Migrating from Reltio to Golden Suite

What carries over, what changes, and the actual migration sequence — written for teams who already know MDM and just need the substitutions.

If you're reading this, you've already decided that Reltio's TCO doesn't match the scope of your MDM problem. Or you want the matching engine in your own infra. Or both. This guide is the substitution table — what to keep, what to drop, what to do in what order.

Before you start, read our Reltio comparison for the honest "where they win" rundown. If your data plane is at Fortune 500 scale or you're locked into Reltio's enterprise integration set (SAP, mainframe, ESB), don't migrate. The rest of this guide assumes you've decided to switch.

What carries over

  • Your source data. Postgres, Salesforce, HubSpot, S3 — we connect to the same upstream systems Reltio does, just via Golden Suite's connector framework (22 types live). The data doesn't have to be re-extracted.
  • Your matching philosophy. If you've tuned blocking rules + scorer weights in Reltio, the same intuitions translate. Golden Suite's goldenmatch exposes the same primitives — blocking, fuzzy scorers, cluster thresholds — just in a Python/Polars API instead of a UI-driven rule editor.
  • Your survivorship rules. "Source priority", "most recent", "most complete" — same vocabulary. You'll re-declare them, but the semantics are 1:1.
  • Your audit posture. Reltio's audit log → Golden Suite's audit_log table, plus a cryptographic chain we don't think they have. SOC2 controls map.

What changes

  • Stewardship UI. Reltio's role-based steward UI is more polished than ours today. Golden Suite has review queues + the workbench, which is a different shape — fewer drag-drop interactions, more "approve / split / merge" decision queues. Some teams prefer it; some find it austere. Try the demo before committing.
  • Connector library. Reltio: 100+. Golden Suite: 22 modern. If you need a legacy mainframe or a niche ESB integration, you'll need to extend Golden Suite's connector framework or use a Kafka tap.
  • Pricing. Reltio's six-figure license + impl. fee → Golden Suite's $99/mo Pro or custom Enterprise. The savings are real; the trade-off is fewer hands-on people implementing for you.

Migration sequence

A realistic timeline: 2–4 weeks for a Pro-tier customer with 5–10 sources and clean Postgres. Add a week per legacy/SAP-shaped source.

Week 1 — Parallel landing

  1. Stand up Golden Suite alongside Reltio. Sign up at bensevern.dev, create a project, add your top 3 most-trafficked sources. Don't touch Reltio.
  2. Compare golden records. Run dedup on the new project. Diff the output against Reltio's current golden table — pick 50 records, look at where they disagree and why. This is where you'll find your survivorship rules drifting from defaults.

Week 2 — Rule parity

  1. Translate survivorship rules. Reltio's UI rule editor → Golden Suite's survivorship editor. One field at a time, replicate the rule, run dedup, diff against Reltio.
  2. Translate matching thresholds. If you've tuned Reltio's match scores, look at goldenmatch's scorer-weight outputs in the postflight panel. The numbers won't be identical (different algorithms), but the order of which fields drive matches should be similar.

Week 3 — Cutover prep

  1. Stand up the remaining sources. All 5–10 should now be live in Golden Suite.
  2. Hook downstream consumers up to Golden Suite's exports (CSV + JSON sidecar) as a secondary feed. Don't cut the Reltio feed yet — run both in parallel for 1 week to catch drift.

Week 4 — Cutover

  1. Audit the parallel-run diff. Anything where Reltio and Golden Suite materially disagree → investigate before cutting. Some disagreements are real bugs in your old rules; some will need new survivorship rules in Golden Suite.
  2. Flip downstream consumers to read Golden Suite. Keep Reltio running read-only for another month as a fallback.
  3. Cancel Reltio. Reclaim the $300k+/yr.

Common pitfalls

  • Don't translate Reltio's match rules 1:1. Reltio's proprietary scorer blend isn't easy to mirror; trying produces a worse output than just running goldenmatch's defaults and tuning from there.
  • Don't skip the parallel-run week. Even with confident rules, downstream consumers find drift you didn't.
  • Don't migrate the legacy mainframe last. It's the one with the most variance from clean data; expose its problems early in the parallel-run period.

When NOT to migrate

  • You're already operationalized on Reltio's stewardship workflow and changing tools would create more org friction than the TCO savings recover
  • Your matching scope is genuinely Fortune 500 — billions of records, hundreds of streams, ten-year horizon
  • You're under an enterprise contract you can't get out of without paying through anyway

For everyone else: most teams who looked at Reltio's quote and went home to build a homegrown system are exactly the audience Golden Suite is built for. Migration is a 2–4 week project, not a multi-quarter one.

Questions? Email ben@bensevern.dev or schedule a call from /enterprise.

← Auditing a RepositoryMigrating from Tamr to Golden Suite →