Full opinion text
CARNES, Circuit Judge: Latecoere International, Inc., unsuccessfully bid for a contract with the Naval Training Systems Center (“the Training Center”), of the United States Department of the Navy, to provide two gravitational forces pilot training systems. After the Navy awarded the contract to Latecoere’s competitor, Environmental Tectonics Corporation (“ETC”), La-tecoere commenced this action seeking declaratory and injunctive relief against the United States, the U.S. Department of the Navy, and H. Lawrence Garrett III, in his capacity as Secretary of the Navy (collectively “the Navy”). The Navy moved for dismissal or judgment on the pleadings. Without holding an evidentiary hearing, but after considering evidence outside the pleadings, the district court affirmed the award to ETC, essentially granting summary judgment for the defendants. Latecoere appeals from that ruling. We find evidence in the record: that Navy officials involved in the procurement process arbitrarily increased ETC’s ratings in order for it to obtain the award; that no rational basis exists for the award to ETC instead of Latecoere; that the award was motivated by bias for ETC and against Latecoere, which was based at least in part upon improper consideration of the fact that ETC is an American company while Latecoere is a French company; and that, but for bias and improper considerations, the Navy would have awarded the contract to Latecoere. Accordingly, we reverse the judgment for the defendants. We begin with a discussion of the facts divided into four subparts. Part LA describes the applicable procedures for evaluating proposals and awarding a contract such as the one involved in this case; it tells how the procurement process is designed to work. Part I.B discusses what happened the first time through in the evaluation, recommendation, and all but the final selection stages of the process; it tells how the process almost did work as designed. Part I.C discusses what happened when the matter reached the office of the Assistant Secretary of the Navy and improper “Buy American,” political concerns were raised; it tells where things went wrong. Part I.D discusses what happened the second time through in the evaluation, recommendation, and selection stages; it tells how the process was manipulated to bring about an improper result. Part II discusses the procedural history of the case. Part III describes the applicable standards of review. Part IV contains our legal analysis. Part V presents our conclusions and directions for the district court. Because there are a number of individuals involved and because some of the terminology is military-procurement jargon, we provide, in Appendix A to this opinion, a combined cast of characters and glossary of terms. I. THE FACTS As we explain later in this opinion, the district court granted judgment against La-tecoere without conducting an evidentiary hearing but after considering matters outside the pleadings. For that reason, we treat the ruling as one of summary judgment, see Fed. R.Civ.P. 12(c), and we review the evidence and all inferences from it in the light most favorable to Latecoere, see Miranda v. B & B Cash Grocery Store, Inc., 975 F.2d 1518, 1532 (11th Cir.1992); Part III.B, below. We will state the facts accordingly, which is not difficult to do because so many of the important facts are undisputed and clearly favor Lateeoere’s position, anyway. In January 1989, the Training Center, which is located in Orlando, Florida, issued Request For Proposals No. N61339-89-R-0004 for the design, construction, and maintenance of two G-Tolerance Improvement Program (“G-Tip”) Training Systems: one at Lemoore, California, and one at Cecil Field in Jacksonville, Florida. In naval procurement, a “request for proposals” is essentially a detailed solicitation of bids, so we will refer to this one as “the Solicitation.” A G-TIP Training System is used to train pilots of high performance jet aircraft to avoid losing consciousness when they encounter strong gravitational forces during sudden acceleration. The system centers on a centrifuge training device, which is essentially a gondola (in which a pilot sits) attached to an arm that propels the gondola (with the pilot inside) in circles. The device is housed in a facility equipped to provide instructional, medical, maintenance, and administrative services. Both the facility and the device are part of the system for which bids were solicited. A. THE PROCEDURES FOR EVALUATING PROPOSALS AND AWARDING THE CONTRACT: HOW THE PROCESS IS DESIGNED TO WORK The Solicitation contained lengthy and detailed technical specifications for the design of the G-TIP Training Systems. Those specifications stated that, when making design compromise considerations, the highest priority would be given to the students’ safety. The Solicitation also set forth the procedures for evaluating offerors’ proposals and for awarding the contract. Offerors were directed to divide their proposals into six volumes, corresponding to major evaluation areas: (I) Device Design; (II) Facility Design; (III) Management Plan; (IV) Integrated Logistics Support; (V) Past Performance; and (VI) Cost. Each proposal volume was further divided into chapters. The first five volumes constituted a technical proposal that was to be evaluated separately from the cost proposal, which was the sixth volume. A Source Selection Evaluation Board (“the Evaluation Board”), consisting of eleven members — and assisted by some thirty advis-ors — all of whom were engineers, scientists, medical doctors, or other specialists, was charged with evaluating the technical proposals. The G-TIP procurement Contracting Specialist, Frank Barbieri, a member of the Evaluation Board, was charged with evaluating the cost proposals. The Solicitation stated that the contract would be awarded to “that offeror submitting the ‘best value’ proposal,” meaning “the best technical proposal with appropriate consideration given to cost and other factors.” With respect to cost, the Solicitation described the appropriate consideration as follows: Cost is not expected to be the controlling factor in the selection of a contractor for this solicitation. The degree of importance of cost as a factor could become greater depending upon the equality of the proposals for other factors evaluated; where competing proposals are determined to be substantially equal, total cost and other cost factors would become the controlling factor. The Solicitation also ranked all of the proposal volumes according to their order of importance, listing the cost volume as the least important. The Training Center specifically reserved, in the Solicitation, the right to award the contract on the basis of initial offers, without negotiations. The Solicitation assigned the five proposal volumes that comprised the technical proposal the following order of importance, beginning with the most important: Device Design, Facility Design, Management Plan, Past Performance, and Integrated Logistics Support. Teams of Evaluation Board members and advisors were organized to evaluate the chapters within each volume of the proposal and, on the basis of those evaluations, to prepare a narrative assessment identifying the strengths, weaknesses, and risks of each chapter of each volume of the proposal and to assign to each chapter a technical rating and a risk assessment rating. Chapter evaluations would be submitted to