Post-award bid protest; motion for preliminary injunction; likelihood of success; discussions; FAR § 15.306; equal treatment; irreparable injury.
OPINION AND ORDER1
The Army held a competition in three phases for a contract to design, develop pro*712totypes of, and ultimately manufacture and supply a new armored but agile all-terrain tactical vehicle to be used in combat zones. Oshkosh Defense, LLC (“Oshkosh”) won the last phase of the competition in August 2015, and it was awarded the manufacturing contract. Lockheed Martin Corporation (“Lockheed”), a losing offeror, has filed suit seeking a court order that would require the Army to reopen the competition and give Lockheed another chance. Lockheed generally alleges that the Army violated procurement law by failing to credit Lockheed’s data relating to performance of its proffered prototype, while unfairly crediting Oshkosh’s data relating to its prototype under similar circumstances. The Army disputes these allegations and avers that the performance of the vehicle Lockheed proposed was not as desirable as that proposed by Oshkosh.
It may take time to resolve this case, especially given that the administrative record reportedly is over 300,000 pages in length and numerous fact-intensive issues have been raised. Because of that prospect, Lockheed has requested a preliminary injunction in the form of a stop-work order, which would bar Oshkosh from doing work while this case is pending. This opinion addresses Lockheed’s request. Unfortunately, the court must decide the pending motion without having in hand all of the evidentiary materials in the Army’s procurement record. Instead, for the purpose of resolving Lockheed’s motion, the parties have submitted excerpts of the records of the procurement, along with some materials from aborted protest proceedings that occurred before the Government Accountability Office (“GAO”).
After reviewing the available materials, the court has decided to deny Lockheed’s motion for a preliminary injunction. The court concludes that Lockheed has not made a sufficiently strong showing that it is likely to prevail on the merits of its protest or that it will be irreparably harmed if Oshkosh conducts the initial work due under the awarded contract.
FACTS2
A. The Army’s Three-Phase Plan to Develop a New Tactical Combat Vehicle
In 2007, the Army and Marine Corps began planning a program to develop a Joint Light Tactical Vehicle (“JLTV’). PI-1155.3 The JLTV would be light, mobile, and capable of airlift, while still providing strong armor and heavy payload capacity. PI-1558. This vehicle would fill a “critical capability gap” in the military’s combat transport assets. PI-1558, Currently, the military has heavily armored vehicles that are difficult to maneuver off-road, and it has light vehicles that lack strong armor and thus fail adequately to protect soldiers and marines from explosives. PI-1558. The JLTV would fill this gap, being fast, relatively light, and well armored. PI-1558.
The JLTV program has three phases, each of which has been covered by a separate *713contract. PI-1155. In Phase One, three companies won contracts to design the JLTV. Phase One lasted from 2007 until 2012. PI-1155. Lockheed and two other companies won this design contract. PI-1155. Phase Two began in 2012 and lasted until 2014. PI-1155. The Army awarded the Phase Two contracts to Lockheed, Oshkosh, and AM General, LLC. PI-1155. In Phase Two, the three companies fabricated 22 JLTV prototypes. These prototype vehicles then were extensively field-tested by the Army, being driven roughly 155,000 miles during those field tests. PI-1155. In the tests, the Army collected data to be used during the Phase Three competition. For Phase Three, only the Phase Two contractors were eligible to compete, and only one could win. Phase Three involves production and deployment of the JLTV, which would become operational by 2019. PI-1560.
In December 2014, the Army invited the three companies who had participated in Phase Two to bid for Phase Three, and each submitted bids. PI-1156. The Army selected Oshkosh for the final award on August 24, 2015. PI-1159. Lockheed filed protests of the Army’s decision with GAO in September 2015. PI-1159. Toward the end of the GAO proceeding, the Army located a substantial number of records of its evaluation of the competing offers which had not previously been included in the record of the procurement as provided to GAO. When GAO indi-eáted that it would proceed to render a decision without considering the newly provided materials, Lockheed filed suit in this court on December 16, 2015. The next day, Oshkosh intervened to defend the award. Mot. to Intervene, EOF No. 8. AM General did not intervene, and it is not a party to this case.
B. During Phase Two, the Army Field Tests the Designs of the Contractors
The Phase Two contract required the winners to manufacture 22 prototype JLTVs and turn them over to the government for testing. PI-1155. The contract provided performance benchmarks, which each contractor sought to achieve during testing. For example, the vehicle needed to start within [* * *] in temperatures ranging from [* * *] Fahrenheit, PI-357, accelerate from [* * *] miles per hour in under [• * *], PI-360, ford [* * *] water, PI-376, and provide air conditioning that could lower the crew compartment temperature from [* * *] Fahrenheit within [* * *], PI-447. Most importantly for this case, the government also evaluated the rate at which the vehicle’s functions became impaired or failed. This rate was determined by applying “failure scoring criteria,” which focused particularly on “mission essential function” failures. PI-503, 506 to - 07.4 A failure was generally defined as an event “in which an item or part of an item does not perform as specified.” PI-504. A failure could be classified as an essential function failure (“EFF”), an operational mission failure (“OMF”), or a hardware mission failure (“HMF”). PI-504.
“An EFF is an event that results in significant degradation of a [mission essential function], yet the vehicle is still partially mission capable.” PI-504. In contrast, “[a]n OMF is an event that results in the loss of a[m]ission [e]ssential [fjunetion ... that reduces its utility to the point that it is deemed ineffective in its role on the battlefield. OMF[s] are failures that cause immediate removal from service in an operational environment or that preclude a vehicle from starting a mission on time.” PI-504. Finally, a HMF “is a subset of OMFs that are chargeable to [the] contractor.” PI-504.
The Army’s field tests counted the mean miles driven between EFFs and the mean miles driven between HMFs. In particular, the vehicle needed to be able to demonstrate 3800 mean miles between hardware mission failures (“MMBHMF”). PI-427.
1. The contractors’ Phase Two demonstrated failure rates.
During the Phase Two contract, both Oshkosh’s and Lockheed’s prototypes failed to *714demonstrate 3800 MMBHMF. PI-212. Each contractor could make changes or “fixes” to its prototype vehicles during the field tests. Lockheed’s vehicles were tested three separate times, demonstrating [* * *] MMBHMF and showing [* * *] during the three field tests. PI-113. In contrast, Oshkosh’s vehicles demonstrated [* * *] MMBHMF and showed improvement over the tests. PI-128. Of relevance, however, during this testing, Oshkosh’s [* * *] experienced an EFF. PI-697.
2. Each contractor presents changes at an Army assessment conference.
After field testing, the government held an “assessment conference,” during which each contractor proposed or implemented “fixes” to their prototype vehicles. PI-115, -129. With these fixes, one could calculate an “assessed MMBHMF.” PI-115, -129.
The effect of a proposed fix was calculated by applying three “fix effectiveness factors” (“FEF”). PI-115. The factors were: (1) understanding the root cause of the failure; (2) certainty of improvement; and (3) test results to verify the effectiveness of the corrective action. PI-115; Compl. ¶33. Each of these fixes was rated from 0.0 to 0.9 in increments of 0.3 such that the maximum fix score was 0.9. PI-115; Compl. ¶ 33. This rating would then be applied against the number of failures, resulting in an “assessed reliability” score. PI-115; Compl. ¶ 33.
An assessment conference fix to a contractor’s design could be either proposed or implemented. A proposed change was just that — a proposal. But if a change had been actually implemented during Phase Two testing, then it would have been on the vehicle for any remaining portion of the field tests, allowing the change to receive potential credit under FEF (3) — test results. Hr’g Tr. 131:23-25, 132:1-4 (Jan. 20, 2016);5 see also Pl.’s Notice of Citation, ECF No. 35. For example, Oshkosh experienced [* * *] types of HMFs. Pl.’s Notice of Citation at 3-4. Of these, Oshkosh implemented changes to address [* * *] of these failures, allowing the changed vehicles to drive for tens of thousands of (and in some cases over one hundred thousand) additional testing miles after the change had been implemented. Id.6
After the assessment conference, the government’s “assessed MMBHMF” for Lockheed’s vehicle was [* * *] and that for Oshkosh was [* * *]. PI-115, -129.
