Trim CRM Excel Exports Before You Load Them into SQL

Salesforce account name trailing space breaks a JOIN

Accounts export carries Acme Corp (trailing space). The Opportunities table has Acme Corp. The join on account name silently drops every Acme opportunity. Trim the Accounts export before staging.

accounts.xlsx (before):     accounts-processed.xlsx (after):
account_name                account_name
Acme Corp·  (trailing)      Acme Corp
Globex                      Globex

SQL after load:
SELECT * FROM opps o JOIN accounts a USING(account_name);
-- before: 0 Acme rows    after: all Acme opps join

HubSpot CHAR 160 defeats Excel TRIM but not this tool

A rich-text company field pasted from a web page ends in CHAR 160. A teammate already ran Excel =TRIM() and it 'looked clean' — but CHAR 160 survives Excel's TRIM and the unique constraint still admits a duplicate. This tool removes the edge CHAR 160.

Before (· = CHAR 160):
company
Initech·
Initech

Excel =TRIM("Initech·")  →  "Initech·"   (still distinct → 2 rows pass UNIQUE)
Whitespace Trimmer       →  "Initech"    (both collapse → constraint catches dup)

Pipedrive deal title trailing newline splits a GROUP BY

A deal title pasted from another tab ends in a newline. After load, GROUP BY stage shows 'Negotiation' and 'Negotiation\n' as two buckets. The trimmer removes the trailing newline.

Before (\n = newline):       After:
deal_title                   deal_title
Negotiation\n                Negotiation
Negotiation                  Negotiation

GROUP BY before: 2 buckets    after: 1 bucket

Full pre-load pipeline for a Salesforce contact dump

Order matters: trim edges first so case-folding and dedup operate on clean keys. Each step is a separate tool; all accept XLSX.

1. Whitespace Trimmer   strip edge spaces/CHAR 160      → contacts-processed.xlsx
2. csv-case-converter   lower-case the email column       (/tool/csv-case-converter)
3. csv-deduplicator     drop duplicate emails             (/tool/csv-deduplicator)
4. \copy staging FROM 'contacts-clean.csv' CSV HEADER;

Why not just TRIM() in SQL on every column?

Wrapping every text column in TRIM() inside your load mapping is brittle (you forget a column; TRIM in SQL also ignores CHAR 160 in many engines) and re-runs on every reload. Cleaning the file once before load is simpler and catches CHAR 160 that SQL TRIM misses.

Brittle:  INSERT ... SELECT TRIM(c1), TRIM(c2), ... (24 columns, easy to miss one)
          -- and Postgres TRIM(' ' ...) does not strip CHAR 160 by default
Clean:    run file through Whitespace Trimmer once → \copy directly

Many database TRIM() implementations strip only the ASCII space by default, leaving the non-breaking space (CHAR 160) that rich-text CRM fields carry. This tool's trim removes edge CHAR 160, so a value that stayed 'distinct' after SELECT TRIM(col) will collapse correctly here before it ever reaches the database.

12 Main St keeps its internal double spaces — the tool trims edges only. If your address-matching depends on single-spaced strings, follow with csv-find-replace using {2,} → before staging.

Acme and acme will still be two distinct keys after trimming — this tool does not change case. Fold case with csv-case-converter if your join or unique key is case-sensitive in the database.

Trimming makes near-duplicates identical, but it does not delete them. Run csv-deduplicator after trimming so the now-identical rows collapse to one before they hit a UNIQUE constraint.

If a single .xlsx has Accounts on sheet 1 and Contacts on sheet 2, only Accounts is processed and returned. Export each CRM object to its own file (Salesforce, HubSpot, Zoho, Pipedrive all let you do this) and trim them separately.

Because the file passes through a plain-text stage, currency/date cell formats are not carried to the output .xlsx. The displayed text is preserved. For loads this is usually fine (you cast types on import), but cast date and numeric columns explicitly in your COPY/LOAD mapping.

If a phone or account number with a leading zero is stored as text in the source, its text is preserved (only edge whitespace is removed). If it was stored as a number in Excel, the leading zero was already gone before export — that is a CRM/Excel typing issue, not a trimming one.

A full contact dump above 5 MB / 10,000 rows is blocked on the free tier. Either upgrade to Pro (50 MB / 100,000 rows) or split the export into chunks and trim each before loading.

A correctly quoted multi-line notes field is parsed as one cell; the trim removes only leading/trailing whitespace of that whole value, leaving internal line breaks intact. If the source CSV had an *unquoted* newline, the row was already malformed at export — re-export with proper quoting.

Some CRMs offer password-protected exports. An encrypted workbook cannot be read by the browser parser — export without encryption (or decrypt in Excel) before trimming.