Last updated: November 17, 2026
Core banking modernization replaces or rebuilds the system of record for accounts, deposits, and loans. When the vendor runs the conversion, your team still owns the integrations, data mapping exceptions, testing, and downstream systems. That work contracts cleanly: a senior QA engineer bills $94 to $114 an hour, median $106 (KORE1 Contract Rate Index 2026-Q4).
The vendor called the first mock conversion a success, and by its own checklist it was. Every account loaded. Balances tied to the penny. A green status page landed in the credit union’s inbox at 6 a.m. on a Sunday, and the conversion team went home. Nobody worried.
By Monday morning the debit card processor was declining a few hundred transactions an hour that it had no reason to decline. The card integration still sent account numbers in the old core’s format, and nobody had been assigned to change it. The vendor’s statement of work covered the conversion, but the card interface was not part of the conversion and belonged to the credit union, which had put no one on it. (The institution is a composite of two I know, and I have changed the details.)
The vendor did nothing wrong. A core vendor sells a conversion and delivers a conversion. What it does not deliver is everything that talks to the core, and that list is longer on signing day than anyone’s slide admits. The rest of this post draws the line between the two sides, prices the seats on yours, and says which of them should be contractors.
I write in partnership with KORE1, a staffing firm that places contract engineers into regulated shops through its banking IT staffing practice, so it benefits when your team needs extra seats. Weigh my view with that in mind, and check my work split against your vendor’s statement of work. Yours is the only one.

What Is Core Banking Modernization, in Plain Terms?
Core banking modernization means replacing or rebuilding the core, the system of record that posts deposits, loans, and payments and feeds the general ledger. Institutions do it three ways, a full replacement run as a conversion, one component at a time, or wrapping the old core with APIs.
The Federal Reserve Bank of Kansas City sorts the options the same way in its 2024 briefing on core banking systems and options for modernization. It calls a full replacement “the most risky and expensive option,” says component-based replacement is “less risky,” and describes the third path as plugging a next-generation core into the legacy one through API connections. A full replacement is the one with a vendor-run conversion weekend, so it is the one I mean for the rest of this post.
Which kind of core do you have today? I sort them by who runs the processing. There is licensed software your own staff operate, there is a hosted arrangement (some people say service bureau) where a core provider runs it for you, and there are cloud-native platforms built to be run that way from the start. The Fed’s briefing adds a second cut by age. Legacy cores are monolithic, with every component intertwined, and next-generation cores are modular, with the components decoupled. It also notes that most smaller institutions outsource to a core provider, which is why so many credit union teams meet a conversion as something done to them by a vendor, on a vendor’s calendar.
No vendor names, on purpose. The work split below holds whichever one you signed with.
What Does the Core Vendor’s Conversion Plan Cover, and What Stays on Your Side?
The vendor’s statement of work covers the conversion programs that move your data, the extracts from the old core, the mock conversions, and its own testing of all of that. It is a real list. Good vendors do it well. Then the plan stops at the edge of the core.
Your side starts there. Every integration into the new core: digital banking, cards, payments, lending, the CRM, the data warehouse, the feed to the general ledger. Each data mapping exception, and the decision behind it. End-to-end and regression testing. Downstream and regulatory reports. The cutover runbook. None of it is exotic. All of it carries a date that someone else set.
Where a vendor’s plan goes quiet, I have assumed the work is yours, because that is how it tends to land. If your statement of work has actually taken a row, move that row across.
| Workstream | The vendor owns | Your team owns | Contract seat | Index bill range per hour |
|---|---|---|---|---|
| Integrations into the new core | The new core’s interfaces and the documentation for them. | Re-pointing or rebuilding every connection: digital banking, cards, payments, lending, CRM, data warehouse, and the general ledger feed. | Integration engineer (senior software engineer) | $118 to $142, median $131 |
| Testing | Its conversion programs and its own testing of each mock conversion. | End-to-end and regression testing across every connected system, with evidence you keep. | Test automation engineer (senior QA tester) | $94 to $114, median $106 |
| Data mapping | The extracts from the old core and the mapping tooling. | Every mapping exception and the decision behind it, plus the downstream and regulatory reports that read the mapped data. | Data mapping analyst (senior business systems analyst) | $95 to $115, median $106 |
| Cutover | The mock conversions and its side of the conversion weekend. | The cutover runbook, the go or no-go evidence, and the order in which downstream systems come back. | Cutover lead (senior project manager) | $93 to $114, median $106 nationally. KORE1-placed IT project managers, $105 to $124, median $116 (n=21, primarily Southern California) |
| The core after go-live | Support under its contract. | A product owner for the core, and the engineers who own each integration. | Permanent employees, not a contract seat | Not priced |
Bill ranges run from the 25th to the 75th percentile of the KORE1 Contract Rate Index 2026-Q4, with the 24% employer burden already inside. The columns for what each side owns are my reading of the common arrangement, and your contract decides.
