Systems Integration & Interoperability

Systems that need to exchange data were built separately, at different times, by different organizations, against different requirements. Nothing made them compatible and nothing since has required it.

GAO looked at this inside the Defense Department in 2025 and listed the reasons plainly: military department-specific operational and acquisition requirements, stove-piped system development, and proprietary contractor designs. Services prioritize their own mission objectives over joint ones. Officials also raised vendor lock, where an organization cannot move to another supplier without substantial cost and rework.

DOD requested over $1.4 billion for joint command and control integration in its fiscal 2025 budget, while the Air Force, Army, and Navy each continued developing their own version of the capability through separate programs.

The eleven unified combatant commands and the eighteen organizations of the intelligence community told GAO they manage data differently from one another, which is a barrier that exists before any technical work begins.

GAO found overly restrictive classification to be a significant hindrance to sharing command and control data, with designators such as NOFORN blocking partner access. Officials it spoke with were not aware of any entity working on a solution, and command and control leadership said the problem sat outside their purview.

The civilian side has the same governance gap, and there it has been measured. No federal agency is designated to set or enforce interoperability requirements. Of nine federal data sources GAO examined in 2026, all nine carried quality problems: missing values, invalid entries, inconsistent formatting, conflicting relationships between records, and duplicates.

Moving data between two systems is the straightforward part of the work. Deciding which one is authoritative when they disagree is the difficult part, and it usually surfaces after the interfaces are built and the schedule is already committed.

Treasury operates a service that checks payments against federal data sources before money leaves, and it accounted for almost $5 billion in improper payments prevented or recovered during fiscal 2025 while drawing on only a fraction of the sources available.

In operational domains the return is harder to express as a dollar figure, and takes the form of a common picture that people can act on without reconciling three systems by hand beforehand.

At OCH we start with data ownership: which office is accountable for a record, what they are permitted to share, and what agreement has to be in place before anything moves. Those questions set the schedule more often than the technical work does.

We design interfaces contract-first, version them from the first release, and publish them to a catalog the agency owns, because when an interface is hard to find teams build their own path around it and that path becomes the one everyone depends on.

We build reconciliation and exception handling at the start, on the assumption that the two systems will disagree, and we agree in advance where a disagreement gets resolved and who resolves it.

We stabilize interfaces before decomposing anything behind them, since breaking a system apart first tends to reproduce the original coupling across more moving parts.