An unsupported component can justify replacing part of a critical application. Broader modernization earns its budget when the system’s design or support constraints prevent the business from closing material exposure and keeping it closed.
A fictional wholesale distributor has an order system that correctly applies years of customer-specific pricing and credit rules. Its runtime is out of support. The team can restrict who reaches the order desk, but the runtime is also shared by warehouse integrations and a partner ordering interface. The CTO has enough budget to fund a first phase. The choice is whether to buy a supported component, move a business capability to a new system, or start a full replacement. The invoices still have to come out right on Monday.
Look for the constraint that keeps risk open
The U.S. Government Accountability Office’s July 2025 review identified 11 federal legacy systems most in need of modernization, drawn from 69 systems identified by agencies. The GAO evaluated factors including vendor support, security risks, operating costs, and age. Agencies reported unsupported hardware, software, or operating systems in four of the 11. Seven reportedly had known cybersecurity vulnerabilities that could not be remediated without modernization. One agency said it isolated components that could no longer receive software updates. The review focused on the systems with the greatest identified need. It shows what a support constraint can demand of an organization that still relies on the affected system. GAO report, methodology and findings on the 11 systems
To make the same distinction in your own application, walk a critical transaction from the first request through its downstream effects. Establish which components control identity, permissions, and data. Check their support status and which builds are actually deployed. Ask who can change and release them, whether the change can be tested against representative workflows, and whether the team can restore service if it goes wrong. For an isolated flaw in supported code, a targeted patch may be the right investment. If incidents or findings keep coming from one replaceable component, price that replacement first.
A broader program starts to make sense when the affected boundary is shared and difficult to change. Several workflows may rely on the same unsupported runtime. Permission decisions may be scattered across jobs and interfaces, with no dependable place to enforce an essential rule. A team might be able to produce a fix but lack a repeatable way to deploy it safely. Each is a different constraint, so each calls for a different first investment. In the distributor’s case, a supported runtime upgrade could be sufficient if it restores a support path and the critical integrations can move with it. Prove that before committing to years of replacement work. Ask the team to estimate the cost of keeping the current system defensible over the same period. Include the maintenance of temporary controls, the impact of an emergency outage, and the staff needed to support the integrations. Those expenses belong beside the migration estimate when leadership decides what it can afford.
Choose the first change that removes an exposure
If the runtime upgrade cannot carry the old integrations, a phased migration may give the distributor a smaller decision. Martin Fowler describes incremental legacy replacement through seams in the existing system, where a portion of behavior can be redirected to a replacement while the rest continues to operate. Matthew Foster of Thoughtworks describes using this approach to ship a modernized subset of a complex mobile application alongside the old one. The team reused components it had already built for other apps. It reported that putting the first one to use in the legacy app took about a month, including the work to make the old app host the new component. That month is a case result tied to reuse. A team creating the replacement from scratch has more work to fund. Martin Fowler on incremental replacement · Thoughtworks’ mobile modernization case study
For the distributor, one candidate is partner order status. It has a smaller footprint than order entry and is a good first phase if moving it lets the team retire a public legacy endpoint. The new service can read a limited set of order data while the old system remains responsible for pricing and writes. Give the pilot a representative partner cohort and a way to compare the answers the two systems produce. Confirm that partners see only their own orders, that the information stays current enough to use, and that support can diagnose discrepancies. Keep order entry on the existing system until the team has evidence to make a separate decision about writes.
A phased move comes with its own costs. If two systems accept writes, routing traffic back may fail to restore the state customers expect. Before moving a write workflow, decide which system owns each record, how to reconcile discrepancies, and what rollback means for changes already accepted. Sometimes there is no safe seam. A coupled runtime and data model may require a coordinated platform upgrade or cutover. That case needs a representative rehearsal, a plan for recovering accepted transactions, and a defensible estimate of the business interruption.
Fund the milestone and its exit condition
GAO found that only three of the 11 selected systems had fully documented modernization plans. It looked for milestones, a description of the work, and plans for what would happen to the legacy system. That last item belongs in a commercial budget as well. If the old interface stays reachable after a successful launch, some of the original exposure remains. Put the shutdown work and its owner in the same funding request. Be explicit about which records, audit history, and operational knowledge must remain available after the interface is retired. GAO report, findings on modernization plans
For the distributor, the first milestone could be to reconcile results for the pilot cohort, move the rest of the status traffic, and verify that the former public status endpoint has been retired. Meanwhile, access restrictions and monitoring can reduce risk on the remaining workflows, with an owner and a date to revisit those controls. The funding decision should identify the exposure the first release will remove and the business behavior it must preserve. If the proposed first phase can show both, the CTO has a basis for funding it and a concrete result to review before spending on the next one.
Sources and further reading
- U.S. Government Accountability Office, Agencies Need to Plan for Modernizing Critical Decades-Old Legacy Systems, GAO-25-107795, July 2025.
- Martin Fowler, Strangler Fig, August 22, 2024.
- Matthew Foster, Thoughtworks, Using the Strangler Fig with Mobile Apps, November 5, 2024.
Want a second look at your own applications?
We find and verify what’s exposed, fix what matters, and modernize what can’t be secured piece by piece.

