Do not merge systems before you merge definitions. A merger migration needs a sandbox, an explicit dedupe and survivorship rule set, and a decision on whose stage model and segment definitions win. Consolidating records under two conflicting data models produces a system neither team trusts.
Most acquisitions fail on integration rather than deal thesis, and go-to-market systems are where that failure becomes visible. Sequence it: agree definitions, map both models to the target, dedupe with a written survivorship rule, migrate in tranches, run both in parallel, then decommission.
Every claim above comes from work LeanScale published. These are the sources.