ListMatchGenie

Match profiles

A match profile tells the Genie which columns matter, how strictly to compare them, and how to break ties. Picking the right profile is the most important decision in any match.

A match profile is a named bundle of settings that controls how the match engine compares two records. It answers questions like: Which columns should we compare? How important is each one? Is an exact match on email enough to accept without checking the name? Should we use phonetic matching? What counts as "close enough"?

You pick a profile on the Configure step of the match wizard. ListMatchGenie ships with preset profiles tuned for the most common customer situations. You can also customize any preset or save your own.

Why profiles exist

Different match problems have different shapes. Matching people by name and address is fundamentally different from matching companies by domain, which is fundamentally different from matching products by SKU. A single "fuzzy match" setting can't handle all three well — what's a 95%-good comparison for a person name is a 60%-good comparison for a company name (because "Inc" and "LLC" matter far less than "Smith" vs "Smyth").

Profiles let you tell the Genie what kind of thing you're matching, and the engine adjusts every downstream decision accordingly:

  • Which columns to compare
  • What weight to give each column in the composite score
  • What counts as a valid "block" key for candidate retrieval
  • Whether phonetic matching is appropriate (it helps for names, hurts for product codes)
  • Whether nickname lookup applies (helps for first names only)
  • Whether abbreviation expansion runs (helps for company and street names)

The built-in profiles

Person

Use when your records represent individual humans — customer lists, lead lists, contact databases, CRM exports.

Comparable fields: first name, last name, email, phone, street address, city, state, ZIP.

Key behaviors:

  • Nickname lookup is enabled: Bill and William match.
  • Phonetic matching is enabled: Smith and Smyth match.
  • Prefix/suffix stripping runs: Dr. Mary Jones, MD compares against Mary Jones.
  • Address abbreviations normalize: 123 N Main St compares against 123 North Main Street.
  • Email and phone, when present, are high-weight near-exact signals.

Best for: CRM dedupe, lead-to-customer matching, list uploads into marketing platforms.

Company

Use when your records represent organizations — supplier lists, account lists, firmographic data, B2B lead lists.

Comparable fields: company name, domain, street address, city, state, ZIP, phone.

Key behaviors:

  • Company suffix normalization: Acme Inc compares against Acme Incorporated and ACME, LLC.
  • Domain, when present, is the highest-weight signal — two records with the same domain almost always represent the same company.
  • Phonetic matching is disabled: company names are too often acronyms or invented, where phonetic codes produce noise.
  • Nickname lookup is disabled.
  • Address comparison is moderate-weight, since companies move and have multiple locations.

Best for: supplier reconciliation, CRM account dedupe, firmographic enrichment matching.

Identifier

Use when your records share a unique, reliable ID — email, account number, SSN, NPI (for healthcare), or any internal system ID.

Comparable fields: the identifier column only, plus optional name/address for sanity-checking.

Key behaviors:

  • Exact match on the identifier is sufficient to accept.
  • No phonetic, no fuzzy, no nickname lookup on the ID itself.
  • Optional "tie-breaker" comparison on name or address for audit purposes.
  • This profile runs in a single pass and is dramatically faster than the others.

Best for: matching new leads against a master where you already share an ID, reconciling cross-system records via email or account number, matching against a public registry (NPI, DUNS).

Contact dedupe

Use to find near-duplicates inside a single list, not across two lists.

Comparable fields: same as Person, but the file is matched against itself.

Key behaviors:

  • Every row is a source; every row is also a master.
  • The engine skips self-matches (row X won't match row X).
  • Results are grouped into "clusters" of probable duplicates rather than 1-to-1 pairs.
  • The confidence threshold is usually lower (60–65%) since within-file duplicates tend to be cleaner than cross-file matches.

Best for: cleaning a list before importing into another system, finding duplicates your CRM's built-in dedupe missed, consolidating records after a system merger.

Healthcare provider

Use when your records represent medical providers (doctors, clinics, pharmacies) and either record set includes an NPI, specialty, or practice identifier.

Comparable fields: first/last name, credentials, NPI, specialty, practice name, address.

Key behaviors:

  • NPI is the highest-weight signal when present — an NPI match is essentially definitive.
  • Credentials (MD, DO, PhD, RN, NP, PA) are stripped from the name for comparison but shown in output.
  • Specialty comparison uses a taxonomy-aware match (pediatrics vs pediatric cardiology are related).
  • Phonetic matching enabled for last names.

Best for: matching against the NPI registry, reconciling provider lists, payer-provider matching.

Product / SKU

Use when your records represent products or inventory items.

Comparable fields: SKU, product name, brand, category, dimensions, price range.

Key behaviors:

  • SKU, when present, is near-definitive.
  • Phonetic matching is disabled.
  • Nickname lookup is disabled.
  • Token-based comparison on product names (handles variations in word order).
  • Numeric tolerance on dimensions (12.5in ~= 12.50").

Best for: catalog reconciliation, supplier product matching, inventory deduplication.

Custom

Start from one of the built-in profiles and adjust:

  • Add, remove, or rename fields to compare.
  • Change the weight of any field (sum doesn't need to equal 100 — the engine normalizes).
  • Override thresholds per field (e.g. "email must be exact, name can be 80% fuzzy").
  • Toggle phonetic, nickname, and abbreviation expansion independently per field.
  • Save the profile under a name for re-use across future matches.

Custom profiles are unlocked on the Starter plan and up.

Picking a profile

If you're not sure, the Genie recommends one based on what it sees in your column profiles. The recommendation appears on the Configure step above the profile dropdown — a one-sentence explanation like "Both files contain first_name, last_name, and email columns. Person profile recommended."

Override the recommendation if you have context the Genie doesn't. Common situations where the recommendation is wrong:

  • Your source file is people but your master file is households — person profile will over-match siblings at the same address. Use custom with address de-weighted.
  • Your source file is leads but some rows are companies and some are sole proprietors — run two passes with different profiles and union the results.
  • Your "identifier" column isn't actually unique in the master — switch from Identifier to Person and use the ID column as a tie-breaker.

When in doubt, start loose and tighten

Run your first match with the default profile at 70% confidence. Look at the review queue — if it's huge, tighten the threshold or pick a more specific profile. If it's empty, loosen. Two runs take ten minutes; getting the profile exactly right on the first try takes an hour of configuration.