C. The Army Issues a RFP for Phase Three, Establishing the Rules of the Competition
The competition for Phase Three began in December 2014 with the Army’s Request for Proposals (“RFP”). PI-1156. The offerors from Phase Two were each invited to bid, and each had an opportunity to improve upon their design to offer an even better vehicle than that tested in Phase Two. PI-1156 to - 57.
1. The RFP establishes a best-value procurement based on three factors.
Pursuant to the RFP, the government would award the contract to the company offering the “best value.” PI-101 (RFP § M.1.1). Value would be determined using a tradeoff process between three factors: Primary Technical; Total Evaluated Cost/ Price (“TEC/P” or “price”); and Small Business Participation. PI-101 (RFP § M.1.1). These three factors were not equally weighted. Instead, Primary Technical was the most important factor; price was second most important; and Small Business Participation was least important. PI-101. Primary Technical referred to the ability of the offeror to meet the government’s performance requirements for the vehicle. PI-104 (RFP § M.4.1). Thus, for example, the requirement that a vehicle demonstrate 3800 MMBHMF was a Primary Technical requirement.
In addition, the Primary Technical and Small Business Participation factors combined were “significantly more important” *715than the price factor. PI-101 (RFP § M.1.1). This meant that price was not dispositive, and the RFP stated that “th[e procurement] determination may result in award to other than the [o]fferor with the lowest TEC/P.” PI-102. Although the RFP established that price was not controlling, it also explained that a scenario existed in which price could be controlling, ie., “when proposals are otherwise considered approximately equal in non-TEC/P factors.” PI-102 (RFP § M.3.1(a)).
*714[* * *] were tested for [* * *] of the field testing’s total miles; and [* * *]% of changes were tested for [* * *]% or less of those miles. Pl.’s Notice of Citation at 1.
*7152. The Primary Technical factor explained.
Pursuant to the Primary Technical factor, the Army would “assess” the “risk that the [o]fferor[’]s JLTV will [not] achieve the [government's defined threshold performance levels for [25 different purchase descriptions listed in the RFP’s Attachment 0061]. Each P[urchase] Description] requirement identified in Attachment 0061 is weighted equally and will be evaluated as such.” PI-104 (RFP § 4.1). Attachment 0061 listed these 25 purchase descriptions by a “PDFOV” number. The 25 purchase descriptions called for a variety of performance requirements. For example, one required that the JLTV “be capable of ascending [* * *],” PI-359 (PDFOV-7478), while another required the JLTV to “be capable of entering and exiting [a variety of different warships].” PI-395 (PDFOV-8452). A risk assessment for a purchase description would be based on an adjectival rating scale of very low, low, moderate, high, and unacceptable risk. PI-1007 (providing definitions for each adjectival rating category).
Of the 25 purchase descriptions, two are relevant to this case. First, PDFOV-2909 provided that the “JLTV shall demonstrate at a minimum, a point estimate of 3,800 Mean Miles Between Hardware Mission Failure (MMBHMF).” PI-427. This PDFOV was termed “Reliability.” PI-427. Second, PDFOV-2918 provided that the “JLTV shall demonstrate the operational availability (Ao) of 95%.” PI-428. PDFOV-2918 was called “Operational Availability,” and was determined by multiplying MMBHMF by other stated factors. PI-428, Thus Operational Availability was linked to MMBHMF, meaning that an error in the MMBHMF calculation could cause an error in the Operational Availability calculation.
Pursuant to Section M.4.1 of the RFP, the government would assess the risk that an offeror would fail to meet these standards by analyzing Army testing data from Phase Two and data submitted by the offerors. PI-104. The RFP explained:'
Government test data which establishes conformance to the proposed design configuration represents the most credible form of substantiating data. Therefore, any substantiating data for a design configuration which meaningfully varies from the offered design configuration may be considered less credible. The greater the extent to which the [o]fferor[’]s proposed design configuration meaningfully varies from the originally tested configuration or testing conducted under different conditions, the greater the probability that the [government may discount the relevance of the test data as substantiating information.
PI-104 (emphasis added). In effect, this meant that the data from Phase Two, including the “assessed MMBHMF” scores, would be the “most credible form” of data. PI-104. Offerors could also submit non-government data, including data generated by the offer-ors themselves. However, non-governmental data would be considered “less credible.” PI-104. Meaningful variation was also not binary, but rather a matter of degree. Greater meaningful variation meant “greater probability” that the government would discount the Phase Two data. This created an incentive for offerors to propose designs that did not “meaningfully var[y]” from the Phase Two design, because they would otherwise need to provide sufficient data for the Army to conclude that the design would still meet the government’s purchase-description requirements.
In addition, RFP § 4.1.2 permitted offer-ors to propose performance that exceeded the minimum thresholds set by the pui’chase descriptions. PI-104. Performance beyond threshold level would be accepted if it was backed by credible substantiating data. If *716so, then the RFP provided that the Army “may” assess the offer at a “reduced risk” of meeting the purchase-description requirements. But if the proposed beyond-threshold performance was not backed by credible substantiating data, then pursuant to RFP § M.4.1.5 it could be disqualified from the competition. PI-104.
3. How MMBHMF affected the price calculation.
Price would be determined by a formula set out in RFP § 4.2. PI-104. One factor in this formula was “life cycle cost,” which predicted repair costs over the life of a vehicle. Included in “life cycle cost” was each offer- or’s “proposed” MMBHMF score under PDFOV-2909, since a vehicle with fewer failures would need fewer repairs. PI-106 (RFP § M.4.2.3); see also PI-524 (attachment to RFP showing the life cycle cost formula). Of crucial importance, the life cycle credit was not based on an offeror’s demonstrated or even assessed MMBHMF. Instead, it was based only on the MMBHMF “proposed” in the final offer. PI-106.
This created an incentive for offerors to propose a high MMBHMF. The higher the proposed score, the cheaper the proposal would look under the Army’s price evaluation. However, RFP § M.4.1.5 prevented offerors from proposing an unrealistically high MMBHMF. PI-104. As described above, if offerors could not support their MMBHMF claims with credible substantiating data, then the offer could be disqualified.
D. Lockheed and Oshkosh Make Initial Offers in Early 2015 1. Lockheed’s initial offer.
In its initial offer, Lockheed stepped up its MMBHMF from [* * *], assessed during Phase Two, to a proposed MMBHMF of [* * *]_ pi-525. Lockheed conceded that it had not yet met this performance standard, but said that it “has committed to achieving” MMBHMF based on future testing and “ongoing RGT design improvements” after award. PI-525 [* * *].7 Lockheed later explained that its [* * *] proposal was also based on [* * *]. PI-862. Lockheed expressly noted that it was proposing [* * *] in the RFP. PI-525.
Lockheed’s offer was premised upon inclusion of a variety of design changes, including changes to [* * *]. PI-114; see also PI-866 (acknowledging changes to many of these systems). Lockheed initially did not provide information relating the effect of these changes to its prototype tested during Phase Two, which could have shown the government the extent to which the design improved upon and varied from that tested in Phase Two. PI-114.
2. Oshkosh’s initial offer.
Oshkosh proposed a vehicle with [* * *] MMBHMF, which roughly matched Oshkosh’s assessed MMBHMF from Phase Two field testing. PI-129. Oshkosh also proposed [* * *] design changes, including changes to [* :i! *]. PI-129; see also PI-220, -589, -843 ([* * *]). Its submissions included substantiating data and information showing the government how Oshkosh’s proposed design' was revised from its Phase Two prototype design. PI-130.
