OSI Collection Services, Inc., B-286597; B-286597.2, January 17, 2001

Decision

File: B-286597; B-286597.2

Date: January 17, 2001

Joseph J. Petrillo, Esq., and Karen D. Powell, Esq., Petrillo & Powell, for the protester.

Jeffrey C. Morhardt, Esq., and Jose Otero, Esq., Department of Education, for the agency.

Tania Calhoun, Esq., and Christine S. Melody, Esq., Office of the General Counsel, GAO, participated in the preparation of the decision.

DIGEST

Protests against award of federal supply schedule task order contracts for private collection agency services are sustained where the record shows that the contracting agency's evaluation of offerors' past performance, which largely relied upon a mechanical comparison of past performance scores for incumbent contractors, was unsupported and unreasonable.

DECISION

OSI Collection Services, Inc. protests the decision by the Department of Education to award federal supply schedule (FSS) task order contracts to numerous other firms under a request for task order proposal (RFTOP) for private collection agency (PCA) services. OSI contends that the agency's evaluation of offerors' past performance was inconsistent with the solicitation's stated evaluation terms, unsupported, and unreasonable.

We sustain the protests.

BACKGROUND

The agency's Office of Student Financial Assistance performs collection and administrative resolution activities on debts resulting from nonpayment of student loans made under numerous federal programs, and on debts resulting from a student's failure to fulfill grant requirements under other federal programs. At the time this solicitation was issued, 17 PCA contractors, including OSI, were performing these services for the agency under a contract that commenced in 1997 and runs through September 2001.

The RFTOP was issued in July 2000 to obtain the services of PCAs with contracts under the General Services Administration's Financial Asset Management Services Schedule, Special Item Number (SIN) 621-4, "Loan and Other Asset Services/Management." Task order contracts were to be issued to 10-12 FSS contractors, with at least two awarded under a small business set-aside. The agency announced that it had about two million accounts, worth about $9 billion, to place under this award. It planned to conduct an initial transfer of 20,000 accounts to each successful contractor, each of whom was to locate and contact the borrowers to demand payment of their debts or to otherwise resolve the account through such measures as wage garnishment, litigation, or other administrative resolutions. Additional account transfers were to occur throughout the life of the contract.

Firms could compete for the task order contracts only if they were specifically invited to do so. Incumbent contractors, such as OSI, were to be invited to compete only if they had performed "consistently well" for the agency based upon its Competitive Performance and Continuous Surveillance (CPCS) evaluation. The CPCS evaluation, performed every 4 months, measures the relative performance of each contractor on all accounts transferred under various performance indicators and is used to determine bonus payments and the transfer of new accounts. Under the CPCS methodology, the contractor ranked the highest under a particular performance indicator receives the maximum number of points available for that indicator, and the remaining contractors receive points in proportion to their standing relative to the leading contractor. Each contractor's overall CPCS score for each 4-month period is the sum of its scores for all of the performance indicators for that period.

Offerors were required to propose a commission or fee for each type of service to be performed under the contract. Since the RFTOP established target rates for these commissions or fees, "quality factors" and the commitment to small business were to be more important than price in making the award selection decision. Section M.1.b. of the RFTOP stated that evaluation factors were to be considered in the following order of importance: past performance, including the past performance of key personnel; technical evaluation; commitment to small business; and price.

The past performance evaluation is the critical issue in these protests. Section M.2. of the RFTOP provided that the following past performance information was to be obtained and considered: a Dun & Bradstreet (D&B) past performance evaluation for all offerors; information obtained when checking references for all offerors; and, "[f]or those companies with a current contract, the Department will use performance data that we have on hand such as the CPCS scores." Recent and relevant information was to receive greater consideration than less recent and less relevant information. The agency considered competitive ranking information to be "extremely relevant." RFTOP M.2.

Twenty-six FSS contractors submitted proposals in response to the RFTOP. Thirteen contractors were large businesses currently performing PCA services for the agency, including OSI. Of the other 13 FSS contractors, 7 were non-incumbents and 6 were small businesses invited to compete for award under the small business set-aside portion of the solicitation.

A source evaluation board (SEB) met to consider offerors' proposals. The SEB considered three elements in evaluating past performance. First, for each offeror, the SEB reviewed the results of a D&B past performance evaluation. These results were in the form of numerical scores on a scale of 1-5, with 1 being "outstanding" and 5 "unsatisfactory". Second, for each offeror, the SEB reviewed the results of interviews conducted with contractor-provided references. These results included subjective comments along with numerical scores on a scale of 1-4, with 1 being "extremely satisfied" and 4 "never satisfied." Third, for incumbent contractors, the SEB considered the CPCS score attained over the life of the contract--the arithmetic total of the seven periodic CPCS scores available at that time. Based upon that total score, the SEB placed the firms in a first or a second CPCS "tier." The second tier was defined as containing firms with an overall performance of average or below. At the conclusion of the evaluation, the SEB assigned each PCA an overall rating of a low, moderate, or high probability of success based on the past performance, technical, and small business commitment evaluations. [¹]

OSI's overall D&B rating was [DELETED], placing its performance between very good and outstanding. The firm's overall reference rating was [DELETED], with two of its three references responding to the agency's inquiries. [²] The evaluators noted that OSI's clients were extremely satisfied with the firm's performance. With respect to the CPCS evaluation, the SEB concluded that OSI was in the second tier of the 17 PCAs. In making its overall assessment, the SEB stated that OSI demonstrated a thorough knowledge of student loan collections--its technical proposal was evaluated as having a high probability of success. [³] The SEB concluded, however, that the most relevant indicator of success was the CPCS evaluation. According to the SEB, while "net back recovery percentages" were important, contractors had to perform the full range of requirements. [⁴] The SEB determined that OSI's performance on the CPCS placed it in the second tier--the lowest five rankings of current PCAs invited to compete. The SEB stated that the information obtained from OSI's references indicated that its performance on other contracts was very good to outstanding, but that the firm's performance on the current contract could not be discounted. The SEB determined that OSI's proposal merited an overall rating of a moderate probability of success.

