MirrorWeb Limited

DOCUMENT FOR PUBLIC RELEASE

The decision issued on the date below was subject to a GAO Protective Order. This redacted version has been approved for public release.

Decision

Matter of: MirrorWeb Limited

File: B-424554; B-424554.2

Date: August 24, 2026

Thomas A. Pettit, Esq., and Craig A. Holman, Esq., Arnold & Porter LLP, for the protester.
Andrew T. Brown, Esq., and Emily Vartanian, Esq., Library of Congress, for the agency.
Jacob M. Talcott, Esq., and April Y. Shields, Esq., Office of the General Counsel, GAO, participated in the preparation of the decision.

DIGEST

1. Protest challenging the agency's evaluation of the protester's technical proposal and the awardee's price is denied where the evaluation was reasonable and in accordance with the terms of the solicitation.

2. Protest challenging the adequacy and fairness of discussions is denied where the record shows that the agency met its obligation to conduct adequate and equal discussions.

DECISION

MirrorWeb Limited, of Manchester, United Kingdom, protests the award of a contract to Internet Archive, of San Francisco, California, under request for proposals (RFP) No. 030ADV26R0010, issued by the Library of Congress, for web harvesting services. The protester contends that the agency unreasonably evaluated its proposal, which resulted in the improper elimination of its proposal from the competition. The protester also argues that the agency conducted misleading and unequal discussions, and unreasonably evaluated the awardee's price.

We deny the protest.

BACKGROUND

On January 22, 2026, the agency issued the RFP in accordance with Federal Acquisition Regulation (FAR) part 12, using part 15 procedures, and sought an offeror to provide web harvesting services for the Library of Congress's Digital Services Directorate.[1] Contracting Officer's Statement (COS) at 1; RFP at 55. The RFP provided for the award of an indefinite-delivery, indefinite-quantity contract for a period of five ordering years. RFP at 5. The due date for proposals, as amended, was March 3, 2026.[2] AR, Exh. 4e, RFP amend. 0005 at 1.

The RFP provided for the submission of proposals in four volumes, one for each of the following evaluation factors: (1) technical approach; (2) corporate experience and capabilities; (3) past performance; and (4) price. RFP at 52‑57. As relevant here, the technical approach factor consisted of two subfactors: technical approach and capabilities; and sample web crawl. Id. at 52, 56.

For the technical approach and capabilities subfactor, offerors were to provide a narrative that demonstrated an understanding of the goals and objectives of the statement of work (SOW), and how the offeror would perform the requirements. Id. The solicitation also required offerors to provide a method for verifying complete transfers of harvested content, noting that the “BagIt” hierarchical packaging format was the agency's preferred method of transfer. Id. at 11. For this subfactor, the agency would assess the degree to which the offeror demonstrated an understanding of the work and feasibility of completing the requirements. Id. at 56. The agency also would evaluate the offeror's methodology and quality assurance approach to assess the degree to which the offeror would perform and oversee web crawls in accordance with the SOW. Id.

For the sample web crawl subfactor, offerors were to perform a “technical test” in accordance with attachment J2a of the solicitation, which required offerors to run a sample web crawl using an agency-provided “seed list” that contained 100 seeds of varying type and complexity.[3] Id. at 52; see AR, Exh. 3(c), attach. J2a at 1. The solicitation further instructed that if an offeror intended to use multiple web harvesting technologies, the offeror “must” repeat the technical test “for each distinct web harvest technology (i.e., crawler).” AR, Exh. 3(c), attach. J2a at 1; see RFP at 52. For this subfactor, the agency would evaluate the results of the sample crawl to determine the extent to which the offeror is capable of providing the services described in the SOW. RFP at 56. The solicitation further provided that each web harvesting technology proposed would be evaluated for the degree to which it illustrated the offeror's understanding and compliance with, among other things, the specifications of the SOW and attachment J2a of the solicitation. Id.

For price, offerors were to provide pricing based on the unit price per terabyte and were permitted to provide discounted pricing for bulk orders of the terabytes purchased. Id. at 54. The agency would evaluate price for fairness and reasonableness. Id. at 57.

