You built a Notion database. It held 300 contacts, 45 active deals, and two years of activity notes. Then you tried to export it. What arrived in your new CRM was a skeleton. Relationship fields that linked contacts to companies went missing. Rollup fields that summed pipeline by stage flattened to null. Formulas that calculated deal age vanished entirely. You've now got raw text, broken references, and no automated way to rebuild what Notion's relational engine created for you. This playbook maps exactly which nine field types orphan on export, which survive intact, and a 30-day automation-driven recovery plan to rebuild your contact graph, deal pipeline, and activity log in a real CRM . The Nine Fields That Vanish (And What Actually Survives) Notion exports to CSV. CSV is flat. Anything Notion did with relationships, lookups, or math dies in translation. Fields that orphan completely: Relation fields (Company linked to Contact): Export shows only the raw link ID, not the name. You get a meaningless string like "rec_9x2k8L9m" instead of "Acme Corp." Rollups (sum of deal value per company): Vanishes entirely. The calculation exists only in Notion's memory. Lookups (pull contact name from linked company): Disappears on export. You lose the denormalized copy. Count fields (how many deals per contact): Exported as static numbers that are now stale and unlinked. Formula fields (calculated days since last touch): All formulas become frozen values—and only if Notion rendered them before export. Many don't, leaving blanks. Created/edited timestamps in Notion: Often export as formatted text, not Unix timestamps. Your CRM's time filters won't parse them. File attachment URLs : Notion exports the file name, not the URL. Links are broken on arrival. Multi-select tags with colors : The color coding vanishes. You get plain text. Email fields with validation : Notion marks invalid emails; the export has no such marker. Duplicates and typos sail through. Fields that survive (mostly) intact: Text fields (name, title, company name) Single-select dropdowns (stage, status, source) Phone numbers (as text) URLs (as plain text links) Numbers (revenue, deal size) — but lose any currency symbol Checkboxes (as true/false) Dates (if formatted consistently; timezone data is lost) The rule: If it required Notion's engine to compute or relate, it doesn't export. If it's inert text or a simple type, it survives. Why This Matters Now: The Hidden 30-Day Cost You're tempted to live with it. "I'll rebuild the relationships manually as I work." That decision costs you roughly 8–12 hours per week for the next month, buried in re-linking contacts to companies, re-sorting deals by owner, and manually dating activity that used to be timestamped automatically. Worse: every deal touched in that 30 days gets a stale "last activity" marker. Your forecast and pipeline metrics lie for a month. A structured 30-day recovery plan—with no-code automations handling the repetitive parts—compresses that work to 3–4 focused hours and keeps data accuracy alive. The 30-Day Recovery Playbook Days 1–2: Audit and map Before you import anything, answer these questions: How many contacts? How many are actual prospects vs. internal notes or test records? How many deals? Which stage field from Notion becomes your CRM's pipeline stage? Which linked records matter most? (Company-to-Contact is almost always first.) Do you have activity logs (calls, emails, meetings logged in Notion)? Export those separately to a second CSV. Which Notion fields have formulas or rollups that drive your business logic? Document the formula so you can rebuild it in your CRM (or live without it). Example: If Notion had a formula field "Days Since Last Activity," note that it's gone. Your CRM will calculate it automatically if you log activities with timestamps—so focus on recovering the activity log cleanly. Days 3–5: Clean and reshape the export Your CSV is messy. Spend these days: Remove test records, duplicates, and incomplete rows (no name, no email). Standardize company names. "Acme" and "Acme Corp" and "ACME CORP" become one. Separate linked-record IDs from text. If a relation field exported as "rec_9x2k8L9m | Acme Corp," split it into two columns: one for the name (which you'll use to re-link), one for a note that it was linked. Convert Notion dates to a CRM-friendly format (YYYY-MM-DD, no timezone). For file attachments, note which records had them. You'll re-upload after import if they're critical. This is manual work, but it takes 4–8 hours and is worth every minute. A clean import prevents 20 hours of cleanup later. Days 6–10: Bulk import contacts and deals Import the cleaned CSV into your new CRM. At this stage: Contacts land with name, email, phone, company name (as text, not linked yet), and static fields like "source" and "status." Deals land with contact name, deal value, stage, and close date (if you had one). All timestamps are null or generic (import date). You'll recover t