In September 2026, the Office of Government Information Services, the federal FOIA Ombuds, released its review of the FOIA websites of 21 major federal agencies.
They conducted their assessment in March against ten basic benchmarks: things as simple as whether a requester can find a request form, a contact for the FOIA Public Liaison, or a clear appeal deadline. Seventeen of the twenty-one websites were missing at least one of them. Worse, when OGIS compared this year's results to its 2021 review, the honest headline is that more of those measures slipped than got better: 24 data points declined between 2021 and 2026, against just 15 that improved.
Four agencies, though, got it right both times, meeting every benchmark in both reviews, which tells you this isn’t an impossible standard – just an inconsistently met one.
I read these findings as a diagnosis of what federal FOIA programs are up against.
I spent four years overseeing FOIA compliance, training, and technology for the Department of the Interior’s decentralized FOIA program. During my fifteen years with Interior, I led or served on teams that procured the tech systems those bureau FOIA teams ran on, participated in the Chief FOIA Officers Council’s Technology Committee, and served as a FOIA Public Liaison.
I’ve been on both sides of this problem: the office that has to answer a requester who’s waited long enough to have to ask “how much longer,” and the side of the table trying to find a system that makes that answer easier to give. Neither side is short on will. Both are short on people, though, and often short on the specific, unusual combination of FOIA knowledge, technology, and technical fluency that would help the work happen more smoothly.
OGIS’s ten benchmarks cover a lot of ground. Of the ten benchmarks, two stand out from my years on both sides of this work: the estimated date of completion and the FOIA log. Both depend on information agencies already have but struggle to surface, and that gap is where the right tools make the biggest difference.
The EDC Problem is a Systems Problem
FOIA has required agencies to provide a real Estimated Date of Completion, an EDC, upon request since 2007. OGIS makes clear that this is not the same as a status update; telling someone they’re a couple hundred requests deep in the queue isn’t the same as telling them when they’ll have their records.
In this year’s review, only about 43 percent of agency websites clearly explained how to get both a status and an EDC. Nearly one in five gave no path to either. Tellingly, OGIS itself couldn’t verify compliance for another 38 percent of websites, largely because the information lived behind portals the Ombuds couldn’t access. When even the federal Ombuds can’t see whether an agency is compliant, that’s a pretty glaring architecture gap.
Part of what makes this so hard is something OGIS's website review wasn't designed to capture: most agencies process far more distinct workflows than the three tracks reportable to OIP and the Annual FOIA Report. Using that reporting as a stand-in for how agencies estimate completion oversimplifies how agencies triage their own work. Producing a real EDC means understanding an agency's resources, backlog, and complexity: its real tracks, not just the reportable ones.
None of this stems from a data problem. The data is certainly there, and all the pieces needed for a real EDC exist somewhere in an agency's own systems. Unfortunately, they’re scattered across the intake system that logs the request, the review queue where the work gets done, the tracking spreadsheet someone built years ago to keep it all straight, and the disclosure process that eventually publishes what's already complete. The estimate is so difficult to pin down and share because that information is scattered across systems that were not built to talk to each other.
Running the entire process, from submission through intake, search, review, response, and disclosure, in one platform gives an agency one up-to-date picture to draw that estimate from, instead of someone piecing it together between disconnected systems every time they’re asked. The more holistically a platform reflects how an agency actually works, the more that date holds up. The purpose of an EDC is transparency: telling the public, as accurately as an agency can, how long the wait will be.
FOIA Logs Have the Opposite Problem
The FOIA Advisory Committee lays out thirteen fields a good FOIA log should carry: tracking number, request date, subject matter, status, disposition, exemptions cited, fee category, and fees charged, among others.
However, a third of the agencies OGIS reviewed don’t publish complete FOIA logs, and even where logs exist, I’ve seen the same failure mode play out for years from the other side of the desk: log summaries that are either so boiled down they’re useless to a requester trying to figure out if their question’s already been answered, or so overloaded with case detail that they’re unusable for the same reason. Both directions defeat the point of publishing at all.
A Staffing and Tooling Story
Early in my career, in one of my first FOIA leadership roles, I surprised the senior leaders I worked for by asking for something that initially appeared counterintuitive: consolidation and more accountability. It worked, and the program came out of it smaller and more effective than it went in. That mindset followed me through the rest of my career. In the program assessments I led or took part in after that, no two situations called for the same fix. There’s no one-size-fits-all answer in this work, but every one of them came back to the same two levers: greater accountability and better technology.
There’s no magic staffing number, no simple ratio of FOIA requests to personnel that is easily quantifiable. Every program deals with turnover and lost institutional knowledge. The solution is in bringing together the right people, the right process, real accountability, and technology that fits how the work gets done. Doing so helps every agency hold onto institutional knowledge so it doesn’t disappear when a particular person moves on.
Many federal FOIA offices are currently balancing fewer people, more requests, increased complexity, and, for a lot of good FOIA professionals, a technology layer that feels like one more thing to manage instead of something built to help.
Read past the list of shortcomings, and the same thread runs through the whole OGIS report. The compliance data almost always exists somewhere in an agency’s systems, but it is fragmented. What’s missing is the connection between where it lives and where the public, or the agency’s own leadership, or OGIS itself, can meaningfully see it.
A real EDC, a real log, a usable summary: none of these are meant to be extra work bolted onto FOIA processing. They’re what FOIA processing looks like when the transparency is built in instead of assembled by hand afterward. It holds for EDCs, logs, and everywhere else we’re pushing this next.
The FOIA log is exactly the problem Relativity FOIA was built to solve. For example, the Case Summary feature – live in the product today – targets subject-matter fields specifically. You can use AI to draft a summary, then empower a person on the FOIA team to decide whether it’s right.
The tool isn’t deciding what the record says or how it should be characterized publicly, but it can help with the first-draft work so a human doesn’t have to build every summary from a blank page. The human still signs off on the deliverable. In every place technology touches throughout the FOIA process, it should work the same way: assist, don’t determine.
If you run a FOIA or records program and you saw your own operation somewhere in that OGIS report, I’d love to hear where the gap bites for you. That’s the input that will help make the next version of what we build even better.
Learn more about Relativity FOIA here.
Graphics for this article were created by Caroline Patterson.