The solicitation provided for award on a best-value tradeoff basis where the evaluation factors were listed in descending order of importance (i.e., technical approach was the most important evaluation factor, and price was the least important evaluation factor). Id. at 55. The solicitation further provided that all subfactors were of equal weight to each other. Id. All non-price factors, when combined, were significantly more than or equally important to price. Id.

The Initial Evaluation and Discussions

The agency received nine proposals by the March 3 due date for proposals, including proposals from Internet Archive and MirrorWeb. COS at 2. On March 31, the contracting officer issued a competitive range memorandum where all proposals were eliminated except for Internet Archive's proposal and MirrorWeb's proposal. Id. at 3. The agency then entered into discussions with Internet Archive and MirrorWeb that same day. Id.

In conducting discussions with MirrorWeb, the agency explained that MirrorWeb's proposal received ratings of marginal under the technical approach and capabilities subfactor and the sample web crawl subfactor due to various weaknesses and deficiencies.[4] AR, Exh. 11, Competitive Range and Discussions at 1. Under the technical approach and capabilities subfactor, the agency explained that MirrorWeb's proposal received a weakness because it failed to provide certain information about the proposed workflow. Id. Under the sample web crawl subfactor, the agency stated that MirrorWeb's proposal received the following three significant weaknesses: (1) MirrorWeb performed per-seed crawls using multiple technologies, but it was unclear which technology was used for which seed, and which reports were generated by which crawler type; (2) MirrorWeb did not designate which seeds were in which crawler and did not crawl all 100 seeds with each technology, thereby making it difficult to determine whether the generated reports met the requirements for parity between reports for the two crawlers; and (3) MirrorWeb did not perform separate crawls per technology but performed a single crawl. Id. at 2.

With regard to the deficiencies, the agency assessed one deficiency in MirrorWeb's proposal under the technical approach and capabilities subfactor and one deficiency under the sample web crawl subfactor. Id. For the first deficiency, the agency stated that MirrorWeb's technical approach did not state that it would use the “BagIt” format as provided in the SOW. Id. For the second deficiency, similar to the significant weakness discussed above, the agency stated that MirrorWeb did not perform separate crawls per technology as required by the solicitation but performed only a single crawl under the sample web crawl subfactor. Id. For both deficiencies, the agency requested that the protester address these issues in its revised proposal. Id.

The Evaluation of Revised Proposals

MirrorWeb submitted its revised proposal on April 2, and the agency completed its evaluation of revised proposals on April 28. COS at 4. The evaluators concluded that, while MirrorWeb's revised proposal resolved almost all the weaknesses noted during discussions, the revised proposal also gave rise to “new concerns and additional weaknesses” and did not address all of the deficiencies. AR, Exh. 16(c), MirrorWeb Final Evaluation Summary Form at 6. Concerning the weaknesses, the evaluators stated that the additional weaknesses “were specifically with the organization of sample crawl,” noting that MirrorWeb's approach was not aligned with the written technical proposal or with the sample web crawl instructions. Id. The agency further explained that there were new concerns regarding MirrorWeb's capacity to conform to the agency's preferred BagIt transfer format and that MirrorWeb's response to questions regarding this matter “indicated a lack of experience in delivering content in this format.” Id.

The agency also noted that there was a “remaining, unresolved deficiency” involving the sample web crawl. Id. Specifically, the agency explained that MirrorWeb failed to conduct the sample crawl in accordance with the instructions provided in Attachment J2A by repeating the technical test for the two technologies that it proposed. Id. at 3. Instead, MirrorWeb used its “proposed pre-crawl analysis” to sort the seeds into one or the other proposed web crawl technology and conducted a single crawl. Id. Due to these deficiencies, MirrorWeb's proposal received a rating of unacceptable under the sample web crawl subfactor. AR, Exh. 20, Debriefing at 3. The agency also eliminated MirrorWeb's proposal from the competition. AR, Exh. 17, Award Memorandum at 16.

As relevant to this protest, the final evaluation results for Internet Archive and MirrorWeb were as follows:

	Internet Archive	MirrorWeb
Technical Approach and Capabilities	Outstanding	Good
Sample Web Crawl	Good	Unacceptable
Total Evaluated Price	$5,130,000 - $11,340,000	$1,480,000 - $1,724,000

AR, Exh. 20, Debriefing at 4.[5] The agency explained that while MirrorWeb's proposal “reflected good value” to the agency, the rating of unacceptable under the sample web crawl subfactor rendered its proposal ineligible for award.[6] Id.

