet, then inspect the resulting CSV. The difficult checks are whether R4 stays separate, R6 remains historical and R8 remains unresolved. Correctly removing R2 alone is a weak test.

An instruction you can try: "Use only the supplied fictional evidence. Identify exact repeated observations separately from entity matches. Preserve distinct jurisdictions and dates. Return a decision and supporting row references for each candidate. Leave unsupported identity links unresolved. Do not search for people or infer beneficial owners."

Re-run the same input and confirm that no new parcel or entity records appear. Then change one observation's date and owner evidence. The workflow should propose a version review, not overwrite the old result without explanation. Report false merges and missed duplicates separately so an attractive reduction in row count cannot hide lost records.

Connect the cleaned list to a useful next step

The public sourcing workflow walkthrough shows the broader research handoff. When reviewing a vendor or a custom workflow, ask to inspect the retained evidence and exceptions alongside the final property list.

To see property-level output for your mandate, request an off-market sample. That page explains the sample boundary and the separate owner-resolution engagement. The useful question is whether the output supports your next research decision, with enough detail to check it.

Frequently asked questions

Is a shared mailing address enough to merge two owners?

No. It can generate a research candidate, but professional agents and service providers can serve unrelated entities. Verify the entity identity and preserve contradictory evidence.

Should one entity holding two parcels become one property row?

No. Preserve the entity-to-parcel relationships. An entity directory can have one entity row while the parcel inventory retains both parcels.

How should an ownership transfer affect deduplication?

Retain the earlier observation as history and verify the later ownership record. A changed owner or observation date is not automatically a repeated import.

Does this exercise identify real beneficial owners?

No. It uses invented records and stipulated registry identities. It neither establishes beneficial ownership nor measures a model against real property records.