E. The Army Corresponds with Lockheed and Holds Discussions
On April 16,2015, the Army sent notices to Lockheed raising questions about its proposal. PI-8543 to -56. First, the Army said Lockheed’s “proposal does not explain the extent of the variances and the impact of *717such variances between the As-Tested [Phase Two] design and the proposed design. The [o]fferor did not provide accurate traceability for the RAM [‘Reliability Availability Maintenance’]-related Corrective Actions, defining how the [Phase Two] tested configuration transitions to the proposed design configuration and the impact to RAM at the system level.” PI-854. Second, the Army indicated to Lockheed that it had not provided sufficient data to substantiate its proposal. In particular, “the offeror did not provide substantiating data to support the proposed [fix effectiveness factor] claims and threshold MMBHMF.” PI-856. Lockheed responded to the Army on April 20 with performance and testing data, but'it continued to rely on future growth and design changes. See PI-858. A few days later, on April 24, the Army sent Lockheed another notice saying that Lockheed’s submission of data on April 20 was deemed insufficient. PI-857 to -58. The Army advised that Lockheed’s “proposed configuration meaningfully varies from the as tested [Phase Two] configuration. This appreciably reduces the relevance [of] the Government Assessment Conference and the final assessed MMBHMF.” PI-858. The government identified four specific concerns:
“[Lockheed] ... [first] has not justified the relevance and validity of [* * *] used to support MMBHMF claims[, second,] has not identified apparent non[-]RAM related configuration ehanges[, third,] has not substantiated apparent non[-]RAM related configuration changes and associated impacts to reliability,” and fourth, Lockheed’s proposed MMBHMF of [* * *] “is considered an[] unsubstantiated technical commitment, per section M.4.1.5. As such ... [Lockheed’s] proposal is in jeopardy of being rejected.” PI-858.8
Lockheed promptly responded. PI-860. Lockheed noted that Phase Two and Phase Three used different scoring criteria for failures. PI-862; see also PI-894 (explaining Lockheed’s understanding that the scoring criteria had changed between Phase Two and Phase Three). Applying Phase Three standards to the data collected in Phase Two, Lockheed believed it had a “start point” MMBHMF of [* * *]. PI-864. With this number as a baseline, Lockheed predicted there would be “new modes that will surface during [Lockheed’s planned] [* * *].” PI-864 (emphasis added). Taking that [* * *] into account, Lockheed reiterated its prediction of an average FEF factor of [* * *]. PI-864. On the basis of these yet-to-be-achieved changes and predictions, Lockheed would contractually commit to [* * *] MMBHMF. PI-864.
Lockheed acknowledged that it had made [* * *] design changes, but averred that each had been analyzed for MMBHMF impact. PI-865. Based on a[!|! * *] Lockheed estimated that the [* * *]. PI-865. This [* * *]. PI-865 to -66 (listing reasons). Lockheed acknowledged making changes regarding [* * *]. PI-866. Lockheed’s response explained that it would continue to make “design improvements configuration changes,” and that based on future design changes and testing, it would achieve [* * *] MMBHMF. PI-866.
The Army sent Lockheed a further notice on May 7, 2015 saying that the company still had not applied the FEF and reliability standards correctly, and that the proposed [* * *] MMBHMF was still unsubstantiated and consequently at risk of disqualification. PI-890 to -91. Lockheed responded with additional data, restating its position that based on the scoring criteria in Phase Three (which differed from Phase Two) and based on [* * *], Lockheed estimated a “start-point” MMBHMF of [* * *]. PI-895. It stood by its proposal to grow MMBHMF in the future.
Army evaluators held an internal meeting on May 19, 2015 at which they discussed Lockheed’s proposal. PI-184. Notes from the meeting show that the evaluators were concerned that (1) Lockheed had not “substantiate[d] the design changes” and was relying on Phase Two data even though the design meaningfully varied; (2) Lockheed proposed to make future design changes; and (8) that Lockheed’s [* * *] MMBHMF was unsubstantiated. PI-184. The evalu*718ators also said “[u]p to this point, the main message has been a lack of substantiating data, leading L[ockheed] to provid[e] more data. Need to address the fact that a number lower than the proposed reliability can be substantiated. L[ockheed] is providing well above the reliability performance that can be substantiated.” PI-184.
The next day, May 20, 2015, representatives of Lockheed and the government had a detailed oral discussion, PI-170 to -82. The Army began by advising Lockheed of problems with its proposal: “I’d like to state upfront that the [g]ovemment still assesses threshold performance for PDFOV-2909 as appreciably increased risk and the proposed value as an unsubstantiated technical commitment.” PI-171. The government further told Lockheed that because of “the complexity of the components and subsystems that were addressed with corrective actions, combined with the quantity of design changes impacting [mission essential function] and ... related components, we believe that the design meaningfully varies from the [Phase Two] as tested configuration.” PI-171 (emphasis added). The government added that it “does not view the [* * *] [MMBHMF assessed in Phase Two] as still[ ] being relevant.” PI-173.
Although the Army did not give Lockheed a list of every change that needed supporting data, or which meaningfully varied, the Army provided an example of one such change: “For example the [* * *] ... experienced multiple issues during [Phase Two] Reliability Testing [but received a FEF of [* * *] during Phase Two based on proposed solutions]____ Within the proposed design, the [* * *] was changed from that which was presented within [Phase Two].... Therefore, the system meaningfully varies and the achieved [Phase Two] [!|i * *] FEF is invalidated[.] Data identifying & substantiating the projected FEF was not provided; this should come in the form of the three components that make up a FEF — Root Cause, Certainty, and Test Results.” PI-171 to -72.
Later, Lockheed representatives asked, “Do[ ] the RFP and subsequent contract include or contemplate reliability growth prior to starting [post-award] Government Reliability Qualification Testing ([post-award] RQT)?” PI-174. Although the RFP required substantiating data, PI-104, Lockheed apparently wanted to rely on predicted future growth. The government explained that “[a] growth curve does not substantiate the performance of a particular design.” PI-175.
Lockheed then asked, “[I]f the government does not recognize reliability growth prior to Govemment.RQT, what is the government’s basis for providing offerors such large TEC/P incentives?” PI-175. This question appears to address the life cycle credit, by which a high MMBHMF number could reduce the government’s evaluated price. The government’s representative reiterated that Lockheed could not rely on future growth or design changes:
As we understand it, L[oekheed]’s RGT significantly influences the proposed value [* * *]. Through RGT, the design will evolve and may derive a configuration that varies from the proposed design. In essence, L[ockheed] is asking the [government to accept and evaluate a design that is different from what is being proposed. Nowhere within the RFP is it stated that we will evaluate a design other than what is explicitly being proposed. It doesn’t appear that RGT concludes prior to the [government taking ownership of the design, [so] it is unclear how L[oekheed] is planning to implement RGT corrective actions to accomplish [the] proposed reliability claim.
PI-175 to -76.
Toward the end of this discussion, Lockheed asked about whether it needed to provide more substantiating data. One Lockheed representative asked, “Did we provide enough substantiating data for the [* * *] [post-Phase Two] changes?” PI-177. To this, the Army responded that “the [government is still evaluating data to verify the [government's position on the [* * *] [post-Phase Two] changes.” PI-177. Another Lockheed representative then said “[Lockheed’s] understanding is that the [government wants data for every change.” PI-177. In response, the government said: “Yes, think of every change is a change is a change.” PI-177.
*719After these discussions on May 20, Lockheed submitted a revised response to the government’s notice dated May 7. PI-896 to -913. In this revised response, Lockheed [* * *] its proposed MMBHMF to [* * *]. PI-899. Lockheed explained that its decision “is largely based on the [g]ovemment’s decision to [* * *].” PI-899. Lockheed acknowledged that its previous proposals had “forecast[ ] design improvements and/or changes” and that the government wanted an “evaluation of only the proposed [Phase Three] design at the time of proposal evaluation.” PI-900 (emphasis in original). Lockheed also explained that it had re-analyzed the FEFs pursuant to the government’s instructions. PI-900.
F. The Army Evaluates Lockheed’s and Oshkosh’s Final Offers
1. Lockheed.
A source selection evaluator described the history of communications with Lockheed, observing that Lockheed originally failed to explain “the extent of the variances and the impact .of such variances between the [Phase Two] as-tested configuration and the proposed design are still unknown.” PI-114 to -15. Later submissions from Lockheed traced the changes, which the Army then understood. PI-116. The evaluator concluded Lockheed’s design meaningfully varied, based on the “complexity of the components and subsystems that were addressed ... combined with the quantity of design changes.” PI-114. This included changes to [* * *]. PI-114. The evaluator also noted that certain changes “have the potential to cause a Mission Essential Function (MEF) failure and therefore meaningfully vary from [Phase Two] test configuration.” PI-114.
