A New Data Provider Can Accidentally Recreate Businesses the Directory Already Has

When a directory adds a partner feed, the incoming dataset may overlap heavily with owner-submitted and manually researched profiles. Publishing every row as new can double the database overnight.

Match before creation

Use the provider's identifiers where available, but also compare phone numbers, websites, addresses, coordinates, and business names against existing records.

No single field should decide every case.

Build confidence levels

An exact phone, domain, and address match may be safe to link automatically. A similar name at the same shopping centre may need manual review. A same-address record in a coworking building may be completely unrelated.

Decide which fields the provider can enrich

Matching a vendor row to an existing profile does not mean every incoming value should overwrite current information. Use source precedence and freshness rules by field.

Preserve the vendor key

Once matched, store the provider's record ID so future updates can target the same profile instead of repeating the matching process from scratch.

Feed deduplication protects both users and business owners. It prevents the directory from presenting multiple versions of the same organisation simply because those versions arrived through different pipelines.

Vendor feeds need a stable match key

Third-party providers may send the same business repeatedly with small variations in name or address. Relying on fuzzy text matching alone can create duplicates over time. Use a stable provider identifier where available and maintain a crosswalk to the directory's own internal ID.