The agency explains that “[o]nce it was clear that Internet Archives was the only remaining contractor for award, it was important to attempt to reduce pricing.” AR, Exh. 17, Award Memorandum at 16. Accordingly, the agency requested that Internet Archive consider reducing its price to reflect “their best possible offer.” Id. Once Internet Archive submitted revised pricing and the agency concluded that the revised price was fair and reasonable, the agency selected Internet Archive's proposal for award, explaining that it demonstrated “exceptional quality and . . . [was] the only one that met all aspects of the solicitation successfully.” Id. at 18.

The agency notified MirrorWeb of the award decision on May 26. AR, Exh. 19, Notice of Competition Outcome at 1. MirrorWeb requested a debriefing on May 27, which the agency provided the same day. AR, Exh. 20, Debriefing at 1. This protest followed.

DISCUSSION

MirrorWeb primarily challenges the evaluation of its proposal and the adequacy of discussions. See Protest at 15‑22. Specifically, MirrorWeb argues that the agency unreasonably assigned its proposal a deficiency under the sample web crawl subfactor. Id. at 15. MirrorWeb also contends that, to the extent the deficiency was reasonably assigned, the agency's discussions were misleading and unequal. Id. at 22; Comments and Supp. Protest at 23. Finally, MirrorWeb challenges the agency's evaluation of Internet Archive's price. See Protest at 25-27. For reasons discussed below, we deny the protest.[7]

Challenge to the Agency's Evaluation of MirrorWeb's Proposal

In reviewing a protest challenging an agency's evaluation, our Office will not reevaluate proposals or substitute our judgment for that of the agency, as the evaluation of proposals is a matter within the agency's discretion. RoadHero LLC, B-423648.2, July 30, 2025, at 4. Rather, we will review the record to determine whether the agency's evaluation was reasonable and consistent with the stated evaluation criteria and applicable procurement statutes and regulations. Id. A protester's disagreement with the agency's judgment, without more, is insufficient to establish that an evaluation was unreasonable. Id.

MirrorWeb first argues that the agency unreasonably assessed a deficiency in its proposal under the sample web crawl subfactor. Protest at 15. Specifically, MirrorWeb asserts that the agency unreasonably penalized it for “selectively crawl[ing]” the seed list by dividing the seeds between its proposed technologies instead of crawling the entire seed list against both technologies. Id. at 18. MirrorWeb contends that the solicitation did not require offerors to run the entire seed list against each crawler but required only that offerors repeat their proposed technical approach for each of the proposed technologies. Id. at 19. The agency responds that it evaluated the protester's proposal in accordance with the terms of the solicitation and reasonably concluded that its proposal was technically unacceptable. Memorandum of Law (MOL) at 2. Specifically, the agency asserts that the solicitation required offerors to run the entire 100-seed list against every proposed technology, which the protester failed to do. Id.

We have no basis to object to the agency's assessment of a deficiency here. As discussed above, the solicitation required offerors to perform a sample web crawl in accordance with attachment J2a of the solicitation. RFP at 52; AR, Exh. 3(c), Attach. J2a at 1. Attachment J2a provided, among other things, that offerors were to run a sample web crawl using the agency-provided list of 100 seeds. AR, Exh. 3(c), Attach. J2a at 1. Attachment J2a further provided that if an offeror proposed multiple web harvesting technologies, “[t]he technical test must be repeated for each distinct web harvest technology (i.e., crawler).” AR, Exh. 3(c), Attach. J2a at 1; see RFP at 52. During discussions, the agency identified a deficiency with respect to MirrorWeb's sample web crawl, stating that MirrorWeb failed to perform the web crawl in accordance with the terms of the solicitation because it performed only “a single crawl.” AR, Exh. 11, Competitive Range and Discussions at 2. The agency then requested that MirrorWeb address this deficiency in its revised proposal. Id. Following the review of MirrorWeb's revised proposal, the agency again noted that MirrorWeb did not perform “a 100‑seed sample crawl per-technology as required in the instructions” but performed only a single crawl. AR, Exh. 16(c), MirrorWeb Final Evaluation Summary Form at 6. The agency therefore concluded that MirrorWeb's proposal was unacceptable because it failed to resolve the deficiency identified during discussions, namely the failure to repeat the technical test for each of the two proposed technologies. Id. at 1.

