Test Case Sensitivity in Company Matching Rules
Separate display names from comparison keys, challenge case-only identifier collisions, and verify CRM matching before normalizing business records.
10/23/20263 min read
Lowercasing every field can make matching appear easier while quietly changing identifiers and display names that should remain untouched. The right question is not whether capitalization matters everywhere. It is which fields may use a normalized comparison value, which must preserve exact text, and how the destination's matching behavior treats them.
Run a case-sensitivity fixture before changing a matching rule. Keep the original value alongside any comparison form. A rule that successfully finds more candidates can still be unsafe if it collapses distinct entities or rewrites authoritative identifiers.
Classify fields before normalizing them
A company display name, an internal opaque key, a website address, and a user-supplied note serve different purposes. Do not apply the same transformation merely because all are stored as text. A verified source specification may permit case-insensitive comparison for one identifier and require exact comparison for another.
Name normalization should normally generate candidates for review, not establish company identity by itself. A name can change without a company becoming a new entity, while two separate companies can share the same normalized name.
Salesforce's external-ID documentation describes supported matching using identifiers from outside the system. Confirm the actual field and matching configuration before assuming how case is handled. Test the chosen import tool rather than relying on a general expectation about the CRM brand.
Construct a fixture that can reveal false matches
Use synthetic records with deliberately similar values. This example's keys are fictional and their intended semantics are defined by the test owner.
C01: display name Harbor Works; comparison name harbor works; opaque key Ab9Q.
C02: display name HARBOR WORKS; comparison name harbor works; opaque key ab9q.
C03: display name Harbor Works North; comparison name harbor works north; opaque key HN12.
C04: a name containing an accented uppercase letter with an approved lowercase comparison form.
C05: a name requiring language-aware case review rather than a simplistic English-only rule.
C06: an evidence URL whose path contains mixed-case text; preserve the supplied verified URL.
For this fixture, C01 and C02 are intentionally different entities because their opaque keys are distinct under the fictional issuing system's specification. That prevents the test from equating visual similarity with identity. If a real issuer defines different semantics, use its documented rule instead.
Test candidate generation separately from merging
First, run the comparison transformation without updating production records. Inspect which pairs become match candidates. C01 and C02 may correctly enter a review queue because their names look alike, but they must not merge automatically. C03 should retain its meaningful North distinction.
Next, test destination matching with the opaque keys. Record whether Ab9Q and ab9q are accepted as distinct, rejected as duplicates, or matched to one record under the selected configuration. If the destination cannot represent the source distinction, redesign the key mapping before import. Do not alter one key arbitrarily until it passes.
Finally, inspect the display fields. The preferred company name should retain its approved capitalization and accents. A comparison transformation should not leak into customer-facing text simply because the same column was reused for both purposes.
Investigate a case-only collision
Suppose the illustrative importer treats the two keys as equivalent and updates C01 with C02's region. That is a failed identity test, not a harmless formatting difference. Stop the affected mode and examine the matching field's configuration with the administrator.
Possible resolutions include using a supported case-sensitive key, introducing an approved namespaced mapping, or selecting a different supported import route. Each requires a new pilot. The choice should preserve source identity and remain understandable to later operators.
Do not repair the problem by lowercasing the source master. That destroys the evidence needed to distinguish the records. Similarly, do not alter URLs indiscriminately; verify links as supplied rather than assuming every component is case-insensitive.
The finished rule sheet should list each field, whether normalization is allowed, the exact comparison operation, the preserved original, expected collision behavior, and human-review conditions. Approve the matching rule only when both false matches and missed candidates have been examined. Case normalization is a controlled comparison technique, not a general license to rewrite business data.
If you are also choosing research access, ask about Leads Automation’s independent Sales Navigator offer at US$15/month on WhatsApp. Confirm the available plan, eligibility and terms before ordering; this is not LinkedIn’s official price or an assurance that any particular integration is included.