Start with the CRM import requirements
Before editing the contact file, open the destination CRM documentation or import screen and identify required fields, accepted file formats, maximum file size, and supported column mappings. Different systems handle names, phone labels, owners, tags, and custom fields differently.
Download a sample template when the CRM provides one. Using its headers and expected value formats can prevent avoidable mapping errors. Do not assume that a spreadsheet accepted by one CRM will import cleanly into another.
Keep one contact per row
Each row should represent one person or organization record according to the destination model. Avoid blank separator rows, merged cells, repeated header lines, subtotals, and explanatory text inside the contact table.
If a source image or PDF produced several people in one row, separate them before import. CRM importers generally cannot infer where one person ends and another begins, and a malformed row can shift values into the wrong fields.
Use clear and stable column headers
Choose headers that map directly to CRM fields, such as First Name, Last Name, Email, Mobile Phone, Company, Job Title, Website, Source, Owner, and Notes. Keep the same header names throughout the complete file.
Remove duplicate columns and decide how to handle fields that the CRM does not support. Useful unsupported information can be combined into Notes or mapped to a custom field, but unrelated interface text should be removed.
Split names only when the destination requires it
Some CRMs accept a Full Name field, while others require first and last name columns. Splitting names automatically can produce errors for compound surnames, prefixes, suffixes, and naming conventions that do not follow a simple two-part pattern.
If separate fields are required, review uncertain names instead of relying entirely on the last word as the surname. Keep the original full name in a backup or secondary column until the import has been verified.
Normalize phone numbers
Use a consistent phone format that matches the CRM and your operating region. International format is often helpful for contacts across several countries. Preserve extensions in a supported field rather than attaching unexplained digits to the main number.
Spreadsheet software may remove leading zeros or convert long numbers into scientific notation. Store phone values as text and inspect several examples after saving the CSV. The visual spreadsheet and exported CSV should contain the same usable digits.
Validate email addresses
Trim spaces, remove accidental line breaks, and check that each email has a plausible local part, at sign, and domain. Extraction from images can confuse punctuation and similar-looking characters, so inspect addresses that appear unusual.
Do not invent missing emails or treat validation as consent to contact someone. The CSV should accurately represent the source and the lawful workflow in which the contacts will be used.
Standardize companies, titles, and categories
Small spelling variations can fragment reporting. Decide whether values such as Inc., LLC, Ltd., and regional office names should remain distinct or map to one standard company name. Apply the decision consistently.
Use controlled values for fields such as lifecycle stage, status, country, source, or lead type when the CRM expects a fixed list. An unrecognized value may be rejected, ignored, or create an unwanted new category.
Add source and ownership information
A Source column should explain where each contact originated, such as CRM screenshot, business card event, PDF directory, or contact list image. Include a batch or date when it helps trace the record.
If records must be assigned to sales representatives or teams, use the identifier expected by the CRM. Confirm whether assignment requires a name, email address, username, or internal owner ID before filling the column.
Remove duplicates conservatively
Compare normalized email addresses and phone numbers, then use names, companies, and sources to review possible matches. Do not delete records solely because two people share a common name or an organization switchboard.
When duplicate rows contain complementary information, merge the strongest fields into one complete record. Keep a separate worksheet or backup containing removed rows and the reason for each decision.
Protect existing CRM data
Understand how the importer treats matches. Some systems create a new record, some update an existing one, and others use email or an external ID to decide. A blank CSV field may overwrite a useful existing value in certain import modes.
Choose create, update, or upsert behavior deliberately. If the CRM supports an external ID, use it when available. Otherwise, test how matching behaves before importing records that may already exist.
Save a clean CSV
Save the final file using the character encoding recommended by the CRM, commonly UTF-8. Reopen the CSV after saving and check accented names, international characters, commas, quotes, line breaks, leading zeros, and multiline notes.
Keep the reviewed workbook separately because CSV does not preserve formatting, formulas, multiple worksheets, or review comments. The workbook is your editable source; the CSV is the controlled import artifact.
Run a small test import
Create a test CSV with five to ten representative records. Include complete and partial contacts, international phone numbers, a company, a title, a source value, and any custom fields. Map each column and review the imported records inside the CRM.
Check field placement, duplicate behavior, ownership, tags, character encoding, and whether blank values changed existing records. Correct the file or mapping before importing the full batch.
Archive the import materials
Save the source images or documents, reviewed workbook, final CSV, import date, destination CRM, and mapping notes. If an import must be corrected later, these files explain exactly what entered the system.
A dependable workflow is therefore: confirm CRM requirements, structure one contact per row, map and normalize fields, deduplicate, save and reopen the CSV, run a test import, then archive the reviewed source. This turns extracted contact data into a controlled business import rather than an uncertain bulk upload.