MirrorWeb's argument that the agency “manufacture[d] a requirement to run each crawler against the entire seed URL list” by “offer[ing] language not found in the Solicitation” is unsupported by the record. Comments and Supp. Protest at 12. The RFP expressly provided that offerors were to “perform a one-time crawl” using the “Library-provided seed list . . . containing 100 seeds” and that the technical test “must be repeated for each distinct web harvest technology.” AR, Exh. 3(c), Attach. J2a at 1; see RFP at 52. As these provisions make clear, the solicitation set forth a test that required offerors to run the agency-provided seed list against their web crawler technology. Id. In the event that the offeror proposed more than one web crawler technology, the solicitation required that the offeror repeat that test for each technology. Id. MirrorWeb concedes that it did not run the 100-seed list against both technologies but split the seed list among its two proposed crawlers. See Protest at 19 (arguing that “[n]othing in those instructions precluded . . . splitting the seed list into small seed crawls”). This test did not align with the requirements set forth in the solicitation because it divided the 100-seed list among the technologies that the protester believed were “best suited for each seed.” Id. We therefore have no basis to object to the agency's decision to assess a deficiency in the protester's proposal for failing to run the 100-seed list against each of the technologies, resulting in an unacceptable rating. This protest ground is denied.[8]

Challenge to the Adequacy of Discussions

MirrorWeb next argues that, even if the agency can justify the deficiency under the sample web crawl subfactor, the deficiency stemmed from the misleading nature of the agency's discussions. Id. at 22. In this regard, MirrorWeb argues that the agency identified three significant weaknesses and one deficiency during discussions, and none of those findings raised the agency's “actual concern,” which was the protester's failure to run the 100-seed list against both of its proposed web crawling technologies. Id. at 25. Instead, MirrorWeb asserts that discussions asked only that the protester explain which crawler MirrorWeb used for each seed. Id. Accordingly, MirrorWeb contends that the discussions were not meaningful because they misinformed the protester as to the true nature of the agency's concerns with its proposal. Id. The agency responds that discussions were meaningful because they clearly identified MirrorWeb's failure to conduct its web crawl in accordance with the solicitation by running the entire seed list against each of the technologies. MOL at 10.

As a general matter, discussions must be meaningful--that is, they must identify deficiencies and significant weaknesses that exist in an offeror's proposal--but that requirement is satisfied when an agency leads an offeror into the areas of its proposal that require amplification or revision. See, e.g., Epsilon Sys. Sols., Inc., B-409720, B‑409720.2, July 21, 2014, at 16. Furthermore, the requirement that discussions be meaningful does not obligate an agency to “spoon-feed” an offeror or to discuss every area where the proposal could be improved. General Dynamics Info. Tech., Inc., B‑418533, June 11, 2020, at 9.

We have no basis to sustain MirrorWeb's argument that the agency failed to conduct meaningful discussions. As discussed above, the agency identified three significant weaknesses and one deficiency related to MirrorWeb's sample web crawl. See AR, Exh. 11, Competitive Range and Discussions at 2. As relevant here, the third significant weakness provided that it was unclear which of the protester's proposed technology was used for each seed and which reports were generated by which crawler type, and further identified MirrorWeb's failure to perform separate crawls for each technology. Id. The deficiency also expressly provided that the protester failed to “perform separate crawls per technology as required by J2a.” Id. Although MirrorWeb asserts that discussions were not meaningful because none of the findings made by the agency addressed MirrorWeb's failure to run each crawler against each seed, Protest at 25, this assertion is belied by the record. As the third significant weakness and deficiency make clear, the agency expressly identified the failure of the protester's proposal to run separate crawls for each technology as set forth in the attachment J2a of the solicitation. See AR, Exh. 11, Competitive Range and Discussions at 2. To the extent that MirrorWeb asserts that the agency should have provided more detail during discussions or should have been more direct in explaining the revisions it expected, we conclude that the agency met its burden of advising the protester of the areas of its proposal that it needed to improve. This protest ground is denied.