proposal volume team leaders for assignment of technical ratings for each volume. Ratings for each volume of the proposal were then to be submitted to the Evaluation Board’s Committee Chairperson for assignment of overall proposal ratings. There were four possible technical ratings: (1) “exceptional” (exceeding specified performance standards with a high probability of success and no weaknesses); (2) “acceptable” (meeting specified performance standards with a good probability of success and no significant weaknesses); (3) “marginal” (either failing to meet performance standards but with deficiencies that were correctable without a major rewrite, or failing to provide sufficient information to determine acceptability); and (4) “unacceptable” (failing to meet specified performance standards due to deficiencies that were not correctable without a major rewrite). There were also three possible risk assessment ratings — “low,” “moderate,” and “high” — that measured potential risks of delay, increases in cost, or poor performance. In other words, the worst possible rating was “unacceptable” with “high” risk. In addition, the Solicitation identified certain chapters of the proposal volumes as “critical areas” of concern to the government, and warned that a proposal rated “unacceptable” for one or more critical chapters might be judged “unacceptable” overall. The Evaluation Board’s ratings and Contracting Specialist Barbieri’s cost proposal evaluations were to be forwarded to a Source Selection Advisory Council (“the Advisory Council”) comprised of high ranking officers at the Training Center. The Advisory Council would then compare the evaluations, recommend which offerors to include in the competitive range, and most significantly, recommend an offeror for the award. Only those offerors in the competitive range could be asked to engage in negotiations with the G-TIP procurement Contracting Officer, Cheryl Hall, and to submit best and final offers if there were negotiations. As previously stated, however, the Training Center reserved the right to award the contract without negotiations, based on initial offers. A Source Selection Authority (“the Selection Authority”) was charged with making what should have been the final selection decision. After that “final” selection decision was made, Contracting Officer Hall was responsible for obtaining the necessary approval of the business aspects of the contract. Pursuant to Navy regulations, the Assistant Secretary of the Navy for Shipbuilding and Logistics had to approve the “business aspects” of the contract. See 4 Gov’t Cont.Rep. (CCH) ¶ 44,043 (Navy 1989). After obtaining the Assistant Secretary’s approval, Contracting Officer Hall was to award the contract to the selected offeror. To summarize the process, the Solicitation set out the specifications for proposals and the rules and guidelines under which the contract was to be awarded. The Evaluation Board and its advisors, some forty technical specialists in all, were to evaluate the technical aspects of the proposals submitted. The Contracting Specialist was to evaluate the cost aspects. The evaluations of the Evaluation Board and the Contracting Specialist were to go to the Advisory Council, which consisted of the Training Center’s senior officials. The Advisory Council was to compare the evaluations, determine which offers were within the competitive range, and recommend one of the offerors for the award. The Selection Authority was to make the actual decision and to award the contract to any offeror within the competitive range, with or without negotiations. Before the award was made, however, the Contracting Officer had to obtain approval of the Assistant Secretary of the Navy, whose concerns were supposed to be limited to the “business aspects” of the contract. B. THE FIRST EVALUATION, RECOMMENDATION, AND SELECTION: HOW THE PROCESS ALMOST DID WORK Five firms submitted G-TIP proposals: Latecoere, ETC, Brown & Root, Wyle Laboratories, and Krug International. All of the firms are headquartered in the United States, except Latecoere which is a French firm. Under a Memorandum of Understanding between the United States and France for reciprocal purchases of defense equipment, the Navy is required to treat all French firms, including Latecoere, as it would an American firm. See Defense Dept. Reg. ¶ 37,165 T-211 (1989). The Evaluation Board evaluated all five offerors’ proposals and gave the ratings that are set out in Appendix B to this opinion. Latecoere and Brown & Root were the only offerors to receive overall proposal ratings of “acceptable”; ETC, Krug, and Wyle received overall proposal ratings of “unacceptable.” ETC’s proposal was woefully deficient: it failed to comply with soldering requirements, an essential safety feature; its foundation design did not incorporate the required drilled piers; its foundation design did not evaluate settlement or cyclic loading; its facility design failed to incorporate the required fire alarm and sprinkler systems; it failed to address data management areas and failed to provide information on computer selection; its proposal improperly redefined the requirements of commercial off-the-shelf equipment; and its building square footage exceeded the maximum allowed. Moreover, ETC’s past performance record was less than exemplary; in the 18 months preceding submission of the G-TIP proposals, government agencies with whom ETC had contracted had initiated three adverse actions against ETC for failing to perform in a timely and competent manner. In one instance, ETC was three years behind schedule. Based on the overall proposal ratings, the Evaluation Board recommended that only Latecoere and Brown & Root were capable of satisfying the requirements for the G-TIP, systems contract and that only they should be eligible for the award. After .reviewing the Evaluation Board’s ratings and comparing cost proposals, the Advisory Council agreed with that recommendation and placed only Latecoere and Brown & Root in the competitive range. As between those two offerors, the Advisory Council recommended that the contract be awarded to Latecoere, without negotiations, because it was both the superior-rated and the lower-cost offeror, proposing a cost of $20,911,166 as compared to Brown & Root’s proposed cost of $35,199,-045. In short, the Advisory Council, which consists of the Training Center’s senior officials, agreed with the approximately forty engineers, scientists, medical doctors, and other specialists who were on or assisted the Evaluation Board. They all agreed that La-teeoere’s proposal was acceptable and ETC’s was not. The Selection Authority, Captain Rowley, agreed as well. He stated: “Based on my independent review of the pertinent documents and discussions, I have reached an independent determination that the Latee-oere International Inc. proposal is the best value technically acceptable proposal with no significant deficiencies.” In a selection decision document, Rowley described Latecoere’s proposal as offering “a superior [technical] evaluation rating ... with the lowest overall cost to the Government at a fair and reasonable price.” He then recited “key factors which led to [his] determination”: [O]f the five offerors submitting proposals, ETC, Krug and Wyle are unacceptable. In my judgment the three offerors failed to meet the requirements of the solicitation and would require major redesign to meet the requirements. It is my