Check Encoding Before Importing Company Names
Use a multilingual UTF-8 fixture to detect damaged company names, distinguish normalization from corruption, and verify the entire import route.
10/23/20263 min read
An international company name can survive an import without remaining the same name. Replacement characters, missing accents, or changed scripts may leave a record searchable only under its damaged form. The difficulty is that a file can contain the right number of records while every affected name has been interpreted incorrectly. Row-count checks do not detect that failure.
Treat character encoding as a contract between the exporter and every reader of the file. The receiving program must interpret the bytes in the intended way. Choosing a font that displays more characters cannot repair bytes that were already decoded incorrectly.
Separate encoding damage from deliberate normalization
Keep the source name, intended display name, and comparison name separate when your matching process needs all three. Removing an accent for search convenience is an editorial transformation. A sequence of garbled characters is a decoding problem. Combining these issues into one cleanup rule makes it hard to establish which name was actually supplied.
HubSpot's format-import guidance specifies UTF-8 for foreign-language characters. That is a destination-specific requirement, not proof that every file labeled CSV uses UTF-8. Verify the actual encoding produced by your exporter and the setting used by your import path.
If the source owner cannot establish the encoding, stop before saving over the file. An approved technical reviewer can inspect a controlled copy. Repeatedly trying encodings and keeping whichever output looks plausible is unsafe for scripts the reviewer cannot read.
Use a fixture with deliberately different failure modes
The following invented business labels are synthetic examples, not references to actual companies. Store their expected Unicode text in the test specification and assign fixed row references.
U01: Côte Atelier. Check the circumflex and final spelling.
U02: Łąka Studio. Check that both Polish characters remain intact.
U03: 東京 Lab. Check that the two Japanese characters survive.
U04: شركة Test. Check the Arabic word and the placement of the Latin label.
U05: Meridian “North”. Check the typographic quotation marks.
U06: two visually identical versions of Café Test, one using a precomposed accented character and one using a combining accent.
Include a blank name and a deliberately invalid input in a separate negative-test file if your technical owner can construct it safely. The expected result should be rejection or an explicit warning, not silent replacement with an apparently acceptable name.
Check both visible text and exact values
First, compare the parsed names with the expected fixture before any CRM import. Next, import the permitted fixture into an isolated test area and export the same records. Match on the U references, not on company names, because names are the values under test.
A native reader or qualified reviewer should check script display where necessary. In parallel, compare exact character sequences with an approved text tool. These checks answer different questions: visual readability and preservation of the intended value. Neither should be substituted for the other.
U06 requires an explicit decision. Two strings may render identically while using different Unicode sequences. If a comparison process intentionally normalizes them, document the normalization form and retain the original where justified. Do not call that transformation a lossless byte-for-byte copy. Conversely, do not merge two companies merely because normalized names become identical.
Diagnose an illustrative mismatch
Imagine U01 exports correctly from the source but appears damaged after a colleague opens and resaves the file in a legacy spreadsheet configuration. The source and CRM are not enough to diagnose the problem. The intermediate file identifies the conversion boundary responsible.
The repair is to regenerate that intermediate from the intact source through the approved UTF-8 route. Hand-editing the six fixture names would make the demonstration look successful without proving the remaining production names are safe. Repeat the whole fixture after changing the setting, including the mixed-direction label and combining-accent case.
Record the source encoding, selected importer encoding, intermediate application, comparison method, and reviewer. Attach pass or fail results for each fixture row. If the destination intentionally restricts a script, obtain a documented alternate representation and retain the original through an approved field rather than silently transliterating it.
Acceptance covers the tested path only. Reopening the export in another application, using a different locale, or feeding a new integration may introduce another conversion. Keep the fixture available for those changes, and do not broaden permissions or collect additional personal information merely to make the test more realistic.
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.