Because of the meaningful variation, the MMBHMF assessed in Phase Two was “not relevant,” and the evaluator needed to look at non-government data. PI-115 (“This risk can only be addressed with substantiating data.”). Lockheed provided data and engineering narratives to support its proposal, but these offeror-süpplied data were deemed “less credible.” PI-117. Nonetheless, Lockheed’s data were considered sufficient to substantiate its [* * *] MMBHMF [* * *] threshold of 3800, meaning its proposed MMBHMF [* * *] was not an unsubstantiated technical commitment. PI-118. But because this [* * *] performance was not “well-supported,” it did not reduce Lockheed’s overall risk assessment. PI-118. The evaluator accordingly rated Lockheed a “high” risk of not meeting PDFOV-2909’s 3800 MMBHMF figure. PI-118. Likewise, the evaluator concluded Lockheed was at a “moderate” risk of meeting PDFOV-2918’s Operational Availability requirements, in part because the proposed MMBHMF [* * *] and because Lockheed had failed to fully substantiate its changes. PI-125 to - 26.
2. Oshkosh.
The evaluator found that Oshkosh’s design did not meaningfully vary from Phase Two. PI-129. The evaluator said that Oshkosh’s final design included [* * *] corrective actions that were implemented during Phase Two testing and “remain constant within the proposed design.” PI-129. [* * *] other corrective actions had not been implemented during Phase Two but were deemed to “minimally vary,” although they did “reduce the relevance” of Phase Two data. PI-129.
Next, the evaluator said that Oshkosh had made [* * *] other “design changes,” but that these were “not related to [Phase Two] HMFs.” PI-129. Of these, [* * *] “address manufacturing efficiencies and supplier changes. As these changes are the same form, fit and function; these changes do not meaningfully vary from the [Phase Two] as-tested configuration ... [and therefore] have no impact on the relevance of [Phase Two government] test data.” PI-129. The remaining [* * *] changes “address requirement changes and reliability enhancements. Some changes are paper only, an accounting exercise, to address B[ill of] Materials] issues. The remaining changes were reviewed for impact to Mission Essential Functions (MEF) and were determined not to meaningfully vary from the [Phase Two] as-tested configuration. Therefore, relevance of [Phase Two] test data is not impacted.” PI-129.
Although Oshkosh’s final design did vary somewhat from its Phase Two design, “the *720impact of such variances” were “understood” by the evaluator. PI-129.. The changes were deemed to have a “slight impact” on the “relevance” of government data from Phase Two, which correspondingly only “slightly increase[d]” the risk assessment. PI-129. Oshkosh provided other data, including engineering analyses, which the evaluator considered “less credible” but which still substantiated Oshkosh’s claims. PI-130. In light of these data, and the fact that government data from Phase Two remained relevant, the evaluator rated Oshkosh a “low” risk of failing to meet PDFOV-2909, Reliability. PI-131. For similar reasons, Oshkosh was rated a “very low” risk of failing to meet PDFOV-2918, Operational Availability. PI-135.
G. The Army Selects Oshkosh for Award
After reviewing the evaluator’s materials, the Source Selection Authority (“SSA”), David G. Bassett, drafted a Source Selection Document that selected Oshkosh for award and explained his reasoning. PI-1004 to -14. Beginning with the Primary Technical factor’s 25 purchase descriptions, the SSA noted that Oshkosh was rated a “low” risk in thirteen categories and a “very low” risk in twelve categories. PI-1011. This summary included Oshkosh’s rating of “low” for PDFOV-2909 and ‘Very low” for PDFOV-2918. PI-1010. Given these results, the SSA agreed with prior evaluators that Oshkosh’s overall Primary Technical risk rating was “low.” PI-1011.
Lockheed contrastingly had received one “high,” five “moderate,” twelve “low,” and seven ‘Very low” risk ratings. PI-1010. Lockheed’s “high” rating was for PDFOV-2909, Reliability. On this point, however, the SSA disagreed with the prior evaluators. PI-1011. The SSA “gave greater technical credibility to the likely effectiveness of the design changes proposed” based on his “reviewing the specific design changes and their description of the substantiating data with the [evaluation] team.” PI-1012. Thus the SSA gave Lockheed a “moderate” rating for PDFOV-2909, bringing Lockheed’s score to six “moderate,” twelve “low,” and seven “very low.” PI-1012. Overall, the SSA assigned Lockheed a “moderate” Primary Technical risk rating. PI-1012.
On the price factor, the SSA noted that Lockheed’s evaluated contract price was $[* * *] billion. Lockheed had received a life cycle cost credit of $[* * *];' a “TDP” (i.e., Technical Data Package) Adjustment of $[* * *];9 a “Tier 1 Adjustment” of $[* * *]; and a “Secondary Technical Adjustment” of $[* * *]. PI-1012. This brought Lockheed’s TEC/P to $3,087 billion. PI-1013.
Oshkosh’s evaluated contract price was $5,587 billion. PI-1012. It received a life cycle cost credit of $1,527 billion; a TDP Adjustment of $511 million; a “Tier 1 Adjustment” of $229 million; and a “Secondary Technical Adjustment” of $622 million. PI-1012. This brought Oshkosh’s TEC/P to $2,689 billion. PI-1013. Thus Oshkosh’s TEC/P was roughly $398 million less than Lockheed.
Overall, the SSA found Oshkosh offered the “best value.” PI-1014.10 Regarding Primary Technical, the SSA noted that Lockheed received a “moderate” overall risk score, while Oshkosh received a “low” risk score. PI-1014. Although the individual ratings appeared to be similar, the SSA wrote that “the proposals were not as close as this might imply as [Oshkosh] is assessed as Low Risk or Very Low Risk for all 25 requirements whereas ... [Lockheed has] a number of Moderate Risk ratings[.]” PI-1014. “The [Oshkosh] proposal met or exceeded requirements, and was supported by the most credible substantiating data supporting a low risk Primary Technical rating.” PI-1014. Although Lockheed had the best small business participation, the SSA noted that this is “the least important factor” under the RFP, and that it “is significantly offset by the [Oshkosh] proposal’s superiority in the two most important factors of Primary Technical and TEC/P.” PI-1014.
*721H. Lockheed Protests at GAO and This Court
The government debriefed Lockheed on September 2, 2015, and six days later Lockheed filed a protest with GAO, triggering an automatic stay of Oshkosh’s performance of the contract. PI-1159. On November 18, 2015, GAO held a hearing and took testimony from government officers, who also submitted declarations. Compl. ¶ 15, According to Lockheed, on October 9 and 14, 2015, the government had told Lockheed that no evaluator notes existed. Compl. ¶¶ 107, 108. Weeks later, however, the government notified GAO that it had located more documents, and on December 4, 2015, GAO directed the government to disclose them. Compl. ¶¶ 112, 113. GAO nonetheless informed the parties that it would not consider these new documents in rendering its decision, which was statutorily required to be issued within a few days after GAO’s order to the government to disclose the newly identified records relating to the procurement. Compl. ¶ 114. Because GAO would not consider the docum'ents in rendering its decision, Lockheed filed a protest in this court on December 16, 2015, and GAO dismissed the protest before it. Compl. ¶ 11. At that point, the automatic stay of performance of the Army’s award to Oshkosh terminated.
In this court, Lockheed filed a motion for preliminary injunction. See Pl.’s Mot. for a Preliminary Injunction (“Pl.’s Mot.”), EOF No. 10. At an initial status conference on December 18,2015, the government informed the court that it could not file the administrative record until February 2016. See Order of December 18, 2015, EOF No. 13. Lockheed opted to proceed with the motion for a preliminary injunction despite not having a complete record of the procurement. The briefing by the parties regarding this motion has drawn upon selected portions of the record. The parties have submitted excexpts from the solicitation, the parties’ offers, the parties’ responses to evaluation notices, government correspondence (including notes and evaluations), declarations provided by government officials to GAO, and transcripts of testimony adduced at the GAO hearing.
