ListMatchGenie

Match companies or products

Matching organizations, suppliers, accounts, or SKUs works differently from matching people. This guide covers the profile choices, field weighting, and gotchas specific to non-person matching.

Matching people is largely about names. Matching companies, suppliers, accounts, or products is about identifiers and context — domains, SKUs, categories, street addresses that mean very different things than they do for people.

This guide covers matching non-person entities end-to-end, with the specific knobs to turn for company and product data.

When to use this guide

Use this workflow when you're matching:

  • Companies — B2B accounts, supplier lists, firmographic data, CRM account records
  • Products — SKUs, inventory items, catalog entries
  • Any non-person entity with identifiers different from person names (email, phone)

For person matching, see Match leads to CRM. For finding duplicates inside a single list, see Deduplicate a customer list.

Why company matching is different

Company and product matching face issues person matching doesn't:

Naming is creative

Acme Inc., ACME INCORPORATED, acme, LLC, Acme Corp., The Acme Company, and Acme (formerly Beta Industries) can all refer to the same company. Person names don't have this — John Smith has much less variability than Johnson & Johnson Inc..

Match profiles for companies handle this with specific rules:

  • Suffix stripping (Inc, Incorporated, LLC, Ltd, Corp, Company, Co)
  • Article removal (The, La, Le, Der)
  • Ampersand normalization (& ↔ and)
  • Tokenization so word order matters less (Smith Jones & Co ↔ Jones Smith & Co)

Domain beats name

If two records have the same primary domain (e.g. acme.com), they are overwhelmingly likely to be the same company — more reliable than any name comparison. The Company profile weights domain as the strongest signal when it's present.

Address is weak signal

Companies move, operate multiple locations, and list different addresses for billing vs shipping vs headquarters. Address carries far less weight in the Company profile than in Person.

Product matching is about SKUs

For products, the SKU (when present) is near-definitive. Product name comparison uses token-based matching so 12oz Blue Mug Ceramic matches Blue Ceramic Mug, 12oz despite different word order. Numeric tolerance handles 12.5in ~= 12.50" ~= 12-1/2 inch.

The workflow for companies

  1. Prepare both files

    For the source and master, include as many of these columns as you have:

    • Company name (required)
    • Domain (huge help — include if you have it)
    • Street address, city, state, ZIP
    • Phone
    • Industry / SIC / NAICS code
    • Any internal ID (your CRM's account ID, DUNS, etc.)

    Clean each file individually first if you suspect internal duplicates.

  2. Upload with correct profiling

    On upload, verify column detection:

    • Company name column profiled as company_name
    • Domain profiled as domain (not generic URL)
    • Address columns correctly identified as their parts

    If the Genie misidentifies a column (e.g. a "parent_company" column as generic text), override it.

  3. Pick the Company profile

    Select Company as your match profile. The defaults are good for most B2B matching. Leave threshold at 70.

    Custom tuning you might want:

    • If domain is present in both files, consider raising domain weight further. Company domains rarely lie.
    • If you're matching branches of companies where the same domain is shared, lower domain weight instead and raise address.
    • If industry code is present in both files, add it as a low-weight tie-breaker.

  4. Run and review

    The match runs the same way as person matching. On Review, pay extra attention to:

    • Same name, different location — often legitimate branches. Use the location context to decide whether to treat them as one or two entities.
    • Same domain, different name — usually a rebrand. These are generally the same entity.
    • Similar name, totally different industry — usually different entities. A "Sunrise Bakery" and a "Sunrise Consulting Group" are unrelated.

  5. Export with enrichment

    The exported file has your source rows plus:

    • Every master column from the matched record (useful for enrichment)
    • The _lmg_ metadata including match method and score
    • Any _lmg_notes you added during review

    For CRM account reconciliation, filter to _lmg_match_status = match to get a clean lookup table of "source account → master account ID".

The workflow for products

Products follow the same overall workflow. The key differences:

Use the Product / SKU profile

This profile weights SKU as the strongest signal when present, with product name next, then brand and category as supporting signals.

Clean SKUs carefully

SKUs often have trailing suffixes (ABC-123-V2, ABC-123-V3 are often the same product). Decide ahead of time whether versioning matters to you and either preserve or strip suffixes in the column settings.

Use token-based comparison for names

Product names in the Product profile are tokenized and compared as sets of terms, not ordered strings. "Blue Ceramic Mug 12oz" and "12oz Blue Mug Ceramic" match because the token set is identical.

Numeric tolerance

For fields with numeric values (dimensions, weights, volumes), the profile applies a 2% tolerance by default. 12.5in, 12.50", and 12.5 inches all match. Override tolerance per-field in custom profiles.

Common gotchas

Parent/subsidiary confusion

Your master might list "Alphabet Inc" while your source has "Google LLC". Same real-world entity, different legal names. Handle this by:

  • Enriching your master with "also known as" columns (e.g. aka, parent, subsidiary_of)
  • Using the Company profile with relaxed thresholds (65), then manually reviewing the ambiguous cases

Suppliers with the same name

"Smith Manufacturing" in California and "Smith Manufacturing" in Texas could be the same company or two unrelated ones. The Company profile uses address and phone as tie-breakers, but if both are unknown, these go to review. Decide case-by-case; consider adding an industry column if your data has it.

Product variants

"Blue Ceramic Mug 12oz" and "Blue Ceramic Mug 16oz" are two products, not one. Make sure size/capacity/variant columns are weighted highly enough that they can discriminate.

Domain enrichment pays off

If your company data doesn't have domains, enriching it once (via Clearbit, ZoomInfo, or public WHOIS) makes every future match run dramatically more accurate. Domain is the most reliable company identifier short of a legal entity ID.