At 300 contacts, Notion starts to feel tight. You've built relational views, rollup fields that count deals-per-contact, formulas that calculate deal velocity. It works until you try to export it. Then Notion's relational model—elegant in the interface—becomes a maze of orphaned data, broken references, and formulas that evaluate to #ERROR in your target CRM. The problem isn't Notion's fault. It's purpose-built for flexibility, not CRM migration. But if you're moving to a proper CRM like Pipedrive, QuickBooks, or Orin , you need to know which fields will survive the journey and which need to be remapped before you hit export. Here are the nine data types that break—and how to handle each one. 1. Relation fields (the core problem) Notion's relation fields link two databases—say, Contacts to Deals. On export, you get a CSV with contact IDs in one column and nothing that connects to the deal database. The relationship metadata vanishes. What breaks: When you import that CSV into Pipedrive or Orin, the CRM sees contact records and deal records, but no links between them. You lose the ability to say "this contact owns three deals." Safe path: Before export, create a formula field in your Contacts database that concatenates related deal IDs as text: =join(prop("Deals"), ", ") . Export that text field alongside your contact data. In your target CRM, map those comma-separated IDs to a "linked deals" field or use a bulk import script to rebuild the relationships programmatically. If you're moving to Orin's CRM , the import wizard can parse these reference strings and auto-link records in a single pass—ask support for the deal-link import template. 2. Rollup fields (calculation loss) You have a Contact record with a rollup that sums deal value across all linked deals. It shows $450K. On export, that field exports as a static number—which is fine until you add a new deal in the target CRM. The rollup doesn't recalculate. It's dead data. What breaks: All rollup fields export as frozen snapshots. They do not become living calculations in your new CRM. Safe path: Identify every rollup before export. Document what it calculates and which field it depends on. In your target CRM, rebuild it as a native field type: Sum rollups (e.g., total deal value): Map to a "Total Pipeline" field and let your CRM recalculate it from linked deals. Count rollups (e.g., number of deals): Export the count as metadata, then let the CRM recount linked records on import. Max/min rollups (e.g., largest deal): Export as reference data, rebuild as a formula in the target system. Most CRMs—Pipedrive, Orin, HubSpot—support native rollup-like fields. Don't try to port Notion's formula logic directly; instead, map the intent of the rollup and rebuild it in the target system's native language. 3. Formula fields with cross-table references You wrote: =if(prop("Deal Status") = "Won", prop("Deals").prop("Deal Value"), 0) . This pulls a field from a related record. On export, Notion can't serialize this—it either exports the evaluated result (a static number) or leaves it blank. What breaks: Any formula that reads from a relation field breaks. The cross-table reference doesn't travel with the export. Safe path: Before export, convert these formulas to helper text fields: Create a new text field called "Formula Export: [Name]". Populate it by exporting the formula result as a string (screenshot it, or use a script to pull the evaluated values). Export that static text field alongside your records. In your target CRM, map it as a reference field or note, then rebuild the logic as a native field type if needed. This is manual, but it preserves the data. Automation (Zapier, n8n) can accelerate this if you have hundreds of records. 4. Multi-select fields with custom formatting You have a Status field with options: "Active", "Inactive", "On Hold". Notion stores these as clean enum values. But you've also added colors and icons in the Notion UI. On export, Notion's CSV drops the UI formatting and exports only the text value. What breaks: The text exports cleanly, but the visual metadata (colors, icons, sort order) is lost. If your target CRM has a different set of status options or a different display order, the import won't auto-map them. Safe path: Before export, audit every multi-select field: List all options in your Notion database. Map them 1:1 to options in your target CRM (Pipedrive's pipeline stages, Orin's deal status, etc.). Export the CSV and use find-and-replace to align any mismatches. Test the import on a small batch (50 records) to confirm the multi-select values map correctly. Most CRMs accept multi-select as comma-separated values in a CSV; make sure your export format matches what your target system expects. 5. Lookup fields (read-only snapshots) A Lookup field shows related data at a glance—e.g., Contact records display the phone number from the linked Company record. On export, Notion exports the looked-up value as static text. It's not a ref