The data-migration E&O minefield — accuracy or liability.

Of all the integration tasks, data migration is the one where a small, invisible error compounds into the largest possible loss. Move a coverage limit wrong, drop an endorsement exclusion in the field mapping, and nothing breaks — the system shows no error, the renewal goes out, everyone moves on. Then a year later a client has a claim, the coverage isn't what the record said, and the agency is holding an errors-and-omissions liability that can dwarf what it paid for the book. The discipline that prevents this is unglamorous and absolute: clean the data first, migrate it carefully, and audit it to 100%.

§ 01 · The four-stage disciplineMap, clean, audit, retain.

Data migration runs in four stages. Data mapping: a field-by-field translation between the source and destination schemas, where every element must have a confirmed destination or a documented handling decision — an unmapped field is a silently dropped one. Data cleaning: resolve inconsistencies, remove duplicates, purge outdated records, and standardize naming — a stage that often takes longer than the migration itself, and the one buyers most underestimate. Quality-control audits: a pre-migration audit to validate source integrity and establish a baseline, plus a post-migration audit comparing a statistically significant sample field by field. And the read-only legacy license: maintain access to the seller's original system for the full 12-month window. The order matters — clean before you map, audit before and after — because each stage catches a different class of error before it becomes permanent. The legacy-license window connects to the liability shield in the E&O liability shield.

§ 02 · Clean at the sourceThe 3–5× rule.

The economic rule that governs the whole effort is the 3–5× cleanup ratio: fixing inaccurate records after they've been imported into the buyer's clean system costs three to five times more than cleaning the source data before migration. The reason is compounding — a dirty record in the source is one problem; the same record imported, then referenced in renewals, endorsements, and client communications, has propagated into many. The data-cleaning stage is therefore not optional housekeeping but the highest-leverage point in the migration: every hour spent cleaning at the source saves three to five hours of post-import cleanup, and avoids the records that would otherwise slip through to become liabilities. A buyer eager to "just move the data and fix it later" has the economics exactly backwards. And the read-only legacy license that holds the original records for 12 months is the safety net — it preserves access to the unmigrated source if a discrepancy surfaces, aligning with the E&O tail-coverage period for the same reason. The legacy-system access discipline is detailed in platform consolidation.

§ 03 · The 100% auditTop-50 by hand.

The accuracy standard is the part buyers find hardest to accept: 100%, not "close enough." Any field-by-field discrepancy in the post-migration audit triggers a hold on that record until the issue is resolved — there's no acceptable error rate, because in insurance a single wrong coverage limit is a claim waiting to happen. Beyond automated sampling, the top 50 clients by premium get line-by-line manual verification: these high-value accounts justify human review because the E&O exposure on a single large commercial account exceeds the cost of the extra audit hours many times over. The logic is risk-weighted — automated sampling catches systemic errors across the book, and manual review of the top 50 catches the concentrated risk where a single error is catastrophic. A buyer who audits a 5% random sample and calls it done has verified the wrong thing: the random sample misses the specific large account whose misfielded data is the actual liability. Sample broadly, verify the top 50 by hand, hold any discrepancy until resolved — that's the 100% standard in practice. The renewal and policy-update workflows this clean data feeds are the subject of workflow harmonization.

§ 04 · The silent failure modesWhat surfaces as a claim.

The reason 100% is the standard is the nature of the failures, which are silent — they produce no error message and surface only when a client files a claim. Four are the most dangerous. Orphaned data: records that lose their links in the migration, invisible until someone goes looking. Truncated coverage limits: a limit that imported wrong, so the policy shows less coverage than the client bought. Lost endorsement-exclusion data: an exclusion dropped in the field mapping, so the record misrepresents what's actually covered. And incorrect client addresses: renewal notices that misroute, so a client lapses without ever knowing their renewal came due. Each is silent at migration and catastrophic at claim time, and the framing that should govern the entire effort is that a single botched data transfer can generate a liability that exceeds the entire purchase price of the acquired book. That's why this is a minefield, not a chore: the cost of getting it wrong isn't a cleanup project, it's an uninsured loss bigger than the deal. Clean at the source, run the four stages, audit to 100% with the top 50 by hand, and hold the legacy license for the full 12 months — and the minefield is crossed safely. The E&O coverage that backstops a pre-migration error is in the E&O liability shield.

◆