The SEB submitted its award recommendation memorandum to the contracting officer, who served as the source selection authority (SSA). The memorandum divided the offerors into six groups. First, six incumbent offerors were recommended for award: Pioneer Credit Recovery, Inc., Van Ru Credit Corporation, National Asset Management Enterprises, Inc. (NAM), Diversified Collection Services, Inc. (DCS), Financial Asset Management Systems, Inc. (FAM), and Aman Collection Services, Inc. The SEB stated that these firms were the top six performers based on CPCS rankings which, when combined with their D&B and reference ratings, gave them high past performance ratings; they also received high technical ratings. Second, three non-incumbent offerors were recommended for award: USA Education Group, Recovery Bureau of America (RBA), and Maximus. The SEB stated that while no CPCS rankings were available for these firms, they had very good to outstanding past performance based upon their D&B and reference ratings and received high technical ratings. Third, two incumbent offerors were recommended for award: NCO Financial Systems, Inc., and Nationwide Credit, Inc. (NCI). The SEB stated that, while these firms were clearly ranked below the nine recommended firms, their overall good past performance rating and high technical rating made both reasonable, although marginal, candidates for selection; their significant commitments to small business made it worth recommending both for award. Fourth, two small business firms were recommended for award under the small business set-aside. Fifth, five incumbent offerors were not recommended for award, including OSI. The SEB stated that these firms had lower CPCS scores than the incumbents recommended for award and that these lower CPCS scores, when considered with the D&B and reference scores, combined to produce significantly lower past performance rankings for these firms than for the incumbents recommended for award. [⁵] The memorandum also shows that the SEB considered the most recent--eighth--set of CPCS scores for some of the incumbent offerors to ascertain upward or downward trends, but determined that inclusion of this set of scores would not have affected its award recommendations.

The SSA concurred with the SEB's recommendations. The final relevant evaluation results for the full and open competition, with the proposed awardees listed in italics, were as follows:

The SSA independently examined the eighth set of CPCS scores to see whether factoring them in would have an effect on the award decisions and concluded that there would be no such effect. [⁶] He also used a formula and a weighted average to combine the CPCS total scores, the D&B scores, and the reference scores into a single past performance score for each offeror. He concluded that the results of this combined past performance score supported the SEB's findings.

OSI filed its initial protest in our Office after its debriefing. The agency subsequently determined that performance of the task order contracts was in the best interest of the United States and that "circumstances exist[ed] which would not permit waiting" for our decision, and executed an override of the statutory stay of performance of these contracts. See 31 U.S.C. § 3553(d)(3)(C)(i) (1994). OSI supplemented its protest after receiving the agency report. OSI alleges that the agency improperly limited its consideration of performance data under the current contract to the CPCS scores, in contravention of the solicitation's terms, and that the agency's reliance on the CPCS scores alone was also unreasonable. OSI also alleges that the agency's selection of the non-incumbent firms' proposals in preference to its own lacked a reasonably supported basis.

DISCUSSION

Under the Federal Supply Schedule (FSS) program, agencies are not required to conduct a competition before using their business judgment in determining whether ordering supplies or services from an FSS vendor represents the best value and meets the agency's needs at the lowest overall cost. Federal Acquisition Regulation § 8.404(a); Amdahl Corp., B-281255, Dec. 28, 1998, 98-2 CPD ¶ 161 at 3. However, when the agency decides to conduct a formal competition for award of a task order contract, we will review the agency's actions to ensure that the evaluation was reasonable and consistent with the terms of the solicitation. Computer Prods., Inc., B-284702, May 24, 2000, 2000 CPD ¶ 95 at 4-5; COMARK Fed. Sys., B-278343;

B-278343.2, Jan. 20, 1998, 98-1 CPD ¶ 34 at 4-5. Our review of the record here shows that the agency's past performance evaluation was both inconsistent with the solicitation's terms and unreasonable.

Initial Protest

The principal issue raised by OSI's initial protest concerns the agency's near-exclusive reliance on the summary CPCS scores. The agency correctly asserts that it is vested with the discretion to determine the type of past performance data from among the various data on hand which was most relevant and reliable in conducting the past performance evaluation. Evaluation of an offeror's past performance is a matter within the discretion of the contracting agency, and we will not substitute our judgment for reasonably based past performance ratings. However, we will question such conclusions where they are not reasonably based or are undocumented. Green Valley Transp., Inc., B-285283, Aug. 9, 2000, 2000 CPD ¶ 133 at 4; see also Future-Tec Management Sys., Inc.; Computer & Hi-Tech Management, Inc., B-283793.5, B-283793.6, Mar. 20, 2000, 2000 CPD ¶ 59 at 7. The critical question here is whether the evaluation was conducted fairly, reasonably, and in accordance with the stated evaluation terms, and whether it is based upon relevant information sufficient to make a reasonable determination of the offeror's overall past performance rating, including relevant information close at hand or known by the contracting personnel awarding the contract. See U.S. Tech. Corp., B-278584, Feb. 17, 1998, 98-1 CPD ¶ 78 at 6; International Bus. Sys., Inc., B-275554, Mar. 3, 1997, 97-1 CPD ¶ 