Fix Notion Database Property Type Conversion Losing Specific Data Points

You changed a property type in a Notion database, and now some of your data points are missing or incorrect. This happens because Notion does not always map every value from the old property type to the new one. This article explains why specific data points get lost during property type conversion and shows you how to recover or prevent data loss.

The root cause is that Notion only converts data that fits the new property type format. Values that do not match are silently dropped or changed to a default. You will learn a safe method to test conversions, back up your data, and restore missing entries.

By the end of this article, you will know how to avoid data loss when changing property types and how to recover lost data points using database backups and formulas.

Why Notion Loses Data When You Change a Property Type

When you change a property type in a Notion database, the software tries to map existing values into the new format. Notion only keeps values that are valid for the new property type. Invalid values are either set to a default or removed entirely.

For example, converting a Select property to a Text property works well because any text can be stored. But converting a Text property to a Number property drops any text that is not a number. Dates, phone numbers, and mixed text disappear. Similarly, converting a Multi-select property to a Select property keeps only the first selected option.

Notion does not warn you before dropping data. The conversion happens immediately when you select the new property type from the drop-down menu. If you do not have a backup, the lost data is permanently gone unless you manually re-enter it.

Property Type Conversion Table: What Gets Dropped

The following table shows common conversions and which data points are lost.

From Property Type	To Property Type	Data Points That Are Lost or Changed
Text	Number	Any non-numeric text, punctuation, or spaces
Text	Date	Text that is not a valid date string
Multi-select	Select	All options after the first one
Email	Text	None, but the email validation is removed
Number	Select	Numbers are converted to text labels; decimal precision may be lost
Phone	Text	None, but formatting is preserved
URL	Text	None, but the link behavior is removed

Steps to Recover Lost Data After Property Type Conversion

If you already changed the property type and lost data, follow these steps to recover it from a database backup or by using formulas.

1. Check if you have a database backup
   Notion does not automatically back up databases. If you exported the database before the conversion, open the exported file in a text editor or spreadsheet application. Look for the original values in the property column you changed.
2. Re-add the original property as a new column
   If you have a backup, create a new property in the database with the same type as the original. Manually copy the values from the backup into this new column. Use the Ctrl+V shortcut to paste multiple values if you have them in a spreadsheet.
3. Use a formula to extract data from the changed property
   If you converted a Text to Number and some text was dropped, create a formula property that shows the original text if it exists. For example, if you still have a separate property with the original text, use prop("Original Text"). If you do not have the original data, you cannot recover it without a backup.
4. Restore from a database copy
   If you duplicated the database before the conversion, open the copy. Find the rows with missing data in the original database and copy the values from the copy into the original. Delete the copy after recovery.
5. Re-enter lost data manually
   If no backup exists, you must manually re-enter the missing data points. Use the page content or other properties to reconstruct the values. For example, if a date was lost, look for the date in the page body text.

If Notion Still Has Issues After the Main Fix

Converted Property Shows Blank Values for All Rows

This happens when the new property type cannot interpret any of the existing values. For example, converting a Text property that contains only words to a Number property results in all blank cells. The only fix is to revert the property type back to the original type. Notion allows you to change the type back, and all original values reappear because they were never deleted from the database — they were only hidden.

Select Property Shows Only One Option After Conversion

If you convert a Multi-select to a Select, Notion keeps only the first option from each row. To recover the other options, revert the property type back to Multi-select. All options reappear. Then create a separate Text property and use a formula to join the multi-select values into a comma-separated list. This preserves the data before you attempt the conversion again.

Date Property Shows Incorrect Values After Conversion

When converting a Text property to a Date property, Notion only recognizes dates in specific formats: YYYY-MM-DD, MM/DD/YYYY, or Month DD, YYYY. If your text uses a different format, the conversion fails silently. To fix this, keep the original Text property and create a formula that parses the date using the parseDate() function. For example, parseDate(prop("Original Text")) works if the text is in a recognized format.

Feature	Duplicating the Database	Exporting to CSV
Preserves all property types	Yes	Yes
Preserves relation and rollup values	Yes	No
Preserves file and image attachments	No	No
Allows testing conversion risk-free	Yes	No
Recovery speed	Fast (copy-paste)	Slow (manual import)

Duplicating the database is the safest way to test property type conversions. Exporting to CSV works as a backup but does not preserve relational data.

You can now safely change property types in Notion without losing data. Always duplicate the database first and test the conversion on the copy. Use a formula property to compare old and new values after conversion. For advanced protection, create a weekly database export using the Export option in the three-dot menu of the database.