Contractors that push an agency into corrective action cannot assume the second evaluation will look like the first. In CSA Global LLC d/b/a Client Solution Architects, B-423877.4, dated Aug. 28, 2026 and posted in redacted form on Sept. 17, GAO denied a challenge to the Army's $299.2 million task order to incumbent Valiant Global Defense Services Inc. for the Army National Guard (ARNG) Mission Command Training Support Program (MCTSP). CSA, of Virginia Beach, Virginia, priced $10,575,753 lower, a gap GAO put at approximately $10.5 million. Evaluators who found zero risks in its staffing plan the first time found one the second time, and GAO said that was permissible.

How the MCTSP Order Moved From GDIT to Valiant

The Army issued task order request for proposals No. W900KK-25-R-ARNG on July 2, 2025, to the large business pool of its Mission Training Complex Capabilities Support multiple-award contract. The order covers training support from all six ARNG mission training complexes (MTCs) in live, virtual, constructive and gaming environments. It runs a 1-month phase-in, an 11-month base period, four 1-year options and a 6-month extension.

Six offerors submitted proposals. In August 2025 the Army selected General Dynamics Information Technology Inc. (GDIT). Valiant, of Herndon, Virginia, and CSA both protested, the Army took corrective action, and GAO dismissed both protests as academic in September 2025.

The Army then reevaluated the technical factor only. In April 2026 the source selection authority (SSA) picked Valiant at a total evaluated price of $299,177,674 over CSA's $288,601,921. Valiant earned Good on all three technical subfactors. CSA earned Acceptable for staffing plan and for recruitment and retention, and Good for phase-in. GAO's docket shows GDIT filed its own protest on May 26, 2026, and withdrew it on July 2. CSA filed on June 1.

The 20-Person Math Behind CSA's New Staffing Risk

The performance work statement requires a simulation support team (SST) large enough to “ensure each MTC can conduct one large (multi-echelon) and one medium scale (single echelon) JLCCTC [joint land component constructive training capability] supported exercise simultaneously.” On reevaluation, the evaluators saw risk in CSA's ability to do that.

CSA called that characterization factually inaccurate. GAO went to the proposal, where CSA wrote that the JLCCTC tool needs an SST lead, eight constructive simulation engineers and a system engineer to operate. In GAO's words, “10 personnel from the TSTs proposed by CSA are required to run a training event that uses the JLCCTC simulation tool.” TSTs are CSA's proposed technical support teams.

Two simultaneous events would take 20 people. The public decision redacts CSA's actual headcounts, but GAO found the proposed teams fell short at each of the six MTCs. Even crediting CSA's redacted workaround, GAO wrote, “there still simply are not enough personnel” to run a medium and a large event at once.

Why GAO Said a Second Read Can Differ From the First

CSA's core argument was about process. GAO quoted the protester's question: “if the original evaluation was thorough enough to support an award decision, why did a second look at the same proposal by the same team produce a materially different risk assessment?”

GAO rejected the framing. “Here, CSA's suggestion that the agency's initial evaluation constitutes a touchstone against which its final evaluation must compare presents a false premise; instead, the overriding concern in our review of a reevaluation following corrective action is not whether the final evaluation is consistent with an earlier evaluation, but whether it is reasonable and consistent with the solicitation,” the decision states.

GAO acknowledged one narrow exception. In eAlliant, LLC, B-407332.6, it held that an SSA facing “starkly different” conclusions about essentially the same proposal had to reconcile or explain them. A shift from one merit and zero risks to one merit and one risk, with the Acceptable rating unchanged, did not qualify. GAO found “no basis to question the reasonableness of the agency's reevaluation where, as here, the record indicates the evaluators simply noticed something they previously had not upon rereading CSA's proposal--similar to the way in which a person can notice a new facet of a character or story line upon rereading a novel.”

Did the Army Justify the $10.5 Million Premium?

CSA argued the tradeoff “failed to look behind the adjectival ratings and compare the qualitative substance of the competing proposals.” GAO found the SSA did compare them. The SSA credited Valiant with “additional benefits related to Key Personnel, Knowledge Management and JLCCTC staffing approaches” under the staffing plan, the most important subfactor, and with a more robust plan to mitigate recruitment and retention risk, the second most important.

The SSA concluded CSA's approach “is not the best value to the Government despite its lower proposed price due to the higher performance risk and fewer technical merits identified in the comparative analysis.” GAO found no basis to question that judgment.

Valiant's Hold on MCTSP: a $138.9M Order, Then an Uncompeted Bridge

Valiant's March 2022 announcement of a $124 million MCTSP task order said the company first won the program in 2018. Recompeted reported that the 2022 order, W900KK22F0013, ran to Dec. 21, 2024, on $138,876,443.89 obligated.

The Army then kept Valiant in place with a bridge. The USAspending record for contract W900KK25C0004 lists a firm-fixed-price award coded “NOT COMPETED,” with one offer received, a period of performance from Oct. 22, 2024, to Jan. 21, 2026, $57,641,685.15 obligated and a base-and-all-options value of $98,115,881.47.

What It Means for Contractors

A protest that ends in corrective action locks in nothing. Every rating, including the protester's own, is open again, and GAO will measure the new evaluation against the solicitation, not the old scoresheet.

When a work statement requires simultaneous events, evaluators will multiply the offeror's own per-event headcount and compare the product to the organizational chart. CSA supplied the 10-person figure GAO used against it. Proposals that lean on shared personnel for concurrent requirements must show the numbers close at every site.

CSA's $10,575,753 price advantage, about 3.5 percent of Valiant's evaluated price, did not overcome Good-versus-Acceptable ratings on the two top subfactors once the SSA documented the technical benefits behind them.

Sources