opinion that added discussions with all the offerors would not insure that the unacceptable of-ferors’ redesign and resubmittal of their proposals would result in an appreciable technical and cost advantage to the Government. [Thus,] ... the technically unacceptable offerors, ETC., Krug and Wyle no longer have a reasonable chance of being selected for contract award. To farther delay this project would ultimately delay the essential training of aircrews and could result in farther loss of life and resources due to the G — Loss of Consciousness problem. ... To specifically identify or speculate on the monetary losses that would substantially effect this project are difficult to delineate. The inherent risks of operating a safe and efficient centrifuge are much too high to award this project to other than the most technically acceptable contractor. The unacceptable offerors’ inability to grasp the technical significance of the turn key G-TIP requirements presents an untenable degree of risk to human life. The geotechnical engineers ..., the facility experts for the Department of Defense, have identified two of the offerors, ETC. and Krug, as lacking the technical capability to handle the complex foundation requirements of the G-TIP. Wyle failed to meet a minimum level of understanding of the device design and operation. The Government cannot afford to allow this type of technical risk in the G-TIP. Two Government centrifuge projects which experienced substantial losses due to device support failures were: a) NASA, Houston, TX and b) NAS Pensacola, FL projects. At NASA a device foundation failure resulted in disassembly of the device and abandonment of facility, and at NAS Pensacola a shift of the device foundation has limited the use of the device. The negative results experienced, due to poor technical planning, at the NASA and Pensacola centrifuge sites could be the type'loss encountered on the G-TIP if selection is driven solely by low price. The magnitude of effort required from each of the unacceptable offerors to meet the minimum requirements of this solicitation would most certainly result in substantial redesign effort and significantly increase the proposed cost. It is apparent that the amount of technical proposal rework by the three unacceptable offerors would result in additional costs that would approach or exceed the current lowest cost technically acceptable offeror. (Emphasis added.) Thereafter, Hall, the G-TIP procurement Contracting Officer, submitted to the Assistant Secretary of the Navy for Shipbuilding and Logistics a pre-negotiation business clearance memorandum (“clearance memorandum”) seeking approval of the Training Center’s' decision to award the G-TIP contract to Latecoere. Attached to the clearance memorandum were: copies of the technical evaluation factors contained in the Solicitation; the Advisory Council’s report and recommendation; the Contracting Officer’s competitive range determination; and Selection Authority Rowley’s selection decision document. The Training Center requested expeditious review of the G-TIP clearance memorandum; there was reason to believe that lives might be at stake. Up to this point, the process had worked well. The board of experts who impartially analyzed the proposals, the Training Center’s senior officials, and the commanding officer of the Training Center who was authorized to make the final selection, had all done their jobs. They were all in agreement: Latec-oere should be awarded the contract; ETC should not. All that remained was for the office of the Assistant Secretary of the Navy for Shipbuilding and Logistics to sign off on the decision. The clearance memorandum and the attached documents that were sent to the office of the Assistant Secretary of the Navy made the superiority of Latecoere’s proposal clear, and expressly warned that choosing an unacceptable proposal, such as that of ETC, would present “an untenable degree of risk to human life.” Thus, there was every reason to expect that the Assistant Secretary’s office would approve the award to Latecoere. It did not. C. THE OFFICE OF THE ASSISTANT SECRETARY OF THE NAVY AND “BUY AMERICAN” CONCERNS: WHERE THINGS WENT WRONG The Assistant Secretary of the Navy for Shipbuilding and Logistics is authorized to disapprove, approve, or conditionally approve a business clearance memorandum setting out the “business aspects” of an award decision. 4 Gov’t Cont.Rep. (CCH) ¶ 44,043, at 1.690-l(a). The Assistant Secretary of the Navy had designated Frank Ford, the Assistant Director of Contract and Business Review, to exercise these responsibilities. Michael McDonald, a senior procurement analyst under Assistant Director Ford, was responsible for initially reviewing the G-TIP clearance memorandum. McDonald has no background in science or engineering and had no experience with G-TIP Training Systems. It was not his job to second guess the Training Center’s technical proposal evaluations; he was supposed to determine only whether the offerors excluded from the competitive range were excluded for good reasons and whether the reasons had been properly documented. Cf. 4 Gov’t Cont.Rep. (CCH) ¶ 44,043, at 1.690-l(a) (requiring the Assistant Secretary to clear the “business aspects” of a proposed contract action). After an initial review of the G-TIP clearance memorandum, McDonald gave Marie Garrett, a procurement analyst under his supervision, responsibility for an in-depth analysis of it. Based upon his “quick review,” McDonald apprised Garrett that he believed approval of the G-TIP clearance memorandum would be a “problem,” because lower-priced offerors (Krug and ETC) had been rated “unacceptable” and excluded from the competitive range and also because the award would go to a French company instead of an American company. McDonald attached the following note to the clearance memorandum: “Marie,. I’m a little leery of awarding a contract to,a French company w/o discussions — Don’t we have to apply Buy American factors? ... This could prove to be a political hot potatoe [sic] if we give to France without giving American firms ‘a chance.’” It is undisputed that the injection of “Buy American” considerations into the procurement process was improper. The United States and France had entered a memorandum of understanding for the reciprocal purchases of defense equipment that obligated everyone involved in, this procurement decision to treat Latecoere as if it was an American company. The Navy has conceded that the country of origin could not properly have been given any weight or consideration at all. But it was. After McDonald wrote the note raising political “Buy American” considerations, he called Captain Kalapos, who was the Training Center’s Director of Contracts. Kalapos was a member of the G-TIP procurement Advisory Council, and McDonald asked him to determine whether ETC and Krug could make their proposals “acceptable” without a major rewrite. Kalapos explained to McDonald why they could not. McDonald asked him to put the explanation in writing. Thereafter, McDonald received a facsimile from the Training Center, stating: [Discussions with [Krug and ETC] ... would require major redesign/rework of the building and structural design proposals] submitted by these offerors. Krug’s proposal presented a simplistic approach to the complex foundation problem at NAS Lemoore which would result in an unusable centrifuge due to settling. ETC’s [sic] acknowledged