In its supplemental protest, MirrorWeb argues that discussions also were unequal. Comments and Supp. Protest at 16. Specifically, MirrorWeb contends that the agency entered into a second round of discussions with the Internet Archive and allowed the awardee to submit a second revised proposal, but it did not offer MirrorWeb the same opportunity. Id. The agency responds that discussions were not unequal because MirrorWeb's proposal was eliminated from the competition once it failed to remedy the areas raised by the agency during the first round of discussions. Supp. MOL at 3. Accordingly, the agency contends that it was not obligated to include MirrorWeb in the subsequent round of discussions. Id.

When an agency conducts discussions, it may limit the competitive range to only the most highly rated proposals and need not conduct discussions with offerors outside the competitive range. See Priority One Servs., Inc., B-415201.2, B-415201.3, Apr. 13, 2018, at 9. Where a proposal is technically unacceptable as submitted and would require major revisions to become acceptable, the agency is not required to include the proposal in the competitive range. Logistics Mgmt. Inst., B-419219, B‑419219.2, Dec. 30, 2020, at 7; Laboratory Sys. Servs., Inc., B-256323, June 10, 1994, at 5.

Although MirrorWeb asserts that our decisions establish a “clear rule” that when an agency reopens discussions with one offeror after the receipt of final proposal revisions, “it must afford all offerors in the competitive range” the same opportunity, Comments and Supp. Protest at 24, this obligation does not extend to proposals that are technically unacceptable and further eliminated from the competitive range. Laboratory Sys. Servs., Inc., supra. As discussed above, following the first round of discussions, the agency concluded that MirrorWeb's proposal was unacceptable due to the deficiency regarding the sample web crawl. See AR, Exh. 11, Competitive Range and Discussions at 2. Specifically, the agency explained that MirrorWeb's proposal failed to perform separate crawls as required by the solicitation, and the agency then instructed MirrorWeb to address this deficiency during discussions. Id. In conducting the evaluation of MirrorWeb's revised proposal, the agency concluded that MirrorWeb did not resolve the deficiency and its sample web crawl still was not conducted in accordance with the terms of the solicitation. AR, Exh. 16(c), MirrorWeb Final Evaluation Summary Form at 3. Accordingly, the same deficiency that was present in the initial proposal was also present in the revised proposal, which made the proposal ineligible for award. See id. Indeed, the agency documented its concerns about MirrorWeb's proposal, and the decision to eliminate MirrorWeb from the competition was made prior to engaging further with Internet Archive, which was then “the only remaining contractor for award.”[9] AR, Exh. 17, Award Memorandum at 16; see also Decl. of Contract Specialist at 1-2.

To the extent that MirrorWeb contends that the agency was required to engage in further discussions with MirrorWeb prior to eliminating its revised proposal from the competition, we find no basis for this position. Instead, our decisions explain that agencies are not required to engage in successive rounds of discussions until all proposal defects have been corrected, nor are agencies required to reiterate concerns that were not alleviated by a firm's proposal revisions. ANHAM FZCO, B-414770 et al., Sept. 14, 2017, at 6. Under these circumstances, this protest ground is denied.

Challenge to the Evaluation of Internet Archive's Price

MirrorWeb also contends that the agency unreasonably evaluated Internet Archive's proposal under the price factor. Protest at 25. Specifically, MirrorWeb asserts that the agency “accepted a price premium that could be as much as . . . 558 percent higher than MirrorWeb's price,” which was unreasonable and should have resulted in finding Internet Archive ineligible for award. Id. at 26. The agency responds that the price evaluation was reasonable and in accordance with applicable regulations. MOL at 12. In this regard, the agency argues that it compared the proposed prices to its independent government cost estimate (IGCE) and market research, and concluded that both MirrorWeb's and Internet Archive's prices were reasonable. Id. at 14.

Based on our review of the record, we have no basis to sustain this protest ground. It is a fundamental principle of federal procurement law that procuring agencies must condition the award of a contract upon a finding that the contract contains “fair and reasonable prices.” FAR 15.402(a), 15.404-1(a); see Crawford RealStreet Joint Venture, B-415193.2, B-415193.3, Apr. 2, 2018, at 9. The purpose of a price reasonableness analysis is to prevent the government from paying too high a price for a contract. Id. The manner and depth of an agency's price analysis is a matter committed to the discretion of the agency, which we will not disturb provided that it is reasonable and consistent with the solicitation's evaluation criteria and applicable procurement statutes and regulations. Torrent Techs., Inc., B-419326, B-419326.2, Jan. 19, 2021, at 6.

