Notion CRM works beautifully at 50 contacts. At 200, you're still fine. Around 500, the cracks start. Filtering on linked fields becomes glacial. Rollups timeout. And when you try to export your data to move to a real CRM, you discover that relationship maps, custom hierarchies, and formula-driven fields simply don't come through on the other side. This isn't a Notion bug—it's a design reality. Notion is a document database with SQL-like queries bolted on. It was never engineered for high-cardinality contact operations. And if you've built your pipeline on Notion because it was free and flexible, the moment you need to scale, you hit a wall that no workaround fixes. Here's what actually breaks, why it breaks, and how to migrate without torching your relationship data. The performance cliff: where Notion CRM degrades The wall appears at different contact counts depending on your setup, but the pattern is consistent: 0–200 contacts: Notion performs. Filters load in under 2 seconds. Linked fields render smoothly. You forget you're in a document database. 200–500 contacts: Slowness creeps in. Filtering on linked records (e.g., "show all deals linked to this account") takes 5–10 seconds. Rollups recalculate visibly. Multi-select filters lag. 500+ contacts: Filtering becomes unusable. A simple query like "all deals linked to accounts in [Region]" can take 20+ seconds or timeout. Syncing via Zapier or Make breaks intermittently. Database views load incomplete data. The root cause is Notion's architecture. It's optimized for flexibility and rich documents, not transactional queries at scale. When you have hundreds of linked records, Notion's database engine has to join and filter across multiple tables in real-time—something relational databases do in milliseconds but Notion does in seconds or not at all. At 500+ contacts in Notion, a linked-record filter that should take 100ms now takes 10+ seconds. That's not a bug—that's the architectural ceiling. What actually breaks on export: the data integrity problem You decide to migrate. You export your Notion CRM to CSV. But not everything comes through. Fields that export cleanly: Text fields (name, email, phone) Single-select fields (status, industry) Multi-select fields (tags, products) Date fields Numbers and formulas that resolve to scalars Fields that orphan or export as broken references: Linked records: Export as IDs or names only. If you have "Account" linked to "Contacts", the export shows the contact name but not the relationship context. You lose the ability to say "this contact reports to this account manager." Rollups and formula fields that depend on linked data: Export as values frozen in time. Formulas like "count of open deals linked to this account" don't port to the new system—you have to rebuild them. Relation hierarchies: If you've modeled a deal pipeline as "Deal → Opportunity → Account → Parent Account", that multi-level hierarchy flattens on export. You get a CSV with one row per deal, but the nesting logic is gone. Rollup functions: "Sum of ACV from linked deals" exports as a number, but the formula itself is lost. If your new CRM has different rollup syntax, you have to manually rewrite it in 50+ accounts. Custom formula fields: "Days since last contact" or "NPS score formula" exports as static values. Formulas don't port. You'll have to recalculate or rebuild them in the new system. Database relations with unique constraints: If you've used Notion relations to enforce "one account per contact", that logic is gone on export. Your CSV has no referential integrity. The practical result: your exported data is a flat file with orphaned foreign keys. Relationship depth is gone. You have to manually map and clean the data before it can load into a real CRM. Airtable: incremental step, still hits a ceiling Airtable feels like Notion's slightly more robust cousin. It has better performance on linked records and rollups, and it's a smoother export because Airtable uses more relational semantics than Notion. Pros: Linked records and rollups work faster up to 1000+ records CSV export includes linked record IDs, which is marginally better than Notion's Automation engine (via Zapier or Airtable's native automations) is more reliable Formula syntax is closer to SQL, so formulas are more portable Cons: Still not a CRM. There's no notion of pipeline, deal velocity, or forecasting. You're building all of it yourself. Airtable scales to 100K records in a single table, but the UI degrades past a few thousand. Filtering and view performance get sluggish around 2000–5000 rows depending on linked fields. Pricing scales with usage. At 500 contacts + 5 tables + team collaboration, you're at ₹3,000–5,000/month. Real CRM pricing kicks in around the same point. You still have to rebuild relationship logic, pipeline stages, and forecasting. Airtable is a platform; you're the developer. Airtable is a good intermediate step if you're at 200–400 contacts and want breathing r