You hit 500 contacts in Affinity and the export looks clean until you paste it into your new CRM. Suddenly half your relationship logic—the paths that led to deals, the strength scores, the who-knows-who map—has evaporated. You have names, emails, maybe notes. What you've lost is the structure that made Affinity work: the relationship graph itself. This is not user error. Affinity's export is flat by design. It was built to map social graphs and interaction chains inside the platform. Move that data anywhere else and the architecture crumbles. A 500-contact migration exposes nine fields that don't port cleanly, and rebuilding them—or losing them—costs four to eight weeks of pipeline clarity and deal velocity. Here's what vanishes on export, why it matters, and how to rebuild it faster than you lost it. The nine fields that orphan on Affinity export Affinity stores relationship depth in nine distinct layers. None of them export as coherent data. Instead you get scattered notes, flattened arrays, or nothing at all: Network depth (degrees of separation): Affinity calculates how many steps you are from a contact—1st degree (you know them), 2nd (friend of a friend), 3rd (connection of a connection). This is metadata Affinity owns. On export, you get raw contact records with no degree field. You must manually re-derive it or lose the insight that separates warm intros from cold calls. Interaction chains: Every email, call, meeting, and action thread in Affinity creates a timeline. Export gives you activity count, not chain logic. You lose the narrative: "met at conference, then emailed twice, then scheduled call." Rebuilding this requires linking records in a way most spreadsheets and even mid-market CRMs cannot express natively. Relationship strength score: Affinity's "relationship strength" metric weighs frequency, recency, and mutual contact overlap. The algorithm lives inside Affinity's platform. Export gives you a number (0–100, usually), but without the formula, you cannot reproduce it, extend it, or trust it in a new system. Most teams lose this immediately. Shared connection IDs: When Affinity identifies that you and a prospect both know Jane Smith, that's a golden warm-intro signal. On export, shared connections are often collapsed into text fields or omitted entirely. In your new CRM, you have no way to query "all prospects I can reach through Jane." The warm-path logic dies. Last meaningful interaction date: Affinity distinguishes between "contact logged" and "last real touchpoint" (email opened, call happened, meeting occurred). Most exports flatten this to a single "last activity" date, or worse, to the raw timestamp without context. You lose the signal that tells you whether a deal is warm or cold-sleeping. Interaction type hierarchy: Not all touches are equal. A 30-minute call weighs differently than a forwarded email, which weighs differently than a one-liner reply. Affinity assigns weight internally; on export you lose the granularity. Your new CRM sees "contact" or "activity" but not the semantics that drive deal scoring. Unilateral vs mutual relationships: Does both sides know each other, or just you? Affinity tracks direction and mutual awareness. Exports often lose directionality. You export a contact and have no record that this person does not actually know you yet—only your intention to reach them. Hidden or archived relationship tags: Affinity lets you tag relationships ("warm," "cold," "venture," "competitor," etc.). These live in a separate relationship record, not the contact record itself. Exports often miss them entirely, or flatten them to unstructured text in a "notes" field. You get "warm" as a word, not as structured metadata you can filter and act on. Relationship context and deal linkage: In Affinity, a relationship is tied to deals and opportunities. That relationship record says "this person is decision-maker for Q3 pipeline deal." On export, you get the contact and the deal separately, with no link showing which role this person played. You must manually re-establish the relationship-to-deal mapping. The core problem: Affinity was designed to map your personal and organizational network. That graph lives inside Affinity. Export flattens the graph to rows and columns. You get the vertices (people) but lose the edges (the relationships themselves). Where these fields scatter—or vanish entirely Affinity's export formats (CSV, API, webhook) handle the nine missing fields differently: CSV export (most common) CSV gives you a contact row with standard fields: name, email, company, phone. Relationship depth, strength, and context fields appear—if at all—in a single text "notes" column, often concatenated and unstructured. Shared connections might be listed as a comma-separated string. Interaction chains are a timestamp list, not a narrative. Rebuilding from CSV requires manual parsing and a lot of guesswork. API export (Zapier, Make, custom integrations) Affinity's API surfaces rel