that a viable design was not submitted for the critical device foundation and that further analysis would be needed before one could be developed. As proposed by ETC the device foundation for NAS Lemoore would provide an unstable base that would have deflected upward (move upward in the ground) in the plastic, expansive clays at that site. Of course, that explanation was in addition to the selection' decision document that had explained that awarding the contract to ETC would not save money but would instead cause delay and create “an untenable degree of risk to human life.” Thereafter,. Assistant Director Ford personally met with the Training Center’s senior officials to discuss the G-TIP clearance memorandum, which is the document that the Training Center had submitted to the Assistant Secretary’s office to obtain its approval to award the contract to Latecoere. At the meeting, Ford expressed concern about the potential political consequences of excluding American companies from the competitive range and awarding the contract to a foreign company without negotiations. Ford also indicated that unless there were negotiations with all of the offerors, regardless of their unacceptable ratings, he would not approve the G-TIP clearance memorandum. Following the meeting, Ford approved the clearance memorandum on the condition that the Training Center “enter into negotiations with the offerors.” D. THE SECOND EVALUATION, RECOMMENDATION, AND SELECTION: HOW THE PROCESS WAS MANIPULATED TO CHANGE THE RESULT As a result of the condition that Assistant Director Ford attached to his approval of the G-TIP clearance memorandum, the Training Center’s Evaluation Board interviewed all of the offerors and solicited revised proposals from them. The offerors were questioned about deficiencies identified in their first proposals. The Evaluation Board had developed approximately 800 such questions, but asked only those that it thought were critical to the re-evaluation. The offerors received copies of the questions shortly before the interviews, and though they had an opportunity to respond during their interviews, they were asked to submit written responses with their revised proposals. Also, pursuant to an amendment to the Solicitation that eliminated the Cecil Field site from the contract, offerors were instructed to delete all references to that site from their revised proposals. The Evaluation Board assessed the answers and the revised proposals that it received from the five offerors, using the same 'rating scheme it had used before. The new proposal volume ratings are set out in Appendix C to this opinion. After the second evaluation, some of the offerors (not Latee-oere) still had “marginal” ratings in critical chapters. ETC had “marginal” ratings for four critical chapters of its Device Design, Volume I of the proposal, and had a “marginal” rating for Past Performance, Volume V of the proposal. When the Advisory Council met during the second evaluation, Kalapos, the Training Center’s Director of Contracts and a member of the G-TIP procurement Advisory Council, told other Advisory Council members that in his opinion the Training Center could not make an award to an offeror who had a “marginal” rating in a critical chapter. If Kalapos were correct, that would have excluded ETC. Kalapos then stated: “[E]ither fix the_thing or I am not going to make an award.” Thereafter, without any new information about or re-evaluation of ETC’s “marginal”-rated critical chapters, the Advisory Council simply increased ETC’s ratings in those chapters to “acceptable.” The Advisory Council ultimately rated all of the offerors’ proposals “acceptable,” except Latecoere’s, which was rated “exceptional.” All offerors were then asked to submit best and final offers, follows: The offers were as ETC Lateeoere Krug Wyle Brown & Root Offers $10,351,541 11,158,665 14,396,202 18,201,695 18,609,915 Difference from Lowest 807,124 4,044,661 7,850,154 8,258,374 Percentage Difference 7.8% 39.1% 75.8% 79.8% Based on these offers, ETC and Latecoere were the only offerors that the Advisory Council considered further. To assist in deciding between the two, the Advisory Council requested a comparative cost analysis. That analysis showed that Latecoere had proposed completing the project three months earlier than ETC and that Latecoere’s earlier completion date might well save both money and human lives. At least one recent aircraft crash had been caused by a lack of centrifuge training and more were anticipated absent increased availability of such training. By contrast to Latecoere’s likely completion date, ETC’s past performance suggested that delays of up to nine months were possible if ETC received the contract; such delays would increase the risk of loss of lives and aircraft. A logistics analyst also compared the long term maintenance costs for the facilities proposed by ETC and by Latecoere. His analysis revealed that, over a ten-year period, the maintenance costs for Latecoere’s proposed facility would be $554,556 less than the maintenance costs for ETC’s proposed facility. After comparing ETC’s and Latecoere’s costs and technical proposals, the Advisory Council again unanimously recommended La-teeoere as the “best value.” Before giving this recommendation to the Selection Authority, however, all references to the comparison of the long-term maintenance costs for ETC’s and Latecoere’s proposed facilities were deleted, possibly at the insistence of Kalapos, the Training Center’s Director of Contracts. Kalapos stated that maintenance costs for a ten-year period should not have been considered, because the Solicitation required offerors to provide maintenance for only one year and gave the Navy an option to contract with the offeror for maintenance for two years thereafter. By the time the Advisory Council made its second recommendation that Latecoere be awarded the contract, Rowley had retired from his position at the Training Center. Robert Urban had replaced Rowley as the Selection Authority. Urban had never before been involved in a best value procurement or acted as a Selection Authority. Selection Authority Urban met with Advisory Council members and G-TIP Contracting Officer Hall on several occasions to discuss the selection. During one of Urban’s meetings with the Advisory Council, Urban stated that the American companies might apply political “heat” if the award went to a foreign company. According to Contracting Officer Hall, during another of those meetings she was asked whether she believed, based on her direct dealings with all of the offerors, that either Latecoere or ETC was likely to protest if the other was selected for the award.. She responded, to Selection Authority Urban, and everyone else present, that Latecoere was not likely to protest an award to ETC, but that ETC was likely to protest an award to Latecoere because “ETC. had been pretty vocal and active in the procurement and ... [it would] have nothing to lose.” Shortly before making his selection, Urban called Assistant Director Ford, whom he had known for years, to ask him about the definition of a best value procurement and to inquire about “a persistent rumor” at the Training Center that Ford had a problem with an award to a French firm. He asked Ford, “[w]as there any truth” in “the rumor about the negative feelings about French firms in procurements?” According to Urban, Ford said he had absolutely no bias against a French