You've built your contact database in Notion. It worked fine at 50 records, still okay at 150. Then you hit 300 and everything feels slower. Queries take five seconds instead of one. Rollups bog down. Filter operations hang. You decide it's time to move to a real CRM—Pipedrive sounds like the obvious next step. Stop. Before you export, you need to know what you're about to lose. Notion's relational model doesn't translate cleanly to Pipedrive's contact structure. Linked records—relationships between your contacts, companies, and deal records—either flatten into text on export or disappear entirely. Custom field history evaporates. Tags sometimes import as a single text blob. And if you've built automation or rollups that depend on those relationships, they'll break on arrival. This isn't a Pipedrive problem. It's a Notion export problem. And if you don't audit your data structure before migration, you'll spend three weeks after the move untangling orphaned records and reconstructing relationship graphs from Slack conversations. The 300-contact performance cliff Notion's contact database starts hitting real friction around 250–300 records, depending on your schema complexity. This isn't a documented limit; it's a physics problem. Notion's database engine wasn't built for transactional, high-query-volume workloads. It's built for collaborative documentation. When you cross 300 records with more than three linked record types and rollups, you'll notice: Filtering takes 3–5 seconds instead of instant Sorting by a rollup field hangs or times out Opening a record with 10+ linked items takes 2–3 seconds to render Bulk updates stall if you're updating across linked tables View load times creep past 10 seconds Users get frustrated. You add a second Notion workspace (mistake). You consider Airtable (better, but still not built for sales). Then you look at Pipedrive and think: this will be fast and structured. It will be. But not if your export destroys your data relationships in transit. What Notion exports—and what it doesn't When you export a Notion database to CSV, the export engine has to flatten everything. A spreadsheet row can't hold a link to another row in a separate table. So Notion makes a choice for each field type: Linked records: Exports as a comma-separated list of record titles, not IDs. Pipedrive can't re-establish the relationship. You get a text field that says "Acme Corp, TechStart, Founder Labs." That's useless for a relational CRM. Rollup fields: Export as the final calculated value, not the formula. If you had a rollup that summed contract values across linked deals, you get the number—but the link is gone, so Pipedrive can't recalculate it. If Notion recalculated that sum, Pipedrive won't know to do the same. Custom fields with history: Only the current value exports. If you tracked when a contact status changed or who last updated a field, that history disappears. Pipedrive has activity timelines, but they won't include Notion's historical context. Checkboxes and multi-selects: Export as text, sometimes concatenated into a single cell. Tags become one garbled field instead of discrete values. Pipedrive's import parser sometimes splits them back out, sometimes doesn't. Formula fields: Export as their result value. If a formula concatenated first and last name or calculated days-to-renewal, Pipedrive imports the output, not the logic. You'll need to rebuild formulas as custom fields or automations in Pipedrive. Created/modified timestamps: Some export, some don't, depending on your Notion plan and export method. If you've been tracking when a contact was added or last updated, that data is at risk. The hard rule: Notion exports data, not relationships. If your CRM schema depends on how records connect, Pipedrive won't import the connections—just the text labels. Mapping your relationship graph before export Before you export, you need to document what relationships actually exist in your Notion database and how Pipedrive will represent them. Do this in a spreadsheet or a shared document: List every linked record type. Companies, deals, projects, contacts-to-contacts, anything. Note how many records are linked on each side (one-to-many, many-to-many, etc.). For each linked type, decide the Pipedrive equivalent. Pipedrive's core objects are organizations (companies), persons (contacts), and deals. If you're linking contacts to companies, Pipedrive handles that natively. If you're linking contacts to contacts (to show relationships, advisors, referrers), Pipedrive doesn't have a native field—you'll need to use a text field, a custom field with a dropdown, or a separate activities/notes entry. Test the export. Export a small test database (20–30 records) with all your field types. Open it in a text editor, not Excel. Check whether linked records are comma-separated, whether multi-selects are intact, whether timestamps survived. Identify what you'll lose. Any relationship that Pipedrive doesn't nati