JURISDICTION
The Court of Federal Claims has jurisdiction to “render judgment on an action by an interested party objecting to a solicitation by a Federal agency for bids or proposals for a proposed contract or to a proposed award or the award of a contract or any alleged violation of statute or regulation in connection with a procurement or a proposed procurement.” 28 U.S.C. § 1491(b)(1), added by the Administrative Dispute Resolution Act, Pub.L. No. 104-320, § 12, 110 Stat. 3870, 3874 (Oct. 19, 1996); see also Systems Application & Techs., Inc. v. United States, 691 F.3d 1374, 1380-81 (Fed.Cir.2012). Accordingly, the court has jurisdiction over this post-award bid protest.
STANDARDS FOR DECISION
A. Preliminary Injunction
To obtain the extraordinary relief of a preliminary injunction, the court considers four factors: (1) likelihood of success on the merits; (2) irreparable harm to the protestor if an injunction is not granted; (3) whether the balance of hardships tips in the movant’s-favor; and (4) the public interest. FMC Corp. v. United States, 3 F.3d 424, 427 (Fed.Cir.1993). Although no one factor is dispositive, a protestor must establish the first two factors, likelihood of success on the merits and irreparable harm, before a preliminary injunction can be granted. See Per Aarsleff A/S v. United States, 123 Fed.Cl. 147, 156-57 (2015) (citing Altana Pharma AG v. Teva Pharm. USA Inc., 566 F.3d 999, 1005 (Fed.Cir.2009)).
B. Standard for Review of a Procurement Action
The Administrative Procedure Act (“APA”), specifically 5 U.S.C. § 706, governs the court’s review of an agency’s contract award. 28 U.S.C. § 1491(b)(4) (“In any action under this subsection, the courts shall review the agency’s decision pursuant to the standards set forth in section 706 of title 5.”). “Section 706 of the APA provides, in relevant part, that a ‘reviewing court shall ... hold unlawful and set aside agency action, findings, and conclusions found to be ... arbi*722trary, capricious, an abuse of discretion, or otherwise not in accordance with law.’” Centech Grp., Inc. v. United States, 564 F.3d 1029, 1037 (Fed.Cir.2009) (alterations in original). Accordingly, the court may set aside a procurement action if “(1) the procurement official’s decision lacked a rational basis; or (2) the procurement procedure involved a violation of regulation or procedure.” Impresa Construzioni Geom. Domenico Garufi v. United States, 238 F.3d 1324, 1332 (Fed.Cir.2001).
When challenging the rationality of an award under the first Impresa ground, “the disappointed bidder bears a heavy burden.” Centech, 554 F.3d at 1037 (quoting Impresa, 238 F.3d at 1332-33). As a practical matter, this burden can be particularly difficult to meet when challenging a negotiated procurement for best value because such procurements inherently entail exercise of a contracting officer’s discretion. See Galen Med. Assocs., Inc. v. United States, 369 F.3d 1324, 1330 (Fed.Cir.2004). Thus, de minimis errors in the procurement process do not justify relief. Grumman Data Sys. Corp. v. Dalton, 88 F.3d 990, 1000 (Fed.Cir.1996). Notably, “[c]ourts have found an agency’s decision to be arbitrary and capricious when the agency ‘entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or [the decision] is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.’ ” Alabama Aircraft Indus., Inc.—Birmingham v. United States, 586 F.3d 1372, 1375 (Fed.Cir.2009) (quoting Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983)).
In particular, an agency’s procurement action is arbitrary and capricious if it treats offerors unevenly. See PGBA, LLC v. United States, 60 Fed.Cl. 196, 207 (2004), aff'd, 389 F.3d 1219 (Fed.Cir.2004); see also CliniComp Int'l, Inc. v. United States, 117 Fed.Cl. 722, 741 (2014) (collecting cases). The agency “must ... evaluate] proposals evenhandedly against common requirements,” CW Gov’t Travel, Inc. v. United States, 110 Fed.Cl. 462, 490 (2013), and must “consistently apply the evaluation factors listed in the solicitation,” TLT Constr. Corp. v. United States, 50 Fed.Cl. 212, 216 (2001). However, evenhanded and fair treatment does not necessarily mean indistinguishable consideration. See FAR § 1.102-2(c)(3) (“All contractors and prospective contractors shall be treated fairly and impartially but need not be treated the same.”). To the extent proposals differ, the government may rationally take different action regarding different proposals.
Turning to the second Impresa ground, “the disappointed bidder must show ‘a clear and prejudicial violation of applicable statutes or regulations.’” Cyios Corp. v. United States, 122 Fed.Cl. 726, 737 (2015) (quoting Impresa, 238 F.3d at 1332-33).
Under either Impresa ground, the court should review an agency’s procurement record existing at the time of decision. Axiom Res. Mgmt., Inc. v. United States, 564 F.3d 1374, 1379-80 (Fed.Cir.2009) (citing Camp v. Pitts, 411 U.S. 138, 142, 93 S.Ct. 1241, 36 L.Ed.2d 106 (1973)). Because the standard of review under the APA requires the court to analyze the agency’s reasoning in light of the facts and circumstances before it, supplementation of the record with new evidence could frustrate this standard. Id. at 1380. Accordingly, “supplementation of the record should be limited to cases in which ‘the omission of extra-record evidence precludes effective judicial review.’” Id. (quoting Murakami v. United States, 46 Fed.Cl. 731, 735 (2000), aff'd, 398 F.3d 1342 (Fed.Cir.2005)). Among other things, supplementation may be appropriate in instances where there is a gap in the administrative record. Midwest Tube Fabricators, Inc. v. United States, 104 Fed.Cl. 568, 573 (2012). Also, the Federal Circuit’s recent decision in Glenn Defense Marine (Asia), PTE Ltd. v. United States, 720 F.3d 901, 911 n. 8 (Fed.Cir.2013), elaborates on required elements of the record. In Glenn Defense, the Federal Circuit observed that pursuant to 31 U.S.C. § 3556, “all of the materials submitted to GAO are part of the administrative record before the Court of Federal Claims.” Id. Accordingly, materials generated in proceedings at GAO *723must be submitted to, and are part of the record in this court, but any “post-hoe rationalizations” in those materials are critically and skeptically evaluated. See Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 420-21, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971), abrogated in part by Califano v. Sanders, 430 U.S. 99, 105, 97 S.Ct. 980, 51 L.Ed.2d 192 (1977); Co-Steel Raritan, Inc. v. International Trade Comm’n, 357 F.3d 1294, 1314 (Fed.Cir.2004); PGBA, 60 Fed.Cl. at 204.
ANALYSIS
A. Likelihood of Success on the Merits
Lockheed makes two principal arguments that the Army.erred in selecting Oshkosh. First, it asserts the Army misled Lockheed during the discussions held on May 20, 2015. Hr’g Tr. at 139:7-8 (“So the problematical misleading discussion is on May 20th.”); Pl.’s Mot. at 2426. Second, Lockheed argues that the Army treated Lockheed worse than Oshkosh for no rational reason, in particular by refusing to credit certain Lockheed data while crediting Oshkosh data under similar circumstances. Pl.’s Mot. at 22, 26-37.
1. Meaningful discussions.
FAR § 15.306(d) governs “discussions” between the agency and offerors. Discussions “are tailored to each offeror’s proposal” and “are undertaken with the intent of allowing the offeror to revise its proposal” so the government may “obtain the best value.” FAR § 15.306(d)(1), (2); Pursuant to this regulation, contracting officers “must, subject to paragraphs (d)(5) and (e) of this section ... indicate to, or discuss with, each offeror still being considered for award, deficiencies, significant weaknesses, and adverse past performance information to which the. offeror has not yet had an opportunity to respond.” FAR § 15.306(d)(3). In doing so, the contracting officer must not “engage in conduct” that “favors one offeror over another.” FAR § 15.306(e)(1). Beyond that, the FAR imposes few other constraints on discussions. Instead, “[t]he scope and extent of discussions are a matter of contracting officer judgment,” and “the contracting officer is not required to discuss every area where the proposal could be improved.” FAR § 15.306(d)(3). .
