You've built a contact list of 500 people in Notion. You've linked deals, mapped relationships, added custom fields. Then you export to CSV—and watch nine critical fields flatten into single text columns, disappear entirely, or leave orphaned references behind. The export dialog doesn't warn you. Your backup CSV is now structurally useless. This post maps exactly which fields break, why Notion's export model fails for relationship data, and a concrete 30-day recovery plan to rebuild your pipeline in a system designed to move. The nine fields that orphan on export Notion's export doesn't fully flatten relations, lookups, and formula fields—it breaks them into comma-separated text that can't be parsed back. Here's what you lose: Deal relations (many-to-many). A contact linked to 3 deals exports as "Deal A, Deal B, Deal C" in a single cell. On import to any other platform, these are strings, not links. Your contact is no longer tied to deals; you have a memo field that once meant something. Company hierarchy (rollups and relations). If you've nested "CEO" under "Company A" and built a rollup to count team size, that rollup vanishes. You export the raw count or a flattened text list of company names—no structure. Custom status fields with conditional formatting. Notion lets you color-code "Negotiating", "Warm", "Cold" with visual rules. Export gives you plain text. Your pipeline-stage color coding, your at-a-glance heat map—gone. Linked formula fields. If you have a formula that calculates "Days Since Last Contact" by looking up the latest activity record, the export can't compute it in real time. You get a static snapshot, and the moment you add new activities, it's wrong. Last-modified timestamps on linked records. Notion tracks when a deal was updated. But the export doesn't capture the relationship between "Contact Alice" and "Deal X updated 3 days ago." You get Contact Alice and Deal X separately, with no time reference tying them. Multi-select fields with relation context. You might tag a contact with "Referred by Partner Y" and also link them to Partner Y's record. Export treats these as two independent fields; the export has no way to say "this person was referred by the linked Partner Y, not just the text 'Partner Y'." Lookup fields that pull from related records. If you've set up a lookup to show "Contact's Company's Annual Revenue" by traversing relations, the export gives you only the final value, not the link chain that created it. Rebuild that lookup in a spreadsheet, and it won't auto-update. Email and phone deduplication logic. If you've used a database relation to flag duplicate contacts by email, the export erases that logic. You get 500 rows with no indication that rows 47 and 203 are the same person. Custom permissions and view filters. If you've hidden "Lost" deals from the default view or restricted edit access to certain fields, none of that survives the export. Your security model and user experience are both reset to zero. Export Notion CRM with 500 contacts, and you will lose the links between contacts and deals, company structure, pipeline stage color coding, formula-based timestamps, and any relationship logic. What you get is nine CSV columns of isolated data, no relationships to rebuild from. Why Notion's export fails for relationship-heavy data Notion is a document database. It's brilliant for capturing related tables and showing you a networked view within Notion itself. But the export format—CSV—is a flat, row-by-row format from 1985. It has no native way to express "this cell points to that cell in another table." When you export, Notion has three choices for each relation or formula field: Flatten it to comma-separated text (useless for re-import). Leave it blank (data loss). Convert it to a static snapshot (broken when source data changes). Notion chooses all three, depending on field type. Your export becomes a Frankenstein's monster: some columns are text, some are empty, some are stale numbers that haven't updated since you hit the export button. At 500 contacts, this breaks. At 5,000, it's a catastrophe. Your pipeline stops being a graph and becomes a pile of disconnected rows. Audit: Mapping what you've actually lost Before you rebuild, you need to know exactly what left with the export. Create a test recovery folder in your workspace and follow this 2-hour audit: Export one table at a time. Export Contacts, Deals, Companies, Activities separately. Don't try to understand a 20-column mega-export. Open the CSV in a text editor, not a spreadsheet. Spreadsheet tools often hide the real structure of comma-separated data. Open it as plain text and look for any cell that contains a comma inside quoted text—that's a flattened relation. Count every deal link that exported as text. Search for "Deal " in the Contacts CSV. For each one, check: is it still linked to the Deal table's database in Notion, or is it now just a string? If it's just a string, you'll need to manually re-lin