Compare two files by deciding what each file represents, mapping identity fields and applying explicit matching rules. Review the resulting record sets before filtering. A match on selected keys does not mean every value in both records is identical.
Start with the decision you need to make
A useful comparison begins with a business question: which incoming leads are not already in the CRM, or which customer identifiers appear in this export but not the previous one? The question determines which file is the working dataset and what result you want to retain.
Use File A for the incoming or working file and File B for the reference. Keep the source files separate. When the reference is a CRM export, check its date and scope; records excluded from that export cannot be recognized as existing matches.
Map fields that represent the same thing
Column names do not have to be identical. Email in File A may map to Email Address in File B. Confirm that both fields carry the same type of identifier, rather than matching by position or by a similar-looking header.
For multiple-field matching, verify every mapped pair. Requiring Email and Company to match can separate contacts whose company value differs. That may be appropriate, but it is a different rule from matching on Email alone.
Choose conservative matching rules
Trim surrounding whitespace or ignore capitalization where the fields support those choices. Inspect missing keys and duplicates within each file before interpreting comparison counts. Repeated keys can make the relationship between rows and unique records more complicated.
A normalized key match is not a cell-by-cell difference report. Two records can match on email while their phone numbers or company names differ. Do not infer that all fields agree, and do not use a key comparison as proof that the newer values are correct.
Read the three result sets
Only in File A means no matching key was found in File B under your selected rule. Matched records appear in both. Only in File B means the reference has a key not represented in File A. These descriptions depend on both the file contents and the matching configuration.
For a simple example with unique keys in each file, File A has 100 records and File B has 80. If 60 keys match, there are 40 new records in A and 20 only in B. The 60 matched keys account for one record in each file; do not add them twice when counting unique keys across both.
Preview the result you want to keep
For a new-lead import, retaining only the 40 new records in the example excludes 60 matched records from File A. The 20 records only in File B are not incoming rows and are not part of that output. Review samples from each result set before confirming.
In RowDesk, keeping only new records changes the working dataset, not the CRM or either original source file. Inspect the applied filtering step, undo during the active session if necessary and export a separate CSV. Use the destination system's own process for the eventual import.
- Verify the File A and File B roles.
- Check mapped fields and normalization.
- Inspect representative new and matched records.
- Confirm the final output before export.