When a credit union replaces its core, the conversion arrives with a vendor, a methodology, and a war room. But the decades of loan files, signature cards, and nightly report archives living outside the core do not convert themselves. That content migration has no default owner, and it is the half of the modernization nobody scopes.
Here is what falls in that seam, and how to close it.
The seamThree vendors. No one owns your documents.
Every document, through one pipeline, exactly once.
Redactions preserved. Relationships intact. Reports split to the member level. And every object written to the log.
Outcomes, on the record.
Table of Contents
What does a core conversion leave out?
Everything in your enterprise content platform, which is most of your institution’s memory.
The core vendor’s statement of work does not include the millions of documents (sometimes billions of pages) in your legacy ECM: signature cards, loan documents with their collateral title work and lien perfections, membership documents, beneficiary designations, wire records, HR files, AP invoices, and share draft images.
Alongside all of that sits an even larger body of content: the report archive. Member statements, notices, general ledger reports, trial balances, and end-of-day core reports, captured every night for twenty or thirty years. All of it subject to audit.
Three parties surround this content, and none of them owns moving it. The new core vendor says, correctly, that it is out of scope. The destination ECM vendor says, correctly, “send us clean, indexed data.” And the platform you are leaving was not necessarily designed to make extraction simple.
The migration falls in the seams unless someone is made explicitly accountable for it. Because it looks like “copying files” from a distance, it is routinely assigned to an internal team already consumed by the core project, or to a general integrator who has migrated file shares but never a credit union document archive. That underestimate is where every failure that follows begins.
You’ll do this migration once. Read this before you do.
“The First Time Is Your Worst Time” is a leadership guide to the half of your core conversion nobody scopes: the content migration. Inside: what’s actually hiding in decades of archives, the six lessons first-time teams learn too late, the seven questions to ask before the workstream begins, and what “done” looks like to an examiner.
Register to get the white paper.
What is actually in a credit union’s document archive?
Not just volume. Accumulated history, and it must be understood before it can be moved. Five things hide in almost every credit union archive:
Index fields that changed meaning over time. A field labeled “officer” may have held three different things across three decades:
- The loan officer
- Then a branch code
- Then a queue identifier
Migrate it blindly and the new system’s search is poisoned from day one.
Document types that multiplied. Archives routinely contain hundreds of document type definitions where the business only ever needed a few dozen, near-duplicates created by different departments and administrators over the years. Migrate as-is and the new ECM inherits thirty years of drift.
Relationships the export won’t hand you. Members relate to accounts, accounts to loans, loans to collateral files, and signature cards to everything. Those relationships live partly in index values, partly in folder conventions, partly in tribal knowledge. Preserving them is the difference between a member-centric archive and a pile of correctly named images.
Annotations, redactions, and versions stored separately. In many legacy platforms, stamps, sticky notes, redactions, and prior versions are sidecar data, not burned into the image.
A naive export migrates the image and silently discards the redaction. Now the redacted content is exposed in the new system, and no one knows.
The report archive, which is the institution’s memory of the old core. The new core will not reproduce the old core’s reports. The day you convert, those nightly reports become the only record of how the institution operated on the old platform.
They are also the hardest content to migrate well: spooled and COLD-format report streams that must be parsed, split into member-level and account-level documents, and indexed from the report text itself.
Treated as an afterthought, they arrive as unsearchable multi-thousand-page blobs. Treated correctly, a member statement from 2011 is retrievable in seconds, forever.
None of this is exotic knowledge. All of it is invisible until you have hit it.
Comparison Table: First Attempt vs. Hundredth
| A first-time migration team | An experienced migration partner | |
|---|---|---|
| The legacy export | Trusts it | Audits the full source population first |
| Quality assurance | Manual spot checks | Automated, document-level reconciliation of every object |
| Annotations and redactions | Often silently dropped | Detected, preserved, correctly applied |
| The report archive | Migrated as giant unsearchable blobs, or skipped | Parsed, split to member level, indexed from report text |
| Timing | Runs alongside the core project and collides with it | Phased: bulk history, delta sync, final sweep |
| Taxonomy | Thirty years of drift, moved as-is | Rationalized in flight; expired records defensibly dispositioned |
| Decommissioning | Legacy system stays on "just in case" | Signed off against an examiner-ready reconciliation package |
| The lessons | Learned on your archive, mid-project | Already learned, on someone else's, hundreds of migrations ago |
What do first-time migration teams learn too late?
Six lessons, each obvious in hindsight, each routinely learned mid-project on a one-way door:
1. The export lies. Legacy extracts arrive with null index values, duplicate identifiers, orphaned pointers, and documents the database references but the disk forgot (and vice versa). Veterans audit the full source population before anything moves.
2. You cannot QA it by hand. Spot-checking ten thousand documents tells you nothing about the forty million you didn’t look at.
Only automated, document-level reconciliation closes the gap: every source object accounted for in the destination or dispositioned with a documented reason.
3. Sequence against the core conversion, not alongside it. Cut over too early and you strand months of new documents in a dying system; too late and you’re running the riskiest data operation of the decade during go-live week.
Veterans phase it: bulk history first, delta synchronization through conversion, final sweep after.
4. Migration is the once-in-a-generation chance to fix the taxonomy and shrink the archive. Done well, a migration often moves fewer documents: expired records defensibly dispositioned, duplicate document types collapsed, index values normalized. You pay to modernize your records, not your junk.
5. Decommissioning is the finish line, and most first attempts never cross it. Unable to prove completeness, the institution keeps the legacy system alive “read-only, just in case,” paying maintenance to the vendor it just left, year after year. The antidote is not courage; it is evidence.