Staff the institution-side seats.
Send the seats, the seniority, the number of months in the vendor’s plan, and the conversion date through the bill rate quote, and you get hourly bill rates by role to set beside the plan.
Which Engineering Seats Sit Outside the Vendor’s Statement of Work?
Four. They are the same four whichever vendor you signed with, and I would name them before the vendor’s kickoff.
Start with the integration engineers, plural, because one is never enough. Every system that talks to the core has its own owner, its own release calendar, and its own test environment, and each has to be re-pointed to the new core or rebuilt around its interfaces. On day one they need access to the test environments of both cores and the vendor’s interface documentation. Without both cores up side by side, nobody can put a response from the old one next to the same response from the new one.
The test automation engineer is the seat I would fill first. Mock conversions repeat. Each time, somebody has to run the same end-to-end and regression suite against a fresh load. Once is a chore. Run by hand on the third mock, by a tired person on a Sunday, it is how a defect gets through. Day one needs the vendor’s test data sets and a place to run the suite that is not production.
The data mapping analyst owns decisions, and I chose that word carefully. The vendor’s tooling maps field A to field B, but it cannot decide what a closed loan with a leftover balance should become, or which of two customer records is the real one. Someone has to make those calls and write down why, in a form you could show an examiner. That person reads the vendor’s mapping documents on day one and has authority to say no to a mapping. The reconciliation side, proving the balances match after the move, has its own treatment in the data migration plan for engineering leaders, so I will not repeat it.
Last, the cutover lead. One person, accountable for the runbook, who is allowed to say stop. They need the vendor’s conversion calendar, the list of every downstream system, and the names of whoever signs off on go or no-go.
That is a short list. It is also the one that gets left off the budget.

Which of Those Seats Contract Cleanly, and Which Should Be Permanent?
Three of four contract cleanly. Test automation, integration build, and data mapping all have a start, an end, and a definition of done that the vendor’s calendar hands you. Bounded work with a date on it is what contractors do best, and it is contract staffing at its simplest.
Two things should stay permanent. One is the product owner for the core, the person who decides what the new platform is for after the vendor leaves. The other is the group of engineers who own the integrations once they are live. Those people hold knowledge that has to last for years, and when the vendor’s support line asks a question at 2 a.m. in month eleven, the answer should not have walked out with a contractor. This is build vs buy for a crew. Buy the sprint. Build the owners.
People skip the handover. Ask the integration contractors for runbooks written as the work happens, and for a recorded walkthrough of each integration from the engineer who built it, before the last month and not during it.
KORE1 IT-role assignments run a median 136 days (n=410; 79% of KORE1 engagements are in California). A contractor you want to keep converts to your payroll at no fee after 1,040 hours, and at 40 hours a week that is 26 weeks. So the decision to keep someone is a month-three conversation. In go-live week it is a negotiation.
What Breaks If the Conversion Date Slips?
Devin Hornick, a partner at KORE1, described where clients are putting their money in an email. “Focus… where companies are still investing and where they simply can’t afford to have projects fail or go unfinished.” A core conversion is that sentence made into a calendar.
The date is booked months ahead with the vendor. A failed mock conversion can push the conversion weekend. When the weekend moves, the old core’s contract keeps running, and every project waiting on the new core waits with it. The mobile features, the payments upgrade, the lending workflow meant to ride the new platform. They all wait.
The stakes are not small. The Kansas City Fed’s briefing says a full core conversion “can take several years and cost millions (if not hundreds of millions) of dollars,” and a month of slip inside a program that size is not a rounding error.
Most slips I picture start the way my opening story did, with a connection nobody owned. The fix was small. Finding it on a Monday after a clean mock was the expensive part, and finding it after the final mock would have been worse. Name every integration’s owner first.
How Do You Oversee the Vendor’s Work Without Owning It?
Responsibility stays put. The 2023 interagency guidance on third-party relationships from the Federal Reserve, the FDIC, and the OCC says a bank’s use of third parties “does not diminish its responsibility to meet these requirements to the same extent as if its activities were performed by the banking organization in-house.” The vendor runs the conversion. The bank still answers.
That is supervisory guidance and not a statute, which the agencies say themselves, and it is written for institutions those three agencies supervise. If you are a credit union, your examiners have their own expectations on vendor oversight, so ask them what they want to see before mock one.
What does oversight look like? Evidence you hold yourself.
- Test results from your own suites, with the date, the data set, and the build under test written on each run.
- A sign-off from a named person for each mapping exception.
- One issue log that you control, with the vendor’s defects entered in it and not left in the vendor’s tracker.
