Six Hundred Institutions Have Disappeared Since 2021. Every One Was an Integration Project.

Consolidation is usually discussed as a strategic story. Underneath it is an engineering workload that nobody puts on a slide.

The number of FDIC-insured institutions fell from 4,839 at the end of 2021 to 4,238 by the second quarter of 2026. In that quarter alone, 36 institutions were absorbed by mergers, four opened, four were sold to non-insured buyers, and one failed.

Six hundred and one charters have gone in under five years.

What a merger actually asks of the systems

Two banks combining is two of everything that has to become one of everything. Two cores. Two general ledgers. Two customer masters, each with its own idea of what a customer is. Two chart-of-account structures, two product code sets, two document repositories, and two sets of integrations to everything else.

The commercial case is agreed in months. The systems work runs for a year or more after close, and most of it is not glamorous. It is field-level mapping, historical balance reconciliation, decisions about which system holds the record for which entity, and the long tail of reports that have to keep producing the same numbers through the transition.

Why the customer master is where it gets expensive

Consider one narrow problem. Both institutions have a customer who banks with each. In one system that customer is a single record with three accounts. In the other, the same person appears twice, because a joint account was opened under a different identifier eleven years ago.

Nobody wrote that down. It surfaces during migration as a duplicate, and somebody has to decide what the merged record should look like, then apply that decision consistently across every case with the same shape.

Multiply by the entity types a bank carries. Individuals, sole proprietors, operating companies, holding companies, trusts, and the relationships between them. Each combination has its own reconciliation rules, and each rule has to be encoded rather than remembered.

The acquirer inherits the acquired bank’s shortcuts

Every institution runs some processes on informal infrastructure. A spreadsheet that reconciles two systems monthly. A report one person builds because the system cannot produce it. A workaround introduced during a conversion in 2019 that was meant to be temporary.

These are survivable at the institution that created them, because the people who understand them are still there and the volume is known. They do not survive acquisition well. The spreadsheet arrives without its author, the report arrives without its logic, and the workaround arrives without the context that explains why it exists.

Diligence rarely surfaces them, because they are not systems and they do not appear on an application inventory. They appear afterward, when a number stops reconciling and nobody can say what used to make it reconcile.

That is the practical argument for mapping the current state of an operation before a transaction rather than during one. The processes that will be hardest to migrate are the ones held in people rather than in software, and they are findable in advance if somebody looks.

The integration debt does not end at conversion

A conversion is declared complete when the balances tie. The integration work continues for as long as both estates keep running.

In practice, few acquisitions collapse to a single stack. A specialist origination platform is retained because the acquired team is good at that product. A servicing system stays because migrating it would disrupt a book mid-term. What was going to be one environment becomes one core plus several survivors, connected by whatever was fastest to build under deadline.

Those connections are the ones that break later. They were written against a schema that has since changed, by people who have moved on, with no monitoring beyond somebody noticing a report looks wrong.

This is why an institution three years past a merger often has higher operating cost than either predecessor, and cannot say precisely why. The cost is distributed across dozens of small manual bridges, none of which is large enough to be a project.

What to build instead of another bridge

The work that reduces this is systems integration done deliberately rather than under closing pressure: bidirectional synchronization between core, CRM, and the general ledger with one system of record named per entity, role-based access carried through the integration instead of bolted on beside it, and hardened interfaces that survive a vendor release rather than failing silently at the next upgrade.

None of that has to wait for the next transaction. A bank carrying integration debt from a merger two years ago is paying for it now, in the reconciliation queues and the reports that take three days to assemble. Those are the places the operating day backs up, and they are the same places a future conversion will hurt most.

Six hundred and one charters have disappeared. Roughly the same number of integration programs happened, and most of them are still running whether or not anyone is funding them as such.

Start with the process that costs you most. A conversation first, and a structured assessment when it earns one. Book a discovery call at https://cxocorporation.com/contact.