6. This is a one-way door. Once the legacy system is decommissioned, whatever the first attempt missed is gone permanently. Every other lesson matters because of this one.
An experienced partner does not avoid these problems. They arrive expecting them, with detection and methodology built from hitting every one before, on someone else’s project.
What does a failed content migration cost a credit union?
A core conversion fails loudly, on day one. A content migration fails silently, over years, and the bill arrives in four currencies:
Members. Industry estimates place credit union member acquisition cost in the hundreds of dollars (recent studies from CU 2.0 and Debbie’s 2026 financial services acquisition report put it roughly between $400 and $700), with the lifetime value of a member relationship estimated as high as $20,000.
“We can’t find your document since the conversion” is exactly the kind of moment that can end a thirty-year relationship.
Litigation. Duane Morris’s Class Action Review reports the top ten data breach class action settlements totaled $593.2 million in 2024, up from $515.75 million the year before.
Documents lost at scale, retention broken uniformly, or chain of custody compromised across the whole membership is precisely the institution-wide failure pattern class litigation is built for.
A single missing lien-perfection document, discovered at charge-off, converts directly into dollars.
Examiners. NCUA guidance on electronic records (12 CFR § 749.5) is explicit: retained records must accurately reflect the information in the record, remain accessible, and be reproducible on demand.
The reconciliation evidence that makes a migration defensible cannot be reconstructed after the fact. It is either designed in, or it doesn’t exist.
And the migration partner itself falls under your credit union vendor management program: examiners will expect the same due diligence, documentation, and oversight you’d apply to any third party handling member data.
The project itself. Every quarter of delay is another maintenance check to the IDP vendor you already left. And the people you would assign to a do-it-yourself migration are the same people delivering the core conversion, the largest project of their careers.
What does “done” actually look like for an ECM migration?
Success is provable, specific, and finished. Five tests:
- One hundred percent accounted for. Every document migrated and verified, or deliberately dispositioned under retention policy with a documented reason, with a processing log that proves it object by object. There is no third category.
- Day-one parity or better. A teller finds the 1987 signature card faster in the new system than in the old one.
- The legacy system actually turned off, on schedule. Leadership signs the decommission backed by an examiner-ready reconciliation package, and the maintenance payments stop.
- Less content, better organized. Expired records defensibly disposed of, taxonomy normalized, member-account-loan relationships intact: a clean, AI-ready foundation for enterprise search, Copilot, and whatever comes next.
- Nobody talks about it. Six months after go-live, no member, no employee, and no examiner has mentioned the content migration at all.
How does BIS run credit union content migrations?
Experience alone is a war story. Experience industrialized, encoded into a platform, a methodology, and a team, is a repeatable outcome. BIS has worked in credit union documents for four decades, through every generation of imaging and ECM technology, and our position is simple: your institution will do this once; we have done it hundreds of times. The first time is always the worst time. Make it ours instead, because we already had it, hundreds of migrations ago.
The engine is Grooper, the intelligent document processing platform we built and operate ourselves. It audits and extracts legacy exports, validates every index value, detects the era-shifts described above, preserves document relationships, and generates import-ready data for virtually any destination: OpenText, Box, SharePoint, OnBase, or the ECM of your choice. It parses decades of COLD and spooled report formats, splitting nightly reports into member-level documents and indexing them from the report text itself.
Every document’s complete journey is written to a processing log: extracted, audited, validated, transformed, loaded, verified, or deliberately dispositioned, with timestamps and reasons. The log is a deliverable, not a byproduct. It is what leadership signs the decommission against, what examiners accept, and what counsel can stand on years later.
Security is designed in: encryption in transit and at rest, least-privilege access, documented chain of custody, and correct handling of the archive’s hidden layers, especially redactions, so nothing the institution once decided to conceal gets silently exposed in the new system.
And because we are the manufacturer of the technology, there is no gap between the software and the service. Our platform, our people, your outcome. Your team is not asked to learn another system or QA millions of documents on nights and weekends in the middle of a core conversion.
The scale claims here are not hypothetical:
- American Airlines Credit Union used Grooper to locate more than 40,000 documents containing PCI data from among roughly one billion stored pages.
- Wings Financial Credit Union eliminated 10,140 hours of manual document work and made data available in downstream systems up to 700 percent faster.
- Del Norte Credit Union has automated 30 document types on Grooper since 2018 and eliminated 97.5 percent of manual data entry; the daily human check takes five to ten minutes.
- And at federal-agency scale, Grooper migrated 39 million pages for the U.S. Nuclear Regulatory Commission in nine months.
One more thing a migration buys you, because every document passes through the pipeline exactly once: reorganization costs almost nothing extra. What arrives on the other side is not a copy of the old archive in a new place. It is a clean, semantically consistent body of knowledge, ready for enterprise search, Copilot, and agentic AI. Institutions that migrate as-is pay to modernize their content twice.
Seven questions to ask before your core conversion
If any answer is unclear, the seam is open:
- Who owns the content migration workstream, end to end?
- What documents and reports live outside the core today?
- Which historical reports must remain searchable after conversion?
- Which records can be defensibly removed before migration?
- What metadata, relationships, annotations, redactions, and versions must be preserved?
- How will every source object be reconciled to the destination, or to a documented disposition?
- What evidence will leadership use to approve decommissioning the legacy system?
Your new core modernizes the transaction engine. The content migration determines whether the institution’s memory arrives in the future intact. You will do this once, through a one-way door. The single variable you control is whether the team walking through it is on their first attempt or their hundredth.