If an examiner asks how you know the conversion was right, “the vendor told us” is not an answer. A reconciliation, a stack of sign-offs, and an open-issue log is. I wrote about the other side of this, what happens when a finding lands and you need hands fast, in fixing an examiner finding on a deadline. The access, review and evidence habits that make the contract seats safe in a regulated environment are in running contract engineers in a regulated shop.

What Does the Institution-Side Team Cost?
Show me the data, so here are five cells from the KORE1 Contract Rate Index and nothing else. I have not seen your vendor’s plan, so I am not pricing the conversion itself, and you will not find a program cost in this post.
| Seat | Geography and source | Bill rate per hour | Median | 3 months | 6 months |
|---|---|---|---|---|---|
| Senior software engineer (integration) | United States national, modeled | $118 to $142 | $131 | $68,300 | $136,600 |
| Senior QA tester and test engineer | United States national, modeled | $94 to $114 | $106 | $54,900 | $109,900 |
| Senior business systems analyst | United States national, modeled | $95 to $115 | $106 | $55,300 | $110,600 |
| Senior IT project manager | United States national, modeled | $93 to $114 | $106 | $54,900 | $109,800 |
| IT project manager, all levels | KORE1 observed placements, primarily Southern California, n=21 | $105 to $124 | $116 | $60,400 | $120,900 |
Ranges run from the 25th to the 75th percentile of the KORE1 Contract Rate Index 2026-Q4, effective October 16, 2026. These are bill rates, billed hourly, with the 24% employer burden already inside every one, and the 3 and 6 month columns assume a 40-hour week.
The national cells are modeled. They start from the Bureau of Labor Statistics OEWS May 2025 wages for software developers (SOC 15-1252), software QA analysts and testers (SOC 15-1253), computer systems analysts (SOC 15-1211), and project management specialists (SOC 13-1082), add the burden, and apply the bill-to-cost range KORE1 observes. The last row is observed, from 21 placements, mostly in Southern California. Twenty-one is a modest sample, so treat it as a data point. The contract rate methodology page has the full method, and the seat-level detail is on the QA tester and test engineer contract bill rates page and the business systems analyst bill rates page.
One of each national seat for six months comes to $136,600 plus $109,900 plus $110,600 plus $109,800, which is $466,900 at index medians. Multiply by the months in your vendor’s plan, and change the seat count to match what your own team cannot absorb. It will not match mine.
Devin Hornick again, on billing. “Hourly, which we prefer. It’s really more of a staffing model to do the hourly billing and you manage the person.” You direct the work. You pay for hours worked.
From KORE1: what we place and how it works
KORE1 staffs the institution-side seats as W2 contract engineers, billed to you hourly and directed by your team. Contractors are KORE1 employees, paid every Friday, and they can convert to your payroll at no fee after 1,040 hours. Contract placements carry no replacement guarantee. We have recruited technology talent since 2005, and 92 percent of the people we place are still in the job twelve months later.
Start with our banking and credit union IT staffing practice if your seats sit inside a bank or credit union.
Questions Engineering Leaders Ask Before Mock One
Can you explain core banking in a simple way?
The core is the bank’s system of record, the program that posts every deposit, loan payment and transfer to the right account and feeds the general ledger. The app, cards, and statements all read from it or write to it, which is why a change to the core reaches every other system.
What are the different types of core banking systems?
By who runs the processing there are three: licensed software your own staff operate, a hosted arrangement where a core provider runs it for you, and cloud-native platforms. The Kansas City Fed adds a second cut by age, legacy monolithic cores against next-generation modular ones. Either cut, same engineering split.
Who is responsible for testing in a core conversion?
Both sides, for different tests. The vendor tests its own conversion programs and runs the mock conversions, and your team tests everything connected to the core and the results the business depends on. Put the split in writing before mock one, and keep your own evidence. Examiners ask for yours.
Do we need contract staff if the vendor runs the conversion?
Wrong question, slightly. The better one is whether your team has four seats of spare capacity for the length of the vendor’s plan, and most teams do not. If your own people can absorb integration, testing, mapping, and cutover without dropping the roadmap, hire nobody.
Write the Split Down Before Mock One
The vendor will run the conversion. It was never going to run your bank. The weekend ends, the vendor’s team leaves, and the integrations, the mappings, and the test evidence stay behind with your name on them.
Take the table above and lay it beside the vendor’s statement of work and calendar, because every row where nobody has a name is a risk to the conversion date. Fill those in before the first mock and the Sunday morning email gets a lot less interesting. Skip it and you end up like the credit union in my opening story, finding out on a Monday.
If you want contract engineers to fill the seats you cannot, talk to the KORE1 team and bring the plan. Or start with hourly bill rates by role and price the seats before anyone signs a conversion date.
As of November 2026. Rates are from the KORE1 Contract Rate Index 2026-Q4, effective October 16, 2026, and the engagement figures reflect KORE1 W2 contract placements, 79% of them in California. This post is reviewed every six months. It is general information and not legal or regulatory advice.