Restated, discussions must be “meaningful” and not “misleading.” See, e.g., CEdge Software Consultants, LLC v. United States, 117 Fed.Cl. 419, 434-35 (2014). “For discussions to be meaningful, they must ‘generally lead offerors into the areas of their proposals requiring amplification or correction, which means that discussions should be as specific as practical considerations permit.’ ” Id. at 434 (quoting Advanced Data Concepts, Inc. v. United States, 43 Fed.Cl. 410, 422 (1999), aff'd, 216 F.3d 1054 (Fed.Cir.2000)). Misleading discussions occur when the agency provides “materially disparate information to bidders on matters that could easily affect their decisions about important aspects of the final competing offers that the agency will be comparing.” Raytheon Co. v. United States, 809 F.3d 590, 596 (Fed.Cir.2015).
Applying these principles here, by May 20, 2015, the government believed Lockheed’s proposal had several “deficiencies” or “weaknesses” within the meaning of FAR § 15.306(d)(3).11 Those were that Lockheed’s [* * *] MMBHMF proposal was unsubstantiated, its design meaningfully varied, and it had failed to submit enough data to satisfy the Army’s evaluators. The government had communicated these problems to Lockheed through a series of notices in April and early May 2015. PI-853 to -866. And during the May 20 discussion, the government reiterated those concerns. The Army explained that it “assesses threshold performance for PDFOV-2909 as appreciably increased risk and the proposed value as an unsubstantiated technical commitment.” PI-171. The government then told Lockheed that it considered Lockheed’s design to meaningfully vary from that tested in Phase Two because of the “complexity” of systems changes “combined with the quantity of design changes.” PI-171. The government asked Lockheed to explain the changes and provide data for them. PI-171. Although the *724government did not give Lockheed a list of changes requiring substantiating data, the government did give Lockheed an example of such a change: [* * *]. PI-171. Thus the government “generally le[d] [Lockheed] into the areas of [its] proposal requiring amplification or correction.” See CEdge, 117 Fed.Cl. at 434.
Lockheed makes four arguments that the discussions were unsound. First, Lockheed contends that the government misled it by saying “a change is a change is a change,” which implied that every design change needed supporting data. Hr’g Tr. 67:19-24, 68:16-22 (arguing that because Lockheed did not know what changes mattered, the company “ha[d] no choice but to substantiate every change”). But that contention overstates the Army’s commentary. The government’s statement that a “change is a change” came at the end of a lengthy conversation, in which the government indicated its reluctance to fully credit the reliability data from Phase Two because of the scope of Lockheed’s changes. Standing alone, the “change is a change” statement might have been ambiguous. But in the context of the [* * *] example and the government’s commentary about “complexity” and “quantity” of changes, the government properly led Lockheed into the area requiring elaboration or correction.
Second, Lockheed argues that the notes of a meeting of Army evaluators on May 19 show the Army did not need more data and thus prove that the Army’s requests for more data were misleading. PL’s Mot. at 24-25. As evidence to support this contention, Lockheed points to several lines in the evaluator’s notes of the internal meeting on May 19 that state “[u]p to this point, the main message has been lack of substantiating data, leading L[ockheed] to providing more data. Need to address the fact that a number lower than the proposed reliability can be substantiated. L[ockheed] is well above the reliability performance that can be substantiated.” PI-184 (cited in PL’s Mot. at 25). These notes do not, however, indicate that the data issue was closed. Rather, they reflect the Army evaluators’ continuing concern about establishing a well-supported reliability figure. PI-184.
Third, Lockheed says that the government failed to disclose its internal assessment of Lockheed’s design’s MMBHMF value. PL’s Mot. at 24. On May 6, Lockheed sent an email to the government asking “[w]hat is the [g]overnment’s assessed value of Lockheed Martin’s reliability against the PDFOV-2909 3800 MMBHMF threshold value?” PI-155. The government replied on May 7, explaining that pursuant to RFP § M.4.1, the government was not assessing MMBHMF values, but was instead assessing the risk that an offeror would meet 3800 MMBHMF. PI-154. Lockheed asserts that a chart in the record belies that reply. The Army chart appears to assess MMBHMF values for Lockheed, based on its proposal. PI-166. Beneath this chart is the number “[* * *].” PI-166. In Lockheed’s view, this “[* * *]” figure must be the government’s assessed MMBHMF value for Lockheed, which means that the government incorrectly represented in the May 7 e-mail that the government was not making calculations. PL’s Mot. at 24. Nonetheless, the 'government properly explained to Lockheed that the RFP required an assessment of the risk an offer would not meet threshold MMBHMF, not a precise or particular MMBHMF assessment. PI-154. The fact that a government employee may have calculated an assessed MMBHMF for Lockheed does not show that was the government’s authoritative position, nor that the calculation supplanted the government’s decision . process’ under the RFP. Moreover, Lockheed does not explain what it would have done differently had it known this information before the significant discussions on May 20.
' Finally, Lockheed argues that the discussion on May 20 caused Lockheed to [* * *] proposal and to instead submit its [* * *] proposal, which [* * *]. PL’s Mot. at 22. But this discussion did not cause Lockheed to [* * *]. Lockheed explained the [* * *]. PI-899. Although the government told Lockheed its [* * *] MMBHMF was unsubstantiated, that does not mean Lockheed had to [* * *]. It could have substantiated [* * *]j or any other figure.
The court finds that the discussions with Lockheed were likely meaningful and not *725misleading.12 The Army explained its concerns to Lockheed, it noted the unsubstantiated technical commitment, and it advised Lockheed that it could not rely on future design changes. It also explained what changes varied and needed supporting data. Thus the Army likely fulfilled its responsibilities under FAR § 15.306.
2. Unequal treatment.
Lockheed contends that: (i) it was similarly situated to Oshkosh yet was required to submit more data; (ii) the Army failed to apply a consistent standard when assessing meaningful variation; and (iii) the Army failed to analyze Oshkosh’s changes for meaningful variation, permitting Oshkosh to rely on Phase Two data despite the fact that Oshkosh made serious design changes. Pl.’s Mot. at 24, 26-34. These arguments essentially turn on whose proffered design differed more from the prototypes evaluated in the Phase Two testing. Accordingly, the question of whether the Army fairly asked Lockheed and not Oshkosh for more data depends on whether the evaluators fairly applied the “meaningful variation” standard to both of-ferors.
Pursuant to RFP § M.4.1, “any substantiating data for a design configuration which meaningfully varies from the offered design configuration may be considered less credible.” PI-104. Thus if a proposed design did not meaningfully vary from the prototype design tested by the government in Phase Two, the offeror could rely on government test data to substantiate performance claims. The phrase “meaningfully varies” was not defined in the RFP, Hr’g Tr. at 82:6-7 (government’s counsel stating this point), 132:16-20 (plaintiffs counsel stating same). “Meaningful variation” is a broad term, and Lockheed does not focus on how it should be interpreted but instead argues that the evaluators applied the concept inconsistently. Pl.’s Mot. at 26-32. Lockheed asserts that for its proposal, the Army found meaningful variation if a change created the “potential for a HMF.” Pl.’s Mot. at 27. As evidence of this, Lockheed cites an evaluator’s comment that “L[ockheed] is modifying [* * *]. These design changes have the potential to cause a Mission Essential Function (MEF) failure and therefore meaningfully vary from [Phase Two] test configuration,” Id. (quoting PI-114). Lockheed asserts that for Oshkosh, the Army applied a more lenient standard, asking whether Oshkosh’s changes related to “form, fit and function” or whether they were “likely” to cause a HMF. Pl.’s Mot. at 30-31 (quoting an Army evaluation at PI-129).
The court’s role is to determine whether the evaluators were evenhanded in applying the meaningful-variation standard. See PGBA, 60 Fed.Cl at 207 (holding that “uneven treatment goes against the standard of equality and fair-play that is a necessary underpinning of the federal government’s procurement process”). Under the APA standard of review, the court does not merely address the evaluator’s explanation. The phrases used by the evaluator cannot mask or supplant the facts. See State Farm, 463 U.S. at 43, 103 S.Ct. 2856 (requiring an agency to “examine the relevant data” and “articulate a satisfactory explanation”), The evaluator’s comments about “complexity,” “potential” HMFs or “form, fit and function” are not conclusive. Instead, the court looks to the evidence before the evaluator, asking whether the agency “entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or [the decision] is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.” Id.
