Recompeted1,688 Protests, and Half of Them Worked
GAO's protest docket just posted its second-quietest year since 2020, while the share of protesters walking away with relief held at 52%. The chart every capture team should keep taped to the wall.
1,688.
That is the number of bid-protest cases filed at GAO in fiscal 2025, per the annual report GAO sends Congress. It is down 6% on the year and 21% from fiscal 2020, and outside one anomaly year, the direction has been the same since the start of the decade.
Here is the number that refuses to move with it: 52% of protests closed with the protester getting some form of relief. Same as last year. Within a few points of every year since 2020 (chart below).
"If protests work half the time, why is everyone protesting less?"
Read the two numbers together
The 52% is GAO's effectiveness rate, and it is the honest statistic in the report. The famous one, the sustain rate, was 14% in FY25: of 380 cases that went the distance to a merit decision, 53 protesters won.
The gap between 14% and 52% is the mechanism. Most protest relief never arrives as a sustain; it arrives as voluntary corrective action, the agency reading the protest, seeing the problem, and pulling the award back before GAO writes anything down.
So a protest is less a lawsuit than a forced second look, and the second look finds something worth fixing about half the time. That rate has been stable through five years of falling volume, which tells you the protests being filed are not getting weaker. There are just fewer of them.
The FY23 bar is the exception that proves the mechanics: filings jumped to 2,025 and the sustain rate hit 31%, driven by a pile of protests against a single procurement (HHS's CIO-SP4). One badly handled mega-vehicle can move a whole fiscal year of national statistics.
What this means for a pursuit decision
For the operator, the chart compresses to a simple prior: a debrief that smells wrong is a coin flip, and the industry is leaving flips on the table.
A few sharper edges on that:
- The 100-day clock held. GAO issued every FY25 decision inside the statutory window. The cost of finding out is bounded and known.
- Relief is not the contract. Corrective action usually means a re-evaluation or amended solicitation. You get another shot at the work; you do not get the work.
- The prior shifts with the paper. 52% is the base rate across every protest filed, weak ones included. A protest built on a specific, documented evaluation error is playing better odds than the average; a reflex protest to buy transition time is playing worse ones, and agencies can tell which is which.
And if you want to watch the statistic being made this quarter: the Army's $50 billion MAPS professional-services vehicle is sitting under a double-digit pile of pre-award protests, with GAO decisions due between mid-August and late October. The Army wanted awards in September. The protesters, collectively, have already moved that.
Final Thoughts
A falling filing count with a flat effectiveness rate is a strange equilibrium: the tool works as well as it ever has, and the industry reaches for it less. Whatever is driving that (protest fatigue, customer-relationship math, the OTA share of spend growing outside protest jurisdiction), it is not the odds.
We track the docket daily and the annual statistics once a year. This is the chart we will update next December, and we would not bet on the orange number moving.
Thanks for reading.