firm. Ultimately, Selection Authority Urban rejected the Advisory Council’s recommendation to award the contract to Latecoere and selected ETC for the award, instead. To support his selection, Urban prepared a source selection decision document that explained his reasons for favoring ETC’s proposal over Latecoere’s. With respect to La-teeoere’s and ETC’s proposals for Device Design (Volüme I of the proposal), Urban stated: “[Although both ETC and Latecoere were rated the same on the volume level, the proposals were not technically equal. Of the 12 critical chapters ... in this volume, Latec-oere received two exceptional ratings in the chapters for Gondola and Arm.” Urban made no mention of the fact that ETC had had “marginal” ratings in four critical chapters in the device design proposal, and that those “marginal” ratings had been arbitrarily increased to “acceptable” without any reevaluation of those chapters. There is no evidence that Urban was aware that those arbitrary alterations had been made. After highlighting the superior features of Latec-oere’s proposals for the Gondola and Arm, Urban stated: “It is my opinion that the technical benefits offered by Latecoere for future requirements, ... exceed the stated requirements of the [Solicitation] and do not equate to value added worthy of the increased monetary difference.” That monetary difference was an additional $818,439 that Latecoere’s device would have cost up front. For Facility Design (Volume II of the proposal), Urban first recognized the chapters in the non-foundation design portion of this proposal for which Latecoere had received “exceptional” ratings — layout, square footage, mechanical system, appearance, quality of materials and equipment, site utilization and parking, acoustic attenuation, termite control, and roof hatches. Then, he stated, without any explanation: “Although these enhancements are in excess of the [Solicitation], they do not appear to provide substantial benefits to the function of the facility.” Turning to the foundation design portion of the Facility Design proposal, Urban explained that Latecoere was rated “exceptional” in this area because of its proposal to consult with a geotechnical engineer. The California site was in an earthquake zone, and the engineer in question had significant experience with seismically-aetive areas. Consequently, the Evaluation Board determined that Latecoere’s proposal had a “ ‘high probability of success.’ ” Urban rejected this determination, stating: “This statement induces some doubt as to the rating if a high probability is all that can be assessed to the construction of the ‘Exceptional’ rated designed device foundation.” Although Latec-oere’s foundation cost proposal was $4,000 less than ETC’s, Urban stated that no “significant improvements [were] to be gained by Latecoere’s proposal.” For Management Plan (Volume III of the proposal), Urban conceded “that the Government would benefit” from Lateeoere’s proposal to complete the project three months ahead of schedule, but stated that he had “no way of apportioning a specified value to earlier delivery,” because “the downing of any aircraft or loss of life due to lack of centrifuge training cannot be discussed in monetary terms of gains or losses.” Urban focused instead on the fact that “[t]here are currently other centrifuge devices operating in the United States which are being utilized for the required G-training.” He concluded that, “I cannot determine the extent or significance of the increased value associated with Latecoere’s proposal which offers delivery sooner than required by the [Solicitation], or that proposed by ETC, which complies with the stated requirement.” Urban also removed the evidence in the Advisory Council’s comparative cost analysis indicating that at least one recent aircraft crash had been caused by a lack of centrifuge training, and that more would likely occur unless access to such training increased. Urban was familiar with this evidence, but contended that it showed that G forces were only a “possible,” not a certain, cause of crashes. He otherwise failed to explain his conclusion that the government would derive no benefit from increasing G-force training, which would in turn decrease the “possibility” — or in the Advisory Council’s words, the “very real probability” — of more mishaps. For Past Performance (Volume V of the proposal), Urban conceded that ETC presented more of a performance risk than La-teeoere, stating: There is no doubt that the data supporting Latecoere’s Exceptional rating is appropriate. There is [sic] also several reasons to support the Marginal rating given to ETC. ... ETC received a Show Cause on 16 May 1988 under contract DLA120-87-C-4861 and on 1 December 1988 under contract N00600-88-C-0504. They were also issued a Cure Notice under contract N62477-82-C-0392. However, all of these actions were satisfactorily resolved and none of these actions were sustained or resulted in further adverse actions. [Also, ETC] was assigned a Marginal rating for poor performance (late delivery) under contract N62477-87-C-0015, N62477-85-C-0257, as well as continued poor performance under the previously referenced contract. Urban stated that: “The essential element for evaluation is [whether] these problems are contractor-caused,” and “[i]n each of these cases there is ho convincing evidence that [ETC’s] performance problems were solely contractor-caused.” Further, he said that any potential problems with ETC’s performance could be cured by “proper Government monitoring,” which the nature of the project would require anyway. Three days before Urban issued his selection decision document, ETC’s contract with the Defense Logistics Agency, DLA120-87-0^1861, was partially terminated for default. Urban was probably unaware of this action, because the Advisory Council had reported that a show cause letter issued to ETC on this contract had been canceled. That information was incorrect; it was not the show cause letter, but part of the contract, that had been canceled. For Integrated Logistics Support (“ILS”) (Volume IV of the proposal), Lateeoere was rated “exceptional” with “low” risk. ETC’s proposal was rated “acceptable” with “moderate” risk, primarily because ETC had not yet developed many of the specifics of its ILS proposal but, instead, had promised to do so if awarded the contract. After considering these ratings, Urban stated: It is obvious that Lateeoere provided an impressive proposal in the amount of detail and documentation already prepared to alleviate any misunderstandings as to the quality and extent of [its ILS analysis program]. What is not made clear, however, is the extent by which this preparation is far superior to that yet to be developed by ETC. Urban pointed out that, despite the fact that ETC’s proposal was not yet fully developed, ETC had estimated an overall cost for ILS that was $121,487 higher than Latecoere’s ILS cost proposal. Based solely on ETC’s higher cost estimate, Urban “reasoned” — and we use that term in the loosest sense — that ETC’s yet-to-be developed proposal would be as good as Latecoere’s.’ He stated: “I cannot support the [Advisory Council’s] conclusions that Latecoere’s ILS program is superior to that of ETC.” ETC, unlike all of the other offerors, had not. submitted a fully-developed ILS proposal. Lateeoere, by contrast, submitted' one deemed “exceptional” because of its impressive detail and documentation. Instead of giving Lateeoere credit for having a superior proposal, Urban chose to assume that ETC’s proposal, when and if it was fleshed-out, would be equal to Latecoere’s. The only basis for that assumption was that ETC had said it would charge more than Lateeoere for whatever proposal it finally developed. Thereafter, Assistant Director Ford approved a “post-negotiation business clearance memorandum” for the award of the G-TIP contract to ETC, and on March 22, 1990, the Training Center awarded the contract to ETC. In re Wyle Lab., Inc., 69 Comp.Gen. 648, 90-2 CPD ¶ 107, 1990 WL 293722 at *3 (1990) (hereinafter “Wyle”). II. PROCEDURAL HISTORY On March 29, 1990, both Lateeoere and Wyle filed protests with the General Accounting Office (GAO), arguing that the award to ETC “was not in accordance with the stated [Solicitation] evaluation scheme” and was unreasonable. Wyle, 1990 WL 293722 at *1, *5. Following a hearing, the GAO denied the protests. Id. at *1. Two months later, Lateeoere filed this action against the Navy in a federal district court, under the Administrative Procedure Act, 5 U.S.C. §§ 701 et seq. seeking: (1) a declaratory judgment that the award ... to [ETC] ... (a) violated applicable procurement laws and regulations, and (b) that the Navy on numerous occasions acted in an arbitrary and capricious manner and/or (c) abused its discretion ...; and ... (2) a preliminary and permanent injunction which (a) orders [the Navy] to terminate said contract award and enjoins any farther performance of said contract; and (b) mandates that [the Navy] award to Latecoere the contract; or in the alternative mandates a reopening of the procurement at the stage immediately preceding the improper Navy conduct[;] and/or (3) award of its attorneys’ fees and costs (a) incurred in pursuing this action and (b) if the Court is unable to fashion adequate equitable relief, [Latecoere] seeks award of its bid and proposal costs as well. Wyle is not a party to this action. The district court referred Latecoere’s case to a magistrate judge. Two weeks after filing the action, Latecoere filed a motion for a temporary restraining order (“TRO”) and for a preliminary injunction against ETC’s performance of the contract. That same day, the Navy moved to dismiss the complaint, or in the alternative, to require the joinder of ETC, as an indispensable party. On November 1, 1990, the magistrate judge issued recommendations on both motions. He concluded that ETC was an indispensable party and recommended its joinder. He also found that Latecoere had failed to demonstrate an “immediate and irreparable injury and a substantial likelihood of success [on the merits] as required for a [TRO].” Based on that finding and on the fact that Latecoere had waited two-and-a-half months before seeking the TRO, he recommended denying it. However, the magistrate judge granted Latec-oere’s request for discovery of evidence to support a preliminary injunction and recommended holding a hearing on that motion as soon as possible. The district court adopted the magistrate judge’s recommendations, denied the TRO, and ordered the joinder of ETC. ETC was added as a defendant. Following a hearing, the magistrate judge recommended denying Lateeoere’s motion for a preliminary injunction. The magistrate judge found that Latecoere still “ha[d] not shown a substantial likelihood of success on the merits.” He further found that, although Latecoere had demonstrated that it would suffer an irreparable injury if the injunction did not issue, that injury would not outweigh the injury that would be caused to ETC and to the Navy by the injunction. Finally, the magistrate judge found that the public interest did not require issuance of a preliminary injunction. Latecoere objected to the magistrate judge’s report and recommendation, arguing: the Magistrate Judge has felt hamstrung by (a) the legal presumptions attending award of Government contracts and (b) an adverse GAO decision which is based on what must be conceded to be a substantially incomplete and, therefore, misleading record. [Further] ... had facts developed through depositions and documentary discovery been available to the GAO at the time of its hearing, its decision would have been for Latecoere. Latecoere moved for the district court to remand to the GAO for an advisory opinion. The district court granted this motion, remanded to the GAO, and reserved ruling on Latecoere’s motion for preliminary injunction. On January 15, 1992, the GAO issued an advisory opinion, conceding that its prior decision was based on an incomplete record, but finding that even the “fuller record” supported its original denial of Latecoere’s protest. In re Latecoere Int'l, Inc. — Advisory Opinion, 92-1 CPD ¶ 70, 1992 WL 15029 at *2 (C.G.1992) (hereinafter “Latecoere”). After considering Latecoere’s evidence that Assistant Director Frank Ford, officials under him, and officials at the Training Center, including Selection Authority Robert Urban, were concerned about potential criticism of an award to a French company, the GAO concluded that this evidence established only general concern about whether the record adequately supported an award to Latecoere and indicated no bias against a foreign firm or for a domestic firm. Id. at *4-*5. Turning to Latecoere’s evidence that the G-TIP procurement Advisory Council had manipulated ETC’s ratings, increasing the ratings in four critical chapters from “marginal” to “acceptable,” the GAO concluded that, even assuming the Advisory Council’s decision to increase the ratings was unreasonable, La-tecoere was not “prejudiced by this change in the scoring of ETC’s [chapters], inasmuch as it apparently had no effect upon ETC’s ... overall technical rating of acceptable] with low risk for the device design [proposal].” Id. at *6. Relying on the GAO’s advisory opinion and on the magistrate judge’s recommendation, the district court denied Latecoere’s preliminary injunction motion. -Latecoere moved for reconsideration, which the district court denied. The Navy then filed a motion to affirm the award to ETC or for judgment on the pleadings. ETC supported the motion and Latec-oere objected to it. Without holding an evi-dentiary hearing, the district court granted the motion. It relied on the Supreme Court’s holding, in Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 420, 91 S.Ct. 814, 825-26, 28 L.Ed.2d 136 (1971), that courts should limit their review of agency actions to the record before the agency when. it made its decision unless the plaintiff has made “a strong showing of bad faith or improper behavior.” The district court found that Latecoere’s evidence did not demonstrate bad faith or improper behavior on the Training Center’s part. The court therefore determined that it should not, through an evidentiary hearing, expand the record beyond that which existed when it ruled on the preliminary injunction motion. The district court then determined that the record did “not show that [the Training Center’s] deeision to award the contract to ETC” lacked a rational basis or that the Training Center had violated “any applicable laws or regulations”; consequently, it denied Latecoere’s requested relief. Latecoere appeals from ■that ruling. Latecoere argues on appeal that, because the district court considered evidence outside the pleadings, this Court should treat the district court’s ruling as a summary judgment for the Navy and ETC. It argues that the