A first post from Sediment Research, a knowledge graph built on the GAO bid protest decision record. This piece is pinned: every subsequent post and every Sediment pattern brief references it as the corpus baseline.
We have structurally analyzed 8,578 opinion-bearing GAO decisions plus 10,972 docket-only filings (19,550 protests total) from 1991 to 2026. The opinion-bearing sample is roughly half to two-thirds of GAO's published merit decisions, depending on the year. Absolute rates therefore run below GAO's published annual aggregates. The within-corpus comparisons that follow are the reliable signal, and that is what these three patterns are built on.
Each is the kind of fact a practitioner recognizes as “obvious in hindsight” the moment they see it laid out. The annual-report cycle and the case-by-case literature both obscure them.
1. The headline sustain rate is a denominator trick
| Disposition | Cases | Share |
|---|---|---|
| Dismissed / withdrawn / unresolved | 12,209 | 62.5% |
| Reached merits decision | 7,341 | 37.5% |
| Total protests filed | 19,550 | 100% |
| Of the 7,341 merits decisions | Cases | Rate |
|---|---|---|
| Sustained (including partial) | ~1,000 | 13.6% |
| Denied | ~6,340 | 86.4% |
Annual reports and roundups cite the merits-only sustain rate. As an all-time average that figure is 13.6% in our corpus; year by year it sits between 6.3% and 12.4%, depending on docket composition.
The all-in rate (sustains as a fraction of all protests filed, not just of merits decisions) is 5.1% (~1,000 of 19,550). The 8.5-percentage-point gap between the two numbers is the cost of the filter the merits-only rate applies. Two-thirds of the corpus never reaches a merits ruling at all.
| Year | Sustained | Merits | Rate |
|---|---|---|---|
| 2014 | 20 | 188 | 10.6% |
| 2015 | 43 | 414 | 10.4% |
| 2016 | 49 | 396 | 12.4% |
| 2017 | 34 | 363 | 9.4% |
| 2018 | 30 | 375 | 8.0% |
| 2019 | 26 | 361 | 7.2% |
| 2020 | 30 | 365 | 8.2% |
| 2021 | 21 | 334 | 6.3% |
| 2022 | 21 | 273 | 7.7% |
| 2023 | 20 | 249 | 8.0% |
| 2024 | 25 | 280 | 8.9% |
| 2025 | 25 | 225 | 11.1% |
The year-by-year merits rate hovers steadily inside that 6%–12% window. Each annual report frames its single number as news. The 12-year record treats it as a constant; the more interesting variable is the filter, not the rate.
Sustain rate denominator: sustained / (sustained + denied). Dismissed and withdrawn protests are excluded from the merits denominator and counted separately above.
2. The ground a protest turns on swings the sustain rate about fourfold
| Ground (doctrine) | Decided | Rate |
|---|---|---|
| Solicitation compliance (own rule applied unevenly) | 570 | 19.1% |
| OCI determination reasonableness | 413 | 13.8% |
| Statutory / regulatory compliance | 1,528 | 11.6% |
| Price / cost realism | 1,352 | 10.6% |
| Discussions adequacy | 1,008 | 10.0% |
| Evaluation reasonableness & documentation | 6,440 | 8.5% |
| All decomposed grounds (baseline) | 12,280 | 7.9% |
| Proposal responsiveness & timeliness | 2,266 | 6.6% |
| Protest-grounds sufficiency / burden of proof | 1,893 | 6.0% |
| Protest timeliness | 1,229 | 5.1% |
| Corrective action adequacy | 316 | 4.8% |
Sustain rates are not even across the grounds a protest can turn on. By doctrine they run from about 5% to 19%, a near-fourfold spread, against a 7.9% rate across all decomposed grounds.
The high end is narrow and specific. Solicitation compliance, where GAO finds the agency applied its own stated requirement unevenly by relaxing it for the awardee, sustains at 19.1%, more than twice the all-grounds baseline. OCI determination reasonableness follows at 13.8%. These are the grounds where the agency departed from a fixed rule, and the record either shows it or it does not.
The grounds practitioners file most sit below the line. Evaluation reasonableness and documentation, the single largest ground at 6,440 decided, sustains at 8.5%; proposal responsiveness and timeliness at 6.6%; and the procedural grounds, protest-grounds sufficiency, timeliness, and corrective-action adequacy, sit near 5% at the bottom. Where a protest lands in the doctrine map moves the sustain rate more than which ground is fashionable in a given year.
Sustain rate is sustained / (sustained + denied), at the ground level; a single decision can raise several grounds and count toward more than one. Dimension figures are from the May 2026 corpus snapshot.
3. Two years past Loper Bright, the corpus sustain rate has not moved
| 24-month window | Sustained | Decided | Rate |
|---|---|---|---|
| Pre-Loper (Jul 2022 to Jun 2024) | 40 | 521 | 7.7% |
| Post-Loper (Jun 2024 to May 2026) | 45 | 485 | 9.3% |
Loper Bright Enterprises v. Raimondo (June 28, 2024) eliminated Chevron deference in Article III courts. The GovCon bar framed it as a paradigm shift for protest outcomes. The corpus shows the merits sustain rate moved by 1.6 percentage points on ~500 decisions per window. That is inside the noise band, and the direction (slight increase) is the opposite of what most pre-decision commentary predicted.
The doctrinal lens makes it obvious in hindsight: GAO reviews agency procurement decisions under its own reasonableness standard, not under Chevron-style deference to agency statutory interpretation. Loper Bright changed the deference doctrine that GAO did not use. The GAO record had no mechanism for it to move.
We refresh these numbers from a dated corpus snapshot as the record extends; the figures above are the May 2026 snapshot. Every Sediment pattern brief links back to this piece as the corpus baseline, then goes one level deeper into a single ground: how GAO has actually defeated and sustained it, with the cases on each side. Those ground-level notes publish here as they ship.
If you work in GAO bid protests and want to see what the structural record says about a specific ground, agency, or counsel of record, fifteen minutes is what we ask for: calendly.com/sediment-research.
— Sediment Research