You've spent three years building Affinity. Warm intros layer on warm intros. Relationship strength ticks up from passive to active to champion. Notes stack up—call recordings, lunch preferences, referral patterns. Then you hit export and watch nine critical fields vanish. The data leaves Affinity flattened, dimensionless, orphaned. Warm intros become binary links. Relationship strength evaporates. Notes either don't export at all or land in a single text block, useless to filter or act on. This isn't a bug. It's the architecture. Affinity's relational model—the part that made it powerful for relationship-driven sales—was never designed for clean exit. And if you're migrating to a CRM that understands relationship depth natively, or simply need portable history before your team loses access, you need a systematic recovery plan. This playbook maps the nine fields that break, why they break, and a 60-day rebuild sequence that restores relationship value without manual re-entry. The nine fields Affinity orphans on export Affinity's export dumps everything into flat tables. What you lose: Warm introduction chains. Affinity stores intros as entities—Person A knows Person B, who knows Person C—with metadata about strength and introduction date. The export creates a CSV row per intro, with no way to reconstruct the chain or understand who can actually reach whom. Relationship strength scores. In-app, strength is a live, filterable field (passive, active, champion). Export treats it as a label, not a sortable value. You can't segment by strength post-export. Interaction history attached to relationships, not contacts. Affinity ties notes, calls, emails to relationship records, not just contact records. The export decouples this—notes float free from context, or attach to the wrong person. Introduction source and date. Who introduced you? When? Affinity stores this. The export doesn't always carry it forward. Custom relationship fields. Any field you built to track relationship-specific data (deal stage, business unit, territory) becomes a generic contact property or vanishes entirely. List membership tied to relationships. Affinity lets you build lists on relationship criteria (e.g., all relationships with Acme in the last 90 days). Those criteria don't survive export. Interaction recency and frequency metadata. Affinity calculates how recently you interacted with a person and through which relationships. That calculated metadata doesn't export. Next step / action flags on relationships. Reminders, follow-up dates, and flags attached to specific relationships drop off. Company hierarchy linked to relationship ownership. In Affinity, you know who owns which relationship at which company, with org chart context. Export flattens this to "person works at company." Why this happens: The relational model vs. the flat export Affinity was built on a graph database foundation—entities (people, companies, relationships) with properties and edges between them. That's powerful for discovery and warm intros. But CSV and JSON exports are flat. Each row is a record, each column a field. A relationship—which is really an edge between two nodes—has to become a row, and when it does, context collapses. A single contact record can't hold multiple relationships. A relationship record can't hold all the metadata about each person it connects. Most CRMs store relationships as rows, too, but they design for it from the start. Pipedrive, HubSpot, and Orin's CRM treat a relationship (or deal, or connection) as a first-class entity with its own set of properties. The export honors that. Affinity's export assumes you're exporting contacts or companies, not relationships. The warm intro graph—the core value—gets sacrificed. The 60-day rebuild roadmap Week 1–2: Audit and categorize the damage Day 1–3: Run the export, catalog orphaned data. Export everything from Affinity: Contacts, Companies, Relationships (as separate CSVs if possible). Open each in a spreadsheet. Identify which nine fields are missing or degraded. For each field, ask: Is this field completely absent from the export? Is it present but flattened (e.g., relationship strength as a text label instead of a score)? Is it separated from its context (e.g., notes without the relationship they describe)? Day 4–7: Prioritize by revenue impact. Rank relationships by deal size, close date, or revenue stage. Your top 20 relationships—the ones attached to the largest or most urgent opportunities—need full recovery. Your next 50 need partial recovery (warm intro chain + relationship strength). The rest can be rebuilt on-demand, over months. Day 8–14: Map your target system's relationship model. If you're moving to Orin, Pipedrive, or HubSpot , understand how each stores relationships: Orin's CRM stores relationships as linked contacts with custom fields for strength, source, and interaction metadata. Multiple contacts can be marked as related to the same organization with distinct roles. Pip