Most CRM damage from an import is not caused by bad records. It is caused by bad mapping. A column lands in the wrong property, a picklist value does not match, and suddenly reports are wrong and workflows fire on the wrong people.
A short field mapping checklist before every import prevents most of this.
The mapping step is where a spreadsheet becomes CRM data. Treat it as a design decision, not a formality you click through. The person running the import should be able to explain where every column goes and why.
Inventory the columns first
List every column in the source file and write its destination next to it. If a column has no destination, decide whether to drop it or to create a property for it. Do not import a column "just in case," because unused properties clutter the object and confuse future users.
Mark each column as required, optional, or ignored. Required columns must have a value on every row or the import will skip records. Knowing this in advance stops a partial import that is hard to reverse.
Count the columns and compare that to the CRM's import limit. If the file is wider than the limit, split the import into two passes on a stable unique key.
Match data types, not just names
A field that looks like the same data can have a different type in the CRM. A phone number stored as text behaves differently from one stored as a number, and a date string in the wrong format imports as blank. Check the type of each destination property before mapping.
Currency and number fields need the right decimal and thousands separators. If the source uses a comma as a decimal separator, convert the column before import or the values will be rejected.
For dates, confirm the expected format in the CRM and reformat the source to match. Ambiguous dates such as 03/04/2026 are a common source of silent errors.
Reuse existing picklist values
Dropdown and picklist properties only accept values that already exist, unless the import is allowed to create new options. Open the property settings and compare the allowed values to the distinct values in your column.
Map each source value to the closest existing option, and list any value that has no match. Creating a new option for every small variation is how a picklist turns into a mess, so consolidate where the meaning is the same.
The HubSpot knowledge base explains how property types and picklist imports behave. Confirm the current behavior for the object you are importing, because the rules differ between contacts, companies, and deals.
Pick the deduplication key
Every import should have a rule for matching existing records. Email address is the usual key for contacts, and company domain is the usual key for companies. Decide whether a match updates the existing record or creates a new one.
For deals, there is rarely a clean unique key in a spreadsheet. Match on a combination such as deal name plus company, and review the preview before committing. A wrong match can overwrite a live deal.
Run the import preview and read the summary of new versus updated records. If the numbers are far from what you expected, stop and fix the mapping before the import runs.
Protect fields owned by automation
Some properties are written by workflows, integrations, or the sales team. Overwriting them with a bulk import can undo hours of work or trigger a workflow loop. Identify those fields and exclude them from the mapping.
If you do need to update an automated field, pause the responsible workflow first and document why. Restore it after the import and confirm it is running again.
Keep a short list of protected properties in your team's import checklist so nobody has to rediscover them each time.
Test with a small sample
Import ten to twenty representative rows into a test view or a sandbox first. Include rows with edge cases such as missing last names, unusual characters, and international phone numbers. Check how each one lands in the CRM.
Verify that the picklist values came through, the dates parsed correctly, and the deduplication rules behaved as expected. Fix the mapping and repeat until the sample is clean.
Only then run the full file. A five minute sample saves an hour of cleanup.
Document the mapping for the next run
Save the final mapping as a template or a note in the import checklist. Next month, the person running the import can start from the same mapping and only adjust for new columns.
Record the import date, the row count, the source file name, and the person who ran it. This is the audit trail you will want if a report looks wrong later.
Keep a rollback note
Before committing a large import, record the current record count and export a backup of the affected object. If the import goes wrong, a backup turns recovery into a restore rather than a manual repair.
Note the exact mapping and the import timestamp next to the backup so the two can be matched later. A rollback note costs a few minutes and removes the fear from a large import.
A consistent mapping process is a perfect task to hand to a trained VA, as long as the judgment calls about new fields stay with an owner. To have this routine covered, review our CRM upkeep service, or request a consultation with a sample of your next import file.