As stated above, the solicitation required offerors to provide pricing information based on the unit price per terabyte, which the agency evaluated for fairness and reasonableness. RFP at 54, 57. The RFP did not specify a method for the agency's price reasonableness evaluation, thereby allowing the agency to exercise its discretion.

In conducting its market research to establish the IGCE, the agency issued requests for information (RFIs) to MirrorWeb, Internet Archive, and other vendors. COS at 6; AR, Exh. 17, Award Memorandum at 15‑16. As relevant here, in response to the RFI, MirrorWeb quoted unit prices of $[DELETED]/terabyte for a “simple” crawl and $[DELETED]/terabyte for an “involved” crawl. COS at 6; see AR, Exh. 8(e), MirrorWeb Pricing Information at 3. Because the agency determined level of effort for the current procurement would be “involved,” it relied on the latter figure to inform its market research and IGCE. COS at 6; AR, Exh. 17, Award Memorandum at 16. The agency concluded that unit prices around $3,000/terabyte for ordering year one and $3,500/terabyte for ordering year five would be a reasonable range. AR, Exh. 17, Award Memorandum at 14.

In reviewing Internet Archive's price proposal, the agency noted that its proposed unit prices, which ranged from $3,040 to $3,400 per terabyte, were “nearly identical” to the prices provided in the market research. Id. at 15. While MirrorWeb's proposed unit prices were significantly lower, they were comparable to the researched unit prices of MirrorWeb's “simple” crawl, which the agency viewed as an “inaccurate representation of the work the Library needs.” Id. at 15. Furthermore, while the agency recognized that it could receive “major savings” with MirrorWeb's proposal, MirrorWeb's proposal was unawardable due to the deficiency assessed under the sample web crawl subfactor. Id. at 16. Once it was clear that Internet Archive's proposal was the only proposal that remained in the competition, the agency requested that Internet Archive reduce its pricing to reflect its market research submissions. Id. Internet Archive responded by submitting revised pricing, reducing the unit price by approximately $100 per terabyte. Id. In light of these findings and revisions, the agency concluded that Internet Archive's price was fair and reasonable, and selected its proposal for award. Id.

MirrorWeb objects to the reasonableness of the evaluation, asserting that its total evaluated price was “$1,480,000 to $1,724,000 depending on the particular services ordered” and the awardee's price contained “a massive price range of $5,130,000 to $11,340,000.” Protest at 26. According to MirrorWeb, the agency unreasonably “accepted a price premium that could be as much as . . . 558 percent higher than MirrorWeb's price.” Id. We find this argument unconvincing. As an initial matter, the fact that the awardee's price was significantly lower than the protester's price does not mean that the awardee's price was unreasonable. Chugach Range & Facilities Servs. JV, LLC, B-420458.6, B-420458.10, Jan. 10, 2024, at 16. Instead, the protester must demonstrate some flaw in the agency's methodology or otherwise establish that the awarded price is unreasonably high. Id.

MirrorWeb has not satisfied this standard. As discussed above, the agency compared the pricing for Internet Archive's proposal to the pricing information from its market research, concluding that Internet Archive's pricing was within the acceptable range of $3,000/terabyte. AR, Exh. 17, Award Memorandum at 16. While MirrorWeb asserts that the evaluation was unreasonable because its pricing was significantly lower than Internet Archive's pricing, the agency points out that MirrorWeb submitted a unit pricing of $[DELETED]/terabyte for “involved” crawls in response to the agency's RFI. MOL at 13 n.21. To now object to a unit price in the range of $[DELETED]/terabyte is unpersuasive as it is essentially a disagreement with the agency's determination that the level of effort required for this procurement would be more akin to an “involved” crawl rather than a “simple” crawl. Because Internet Archive's unit pricing was within the $3,000/terabyte range, we have no basis to object to the reasonableness of the agency's evaluation and its conclusion that Internet Archive's price was reasonable and eligible for award. This protest ground is denied.

The protest is denied.

Edda Emmanuelli Perez
General Counsel

---