The evaluator explained that Lockheed’s Phase Three proposal meaningfully varied because as a matter of fact it had a “quantity of complex” changes from its Phase Two *726design. PI-115 (“[T]he quantity of complex Corrective] AOtion]s results in a design that meaningfully varies from the [Phase Two] as-tested configuration and decreases the relevance of the [Phase Two government] test data.”). The evaluator reiterated this point on a number of occasions, saying, for example, that the “complexity of the components and subsystems that were addressed with [Corrective Actions], combined with the quantity of design changes, results in a proposed design that meaningfully varies from [the Phase Two] test configuration.” PI-114.
The evidence before the evaluator supports her finding that Lockheed made complex changes. Lockheed redesigned [* * *]. PI-915 to -16. This redesign included a change from [* * *]. PI-916. Lockheed also redesigned [* * *]. PI-925 to -27; see also PI-211 (explaining these functions). Next, Lockheed made changes to the [* * *], PI-937 to -39, [* * *], PI-958 to -60, and the [* * *], PI-961 to -65. Lockheed redesigned the [* * *], including the [* * *]. PI-944 to -55. In due course, for all of these changes, Lockheed submitted data to support them. PI-915 to -71.
The evaluator took all this into account in her analysis and concluded that these changes resulted in meaningful variation, which reduced the relevance of Phase Two data. PI-114 (identifying changes to the [*. * *]). The absence of Phase Two testing data for a vehicle reflecting these changes was critically important because, pursuant to RFP §. M.4.1 the evaluator considered Lockheed’s supporting data to have less credibility than that established during the Phase Two testing regime. PI-114 to -16. Given the number of changes Lockheed made to vehicle components, the court cannot discount the evaluator’s findings that these changes were complex as a matter of fact, which findings supported her conclusion that they “meaningfully var[iéd]” under the RFP. The court cannot say the evaluator “offered an explanation for its decision that runs counter to the evidence.” State Farm, 463 U.S. at 43, 103 S.Ct. 2856.
The evaluator did not use the phrase “quantity of complex changes” when discussing Oshkosh’s proposed design. But that does not mean that the evaluator’s analysis of Oshkosh’s proposed design was inconsistent with the analysis she applied to Lockheed. The evaluator explained that Oshkosh’s changes did not meaningfully vary because as a matter of fact the changes addressed “manufacturing efficiencies” and were “the same form, fit and function.” PI-129; see also PI-562 to -82. The government and Oshkosh argue that Oshkosh’s Phase Three changes were not as meaningful as those proposed by Lockheed, because Oshkosh replaced parts with similar but proven parts and did not redesign systems. See generally PI-220, -299, -842-843.
Lockheed disagrees, saying the record shows Oshkosh made three significant design changes that should have drawn more scrutiny from the Army. First, Lockheed points out that Oshkosh changed at* * *] in its proposed design. PL’s Mot. at 33 (citing PI-843). If this part failed, a HMF “would” occur, PI-299, and so Lockheed argues this change is significant. Pl.’s Mot. at 33. But the record shows that the [* * *] was tested on a JLTV during Phase Two and experienced no failures. PI-842. Second, Lockheed argues that Oshkosh changed its [* * *] which is a system that provides [* * *]. Pl.’s Mot. at 34 (citing PI-577). Lockheed says this change should have drawn scrutiny. Pl.’s Mot. at 34. But evidence before the evaluator shows Oshkosh’s change was to [* * *]. PI-220. Given this evidence, the court cannot say it runs counter to the evaluator’s conclusion that the change was not meaningful.
Third, Lockheed argues that Oshkosh changed its [* * *] without drawing scrutiny. Pl.’s Reply Br. in Support of Pl.’s Mot. (“Pl.’s Reply”) at 28, ECF No. 31. The [* * *] is a[* * *] component, and during Phase Two, Oshkosh’s [* * *] which was scored as an EFF. PI-589 to -90. Although aspects of this change were apparently presented during the Phase Two assessment conference, held at the end of the Phase Two testing, this change by Oshkosh was not actually tested on a prototype vehicle during Phase Two. Oshkosh’s change to the [* * •] in large measure involved replacing [* * *], eliminating some [* * *] disparities in Oshkosh’s *727system. PI-589. Oshkosh provided data to support this change. PI-591. It is difficult for the court to analyze the significance of the change in Oshkosh’s [* * *], given the relative absence of details in the record before the court. Although the evaluator seemingly credited Oshkosh with a system that was effectively tested during Phase Two, that assessment appears incorrect. And, the RFP’s attachments indicate that a total failure of this system would result in a HMF. Given that Oshkosh changed some [* * *] in the system, this change may be similar in magnitude to Lockheed’s changes to the [* * *]. But overall, a[* * *] change to Oshkosh’s [* * *] is likely not as meaningful as Lockheed’s redesign of a number of systems, including the [* * *].
In summary, the limited record before the court suggests that the evaluator focused on whether Lockheed’s and Oshkosh’s Phase Three designs varied from their Phase Two designs in meaningful ways. The issue is whether the record supports the evaluator’s conclusion that Lockheed made more changes to important systems, reducing the relevance of Phase Two test data. The record before the court generally supports the evaluator in that regard.13
Because the evaluator found Lockheed’s design meaningfully varied from that field tested during Phase Two, the court cannot say the Army unfairly demanded more substantiating data from Lockheed than Oshkosh. Lockheed’s initial proposal did not submit new data to support its reliability claims. This meant the Army had to ask for substantiating data. In response, Lockheed submitted data to support its changes, but the Army ultimately relied on the degree of change from the prototypes tested in Phase Two to conclude that Lockheed’s data did not deserve full credit. PI-117; see also PI-118 (observing that Lockheed’s changes “decrease[d] the relevance of [g]overnment [Phase Two] data”). In contrast, Oshkosh made fewer changes to important systems. For the changes it did make, it submitted data up front, meaning the Army did not have to ask for more. The Army ultimately concluded that Oshkosh’s changes in fact caused a “slight impact to the relevance of [Phase Two] test data” for Oshkosh. PI-129; see also PI-1011 (noting Oshkosh “generally” relied on government test data). But because those changes were more limited, the impact was only “slight.”
Based on the limited record here, the court finds it likely that the Army treated Lockheed and Oshkosh fairly.14 To the extent the Army treated Lockheed differently, it acted rationally because Lockheed’s proposal raised issues that Oshkosh’s did not. See Chenega Mgmt., LLC v. United States, 96 Fed.Cl. 556, 586 (2010) (finding agency acted rationally when it refused to engage in clarifications with plaintiff, because plaintiffs ratings were “significantly inferior” to other offerors).
3. Adjectival rating.
Lockheed argues the Army erred by assigning it an overall risk rating of “moderate.” Pl.’s Mot. at 35. For the 26 purchase-description requirements, Lockheed ultimately received six “moderate,” twelve “low,” and seven “very low” scores. PI-1010. The RFP required these factors to be “weighted equally.” PI-104 (RFP § M.4.1). Thus, in Lockheed’s view, these scores should average out to a “low” overall risk rating, not “moderate” as the Army found. Pl.’s Mot. at 35-37. As an example, Lockheed says that if one assumes a “very low” is worth four points, a “low” is worth three, a “moderate” is worth two, a “high” is worth one, and an “unacceptable” is worth zero, then Lockheed’s average score was greater than three, proving Lock*728heed deserved an overall low-risk rating. Pl.’s Mot. at 36 n.5.15
Lockheed’s argument assumes a quantitative analysis, rather than qualitative. Although an Army chart in the record suggests some point values may have been internally assigned to the risk categories, see Pl.’s Reply at 13 (citing PI-1585 (in turn showing a chart that appears to average risk scores)), this does not reinforce Lockheed’s assumption that the risk categories would be scored zero to four. Even though the RFP said the 25 requirements were equally weighted, and even assuming the Army did a numerical averaging, any weights for the risk categories have not been disclosed. In that respect, the contracting officer has broad discretion when assigning adjectival ratings. See Wackenhut Servs., Inc. v. United States, 85 Fed.Cl. 273, 293 (2008) (giving deference to evaluator’s assignment of adjectival ratings).
