Notion works fine until it doesn't. You've got 400 contacts, 800 interactions, and relationship tags sprawled across three databases. Then you realize Notion's relational fields—the colleague-of, refers-to, competitor-of links that actually drive your pipeline—don't export as CSV without collapsing into text strings. You're now choosing between losing relationship depth or manually rebuilding 200 contact links in your new CRM. Neither is acceptable. This playbook shows you how to migrate from Notion to Orin's CRM without that loss. The core problem is Notion's architecture: relations live in a separate layer. Export your contacts table and the links evaporate. We'll solve it by exporting separately, mapping explicitly, and verifying completeness before you delete anything. Why Notion's CSV export breaks your relationships Notion stores relational data in a lookup layer, not in the row itself. When you export a contacts table to CSV, you get columns like name, email, phone—but the "related accounts" or "referred by" field becomes either blank, a text string of names, or a JSON array that your CRM can't parse natively. This isn't a Notion flaw; it's how relational databases work. But Orin's contact relationship model is built to handle this. It supports native relationship types (colleague, referrer, decision-maker, competitor, etc.) and links them by contact ID, not by text guessing. The migration path respects both systems. Export strategy: four separate datasets Don't export your Notion workspace as one mega-file. Instead, export four tables, each one atomic and verifiable: Contacts table (name, email, phone, company, title, custom fields). This is your anchor. Interactions table (contact ID, date, type—call/email/meeting, summary, linked contact if applicable). This preserves your conversation history. Relationships table (source contact, relationship type, target contact, notes). This is the bridge. Notes/activity log (contact ID, date, note text, owner). This is your institutional memory. Why separate files? Because Notion's rollup and lookup fields don't flatten cleanly into one export. By splitting, you create a staging area where you can verify each dataset independently before import. How to export each table In Notion: Open your Contacts table. Click the three dots (top right). Select "Download" and choose CSV. Open your Relationships table (or embedded relations view). Do the same. For interactions tied to contacts, create a filtered view that shows source + target. Export as CSV. For notes, export your activity log or create a rollup view of linked notes. CSV export. You now have four CSVs. Open each in a spreadsheet. Check for: Empty rows (delete them) Text fields that should be IDs (e.g., "John Smith" instead of a contact ID). Flag these for mapping. Lookup/rollup columns that came through as semicolon-separated lists. These need parsing. Timestamps in non-standard formats (Notion often exports them as text). Clean these now. It takes 30 minutes for 500 contacts and saves hours of import debugging. Map Notion relationships to Orin's native types Notion's relations are free-form; you probably named them "Referred by", "Competes with", "Team member", etc. Orin's CRM offers a bounded set: colleague-of, referrer-of, decision-maker-at, competitor-of, customer-of, vendor-of, and custom types. Create a mapping table in your spreadsheet. Two columns: Notion relationship label: "Referred by" Orin relationship type: "referrer-of" (and note direction: is it "John referred Alice" or "Alice was referred by John"? Direction matters for bidirectional relationships) Example for a 500-contact base: "Colleague" → colleague-of (bidirectional) "Referred by" → referred-by (directional; Alice was referred by John) "Works at" → colleague-of (at same company; can also be a company field) "Competitor" → competitor-of (usually bidirectional) "Decision maker at" → decision-maker-at (directional) "Vendor" → vendor-of (directional; John is vendor of Alice's company) The mapping step is where you'll discover data ambiguity. "Referred by" might mean "this person referred a deal" or "this person was referred to us by someone." Clarify now, before import. A single wrong direction across 50 links will take hours to untangle in Orin. Import process: staged and verified Log into Orin. Go to your CRM module. Use the bulk import feature (or ask Orin support for a batch upload if you're migrating 500+ contacts). Step 1: Import contacts first Upload your cleaned contacts CSV. Map columns: name → Name, email → Email, phone → Phone, company → Company, title → Job Title, custom fields → their Orin equivalents. Leave relationship fields blank for now. Import 50 records first as a test. Check that they land correctly. Look for: Names that split incorrectly (Smith, John vs. John Smith). Fix your CSV mapping. Phone numbers that lost formatting. Pre-format or use Orin's phone field type. Emails flagged as duplicates. Dedup your Notion export