record strongly suggests that, during the procurement process, Training Center officials committed acts that were irrational and in violation of procurement regulations. In light of this evidence, Latecoere argues that the district court should have proceeded to an evidentiary hearing on its demands for a permanent injunction and other relief. It also argues that this Court should order the district court to grant such relief. The Navy and ETC reply that the district court should be affirmed. III. STANDARDS OF REVIEW A. THE ADMINISTRATIVE PROCEDURE ACT Judicial review of the Training Center’s action is governed by the Administrative Procedure Act (“APA”), 5 U.S.C.A. § 706(2) (1977). -The APA provides in pertinent part: “The reviewing court shall ... (2) hold unlawful and set aside agency action, findings, and conclusions found to be — (A) arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law....” This standard requires a disappointed bidder to show “‘either that (1) the procurement official’s decisions on matters committed primarily to his own discretion had no rational basis, or (2) the procurement procedure involved a clear and prejudicial violation of applicable statutes or regulations.’ ” Kinnett Dairies, Inc. v. Farrow, 580 F.2d 1260, 1271 (5th Cir.1978) (quoting Kentron Hawaii, Ltd. v. Warner, 480 F.2d 1166, 1169 (D.C.Cir.1973)), quoted in Choctaw Mfg. Co. v. United States, 761 F.2d 609, 616 (11th Cir.1985). This deferential standard reflects the respect that reviewing courts are required to accord to agencies in their evaluation of bids and in their interpretation and application of procurement regulations. Kinnett Dairies, 580 F.2d at 1272. “While contracting officers may not opt to act illegally, they are entitled to exercise discretion upon a broad range of issues confronting them, including ‘considerations of price, judgment, skill, ability, capacity, and integrity’ in the selection of businesses with whom the government will enter into contracts.” Id. at 1277 (quoting Scanwell Laboratories, Inc. v. Shaffer, 424 F.2d 859, 869 n. 10, 874 (D.C.Cir.1970)). Accordingly, reviewing courts should be concerned with whether the contracting agency provided a coherent and reasonable explanation of its exercise of discretion. See, e.g., MCI Telecommunications Corp. v. FCC, 675 F.2d 408, 413 (D.C.Cir.1982); Environmental Defense Fund, Inc. v. Costle, 657 F.2d 275, 283 (D.C.Cir.1981). Proof that the award lacked a reasonable basis generally establishes arbitrary and capricious action. Keco Industries, Inc. v. United States, 203 Ct.Cl. 566, 492 F.2d 1200, 1203-04 (1974). Thus, if a reviewing court: finds a reasonable basis for the agency’s action, the court should stay its hand even though it might, as an original proposition, have reached a different conclusion as to the proper administration and application of the procurement regulations. ... Only when the court concludes that there has been a clear violation of duty by the procurement officials should it intervene in the procurement process and proceed to a determination of the controversy on the merits. M. Steinthal & Co. v. Seamans, 455 F.2d 1289, 1301-03 (D.C.Cir.1971), quoted in Kinnett Dairies, 580 F.2d at 1271. Also, proof of subjective bad faith by procuring officials, depriving a bidder of fair and honest consideration of its proposal, generally constitutes arbitrary and capricious action. Keco, 492 F.2d at 1203. Bad faith includes “predetermining the awardee or ... harboring a prejudice against the plaintiff.” Id. at 1204. Additionally, “courts should be extremely reluctant to overturn” agency actions validated by the GAO. Kinnett Dairies, 580 F.2d at 1272. However; “uncritical” deference to the GAO’s decisions is not proper, because it “would, in effect, repeal Congress’ grant of jurisdiction via the APA.” Shoals American Indus., Inc. v. United States, 877 F.2d 883, 889 (11th Cir.1989). “[T]here certainly may be instances where the District Court will find procurement illegality that the GAO failed to recognize, or at any event failed to correct.” M. Steinthal, 455 F.2d at 1305; cf. Delta Data Sys. v. Webster, 744 F.2d 197, 201-02 (D.C.Cir.1984) (stating that, although the GAO may have contracting expertise, courts have no obligation to defer to its conclusions); Aero Corp. v. Dep’t of the Navy, 540 F.Supp. 180, 205-06 (D.D.C.1982) (stating that courts have a duty independently to judge the legality of procurement decisions). B. SUMMARY JUDGMENT In ruling on the Navy’s motion to affirm or for judgment on the pleadings, the district court considered depositions and documents outside the pleadings. Fed.R.Civ.P. 12(c) provides, in pertinent part: If, on a motion for judgment on the pleadings, matters outside the pleadings are presented to and not excluded by the court, the motion shall be treated as one for summary judgment and disposed of as provided in Rule 56, and all parties shall be given reasonable opportunity to present all material made pertinent to such a motion by Rule 56. Thus, the district court was required by Rule 12(e) to treat the Navy’s motion as one for summary judgment. We review its decision as if it did. In considering a motion for summary judgment, a court must “review the evidence and ah factual inferences therefrom in a light most favorable to the party opposing the motion,” Thrasher v. State Farm, Fire & Casualty Co., 734 F.2d 637, 638 (11th Cir.1984), and grant summary judgment “only if no genuine issues of material fact exist” and the moving party is entitled to judgment as a matter of law, Miranda v. B & B Cash Grocery Store, Inc., 976 F.2d 1518, 1532, 1534 (11th Cir.1992). This Court reviews de novo the grant of a summary judgment motion. Chapman v. Klemick, 3 F.3d 1508, 1509 (11th Cir.1993). Thus, if Latecoere has presented evidence demonstrating the existence of genuine issues of material fact over whether the Training Center’s award to ETC lacked a reasonable basis, was done in bad faith, or involved a clear and prejudicial violation of procurement regulations, we must reverse and remand. Our review of the Navy’s decision to award the contract to ETC will not be limited to the formal record before the various Navy officials at the time the decision was made, because Latecoere has made “a strong showing of bad faith or improper behavior,” Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 420, 91 S.Ct. 814, 825-26, 28 L.Ed.2d 136 (1971). Likewise, while our standard of review is deferential, we will not abdicate our duty to review. IV. ANALYSIS A. THE IRRATIONAL ACTS AND ACTS IN VIOLATION OF PROCUREMENT REGULATIONS Latecoere contends that the record before the district court demonstrates that the G-TIP procurement process was corrupted by three irrational acts that were clear and prejudicial violations of procurement regulations and, accordingly, that the award to ETC should be set aside. Those three acts are: (1) the Advisory Council’s increase of ETC’s ratings in four critical chapters of the procurement proposals from “marginal” to “acceptable” without re-evaluation of those chapters; (2) Selection Authority Robert Urban’s use of cost as the most important factor in the selection, contrary to the terms of the Solicitation; and (3) Selection Authority .Urban’s failure to provide adequate support in his selection document for his determination that Latecoere’s superior technical features were not worth the minimally higher cost of Latecoere’s proposal. Latecoere maintains that based on its evidence of ratings manipulation and other improper behavior , in the procurement process, the district court should not have granted judgment against it, at least not without an evidentiary hearing. 1. The Advisory Council’s Increase of ETC’s “Marginal” Ratings Latecoere asserts that the evidence establishes that the Advisory Council acted irrationally and in violation of procurement regulations when, during its review of the Evaluations Board’s evaluations and ratings of the offerors’ revised proposals, the Advisory Council increased ETC’s ratings in four critical chapters from “marginal” to “acceptable” without re-evaluating those chapters. Latec-oere points to 48 C.F.R. § 15.608(a) (1993), which provides that “[pjroposal evaluation is an assessment of both the proposal and the offeror’s ability (as conveyed by the proposal) to successfully accomplish the prospective contract.” Latecoere contends that, by increasing ETC’s chapter ratings without reevaluating the chapters, the Advisory Council improperly based its evaluation of ETC’s proposal on factors other than those conveyed in its proposal. Latecoere maintains that the elimination of ETC’s “marginal” ratings prejudiced Latecoere by making ETC eligible for the award when it otherwise would not have been. The Navy does not deny that the Advisory Council increased ETC’s ratings. Rather, it contends that the Advisory Council’s act was not irrational or in clear violation of § 15.-608(a). Alternatively, it argues that, even if increasing the ratings was irrational and clearly violative of § 15.608(a), it was not prejudicial to Latecoere. After considering Latecoere’s allegations and the evidence regarding the ratings increases, the GAO concluded: While [increasing ETC’s ratings] is indicative that the evaluation of these [chapters] was unreasonable, the record also shows that ETC’s overall technical rating for the device design factor was acceptable with low risk, even with the four critical ... marginal ratings. Latecoere does not contend that ETC’s overall acceptable rating for the device design factor was unreasonable. Since the [Selection Authority] was aware of the marginal ratings for these [chapters] and concurred with the [Advisory Council’s] determination to change them to acceptable, it follows that he weighed them in making his selection decision. Under the circumstances, we are not persuaded that Latecoere was prejudiced by this change in the scoring of ETC’s [chapters], inasmuch as it apparently had no effect upon ETC’s unchallenged overall technical rating of acceptable] with low risk for the device design evaluation factor. Latecoere, 1992 WL 15029 at *6. We agree with the GAO that the Advisory Council’s increase of ETC’s ratings from “marginal” to “acceptable” without re-evaluation of those chapters was unreasonable. There is no other way to describe an action taken in order to make an award to ETC to which it was not otherwise entitled on the merits. If fairly evaluated, ETC’s proposal would not have been selected, so the books were cooked to ensure that it was. We disagree with the GAO’s conclusion that the cheating on behalf of ETC did not prejudice Latecoere. The GAO based its conclusion that Latecoere was not prejudiced by the arbitrary elevation of ETC’s ratings on two assumptions, one of which has no support in the record, and the other of which is flatly contradicted by the record. The GAO’s first assumption was that Selection Authority Robert Urban was aware that ETC had had “marginal” ratings in critical chapters and must have taken those “marginal” ratings into consideration when making his selection. That assumption is not supported by the record for two reasons. First, the record contains no evidence that Urban was aware when he selected ETC that the Advisory Council had changed ETC’s “marginal” ratings in critical chapters. Second, even if Urban had known that the Advisory Council had arbitrarily upgraded ETC’s “marginal” ratings in critical chapters, the record provides no reason at all to assume that Urban would have considered that fact against ETC. ETC’s “marginal” ratings in critical chapters were changed for arbitrary and political reasons, and there is evidence that Urban’s own decisional process was driven by that same improper motivation. Urban’s overall course of conduct and his statements permit and perhaps compel the conclusion that, in keeping with the wishes of Assistant Director Ford, Urban was determined to favor ETC. Consequently, there is no indication that Urban considered the fact ETC actually should have had “marginal” ratings in critical chapters. The GAO based its conclusion that the arbitrary increase of ETC’s “marginal” ratings did not harm Latecoere on a second assumption as well: that this increase had no effect on ETC’s overall rating. The evidence contradicts that assumption. Captain Kala-pos, the Training Center’s Director of Contracts, told the Advisory Council members that there would be no award to an offeror with a marginal rating in a critical chapter of the Device Design volume of the proposals, even if the firm had an overall rating of “acceptable.” Thus, the evidence shows that if the Advisory Council had not cheated on ETC’s behalf, ETC would have had “marginal” ratings in four critical chapters and would not have been awarded the contract. The Advisory Council’s action prejudiced Latec-oere, and the GAO’s contrary conclusion is clearly wrong. The only statement by the magistrate judge pertaining to this issue is that: [T]he record does not currently ... support the conclusion that [ratings] were manipulated only to accept a lower cost proposal from a domestic corporation. Rather the record establishes only that ETC’s proposal, previously classified as technically unacceptable, was reclassified as technically acceptable due to the reclassification of certain “critical” chapters. Because the magistrate judge’s finding ignores evidence that the Advisory Council increased ETC’s ratings in critical chapters without re-evaluating those chapters for no reason other than to award the contract to ETC, we reject the magistrate judge’s finding on this issue. The record contains overwhelming evidence that the Advisory Council’s manipulation of ETC’s ratings was arbitrary, irrational, improperly motivated, and a clear and prejudicial violation of 48 C.F.R. § 15.608(a). That evidence alone requires that we reverse the district court’s grant of summary judgment against Latecoere. 2. Urban’s Use of Cost as the Most Important Factor Latecoere also argues that the record demonstrates that Urban used cost as the most important factor in his selection, in violation of 41 U.S.C. § 253b(a) (1987), which provides: “An executive agency shall evaluate sealed bids and competitive proposals based solely on the factors specified in the solicitation.” See also In re Dewberry & Davis, 92-1 CPD ¶ 421, 1992 WL 103406 at *3 (C.G.1992) (“[Agencies] do not have the discretion to announce in the solicitation that they will use one evaluation plan, and then follow another; once offerors are informed of the criteria against which their proposals will be evaluated, the agency must adhere to those criteria ... or inform all offerors of any significant changes made in the evaluation scheme.”) The G-TIP Solicitation designated itself as a best value procurement in which cost would be the least important factor in determining the award. The Solicitat