Most importantly, this adjectival rating appears to be a guide, not a dispositive consideration, for the agency, as the Army’s RFP does not say that any particular ratings require any particular results. The rating itself is less important than the rationale behind it. Thus “[e]ourts should look beyond the adjectival ratings because proposals awarded the same adjectival ratings are not necessarily equal in quality.” Blackwater Lodge & Training Ctr., Inc. v. United States, 86 Fed.Cl. 488, 514 (2009). Whatever overall rating the Army chose here, the record shows that the Army considered Oshkosh’s proposal to be technically less risky across the 25 factors, a fact the SSA discussed in his final selection decision. See PI-1014 (stating that although the overall risk rating for Oshkosh was low and that for Lockheed was moderate, “the proposals were not as close as this might imply as [Oshkosh] is assessed as Low Risk or Very Low Risk for all 25 requirements whereas ... [Lockheed has] a number of Moderate Risk ratings.”). Oshkosh received thirteen “low,” twelve “very low,” and no “moderate” ratings. It thus demonstrated lower risk than Lockheed, which had six “moderate ratings.” Given these differences, it is difficult for the court to say the contracting officer acted arbitrarily-
4. Prejudice.
The court may grant relief in a bid protest only if the plaintiff demonstrates it was prejudiced by the government’s action. See Information Tech. & Applications Corp. v. United States, 316 F.3d 1312, 1319 (Fed.Cir.2003); see also 5 U.S.C. § 706 (last sentence) (requiring the court to take “due account” of “the rule of prejudicial error”). To be prejudiced, a protestor must have had a substantial chance of receiving the award but for the government’s error. Id.
Here, the Army selected Oshkosh after weighing which offeror would provide the “best value” to the government. The Primary Technical factor weighed more than price, which weighed more than Small Business Participation. Against this Primary Technical standard, the SSA found that Oshkosh had better risk scores, as detailed above. The SSA also found that Oshkosh was cheaper by nearly $400 million. PI-1009. As for small business participation, Lockheed was rated “outstanding” while Oshkosh was rated “good.” PI-1009. Accordingly, the SSA selected Oshkosh because it had the best Primary Technical rating and the lowest price. PI-1014. Although Lockheed had better small business participation, this was the weakest factor and did not outweigh the other two.
Lockheed argues it was prejudiced because “[h]ad Lockheed Martin been properly evaluated with respect to Reliability and not been misled, it would have made a substantially different MMBHMF proposal. If Lockheed Martin had made such a proposal — or Oshkosh’s proposal had been given equivalent scrutiny — a different, more favorable [life cycle cost] adjustment would have resulted in an award to Lockheed Martin.” Pl.’s Mot. at 37.
*729The court is not persuaded. Even with adjustments to the Reliability and Operational Availability elements, Oshkosh would still be technically superior. And, the RFP provided that the Primary Technical factor weighed “more” than price, unless non-price factors were approximately equal, PI-101, and that “this determination may result in award to other than the [ojfferor with the lowest TEC/P.” PI-102.
Therefore, to establish prejudice Lockheed cannot only show that the technical evaluation was flawed. It must also show that flaws affected the price calculation enough to make price almost determinative. Cf. Weston Solutions, Inc. v. United States, 95 Fed.Cl. 311, 327 (2010) (finding prejudice when, but for an error in the technical evaluation, plaintiff and the awardee could have been technically equal, allowing the government to use non-technical factors as the tie-breaker), aff'd, 440 Fed.Appx. 926 (Fed.Cir.2011). Lockheed has not done so at this point. There is no evidence in the record showing what alternative life cycle credit might have been given. See Industrial Property Mgmt., Inc. v. United States, 59 Fed.Cl. 318, 324 (2004) (finding that when offerors were technically equal, plaintiff did not have substantial chance of award when his bid was 20% more expensive than the winner); see also CMI Mgmt., Inc. v. United States, 115 Fed.Cl. 276, 298 (2014) (finding no prejudice on alleged error in rating of a subfactor, because “even if the court were to find that [plaintiffs] rating for Subfactor Four as Acceptable rather than Good was ill-founded, ... this subfactor would have no material bearing on [plaintiffs] overall rating of Acceptable.”).
Analytically also, on the merits, Lockheed has not yet satisfactorily addressed the changes it made to its MMBHMF figures during the discussions. Lockheed ultimately proposed a MMBHMF of [* * *]. Although in its reply brief Lockheed suggests that it could have substantiated a MMBHMF of [* * *] but the government prevented it from doing so, Pl.’s Reply at 9, that suggestion has limited support in the record. Lockheed’s final offer explained that the [* * *] figure “is largely based on the government’s decision to not recognize any significant value for continued reliability growth activities that occur after contract award, even though a contractor has a contractor commitment to meet the proposed value.” PI-899. That concession diminishes Lockheed’s overall posture and the likelihood that it could show prejudice.
B. Irreparable Harm
In the context of a preliminary injunction, a protestor can prove irreparable injury by showing that an injunction is necessary to preserve the plaintiffs remedies should it succeed on the merits. Eskridge Research Corp. v. United States, 92 Fed.Cl. 88, 98 (2010) (citing Qingdao Taifa Grp. Co. v. United States, 581 F.3d 1375, 1379-80 (Fed.Cir.2009)). But in general; a plaintiff may not show irreparable harm by claiming generically that the winner’s transition into performance will give it advantages. See IBM Corp. v. United States, 118 Fed.Cl. 677, 684 (2014) (finding no irreparable harm from winner’s beginning performance).
Lockheed makes two arguments that it will be irreparably harmed without a preliminary stop-work, order. First, Lockheed asserts that without temporary relief, this court may be unable to provide a remedy if Lockheed later prevails on the merits. See Pl.’s Reply at 33. All agree that Oshkosh will perform largely preparatory work in the coming months, thus limiting potential termination costs. See Def.—Interv.’s Opp’n at 39 n.8 (quoting a declaration from a Vice President of Oshkosh explaining that “termination costs would be very modest relative to the value of this procurement, and would not preclude an effective remedy”); Def.’s Opp’n at 35, EOF No. 26 (“The amount of work to be performed over the next several months is very small relative to the overall contract.”); PL’s Mot. at 39 (“[T]here is no urgency here.”). Given the proportionately small commitment of resources compared to the overall value of this contract, Oshkosh’s continuing performance would not prevent this court from giving Lockheed an appropriate remedy should Lockheed ultimately prevail. See Turner Constr. Co. v. United States, 645 F.3d 1377, 1388 (Fed.Cir.2011) (noting that *730the Court of Federal Claims “has broad equitable powers to fashion an appropriate remedy”).
Second, Lockheed argues that “Oshkosh may be well into performance when this matter is resolved. Oshkosh may have the opportunity to further develop, change, and refine its solution under [glovernment funding and testing before any corrective action, which will adversely affect Lockheed Martin’s prospects in any re-competition ordered by the [c]ourt.” Pl.’s Mot. at 38. The only evidence of this harm cited by Lockheed is a chart, drafted by Oshkosh during the competition, showing that Oshkosh will [* * *] post-award to test its own vehicles. Hr’g Tr. 142:6-12 (citing PI-273 to show that Oshkosh will conduct reliability testing). This does not show how Lockheed will be harmed.
C. Remaining Factors
Because Lockheed has failed to establish a likelihood of success on the merits and irreparable harm, the court cannot grant a preliminary injunction. See Altana Pharma, 566 F.3d at 1005. Accordingly, the court need not consider the remaining factors pertinent to preliminary injunctive relief.
CONCLUSION
Plaintiffs motion for a preliminary injunction is DENIED. On or before February 24, 2016, the court requests that the parties submit a joint status report addressing further proceedings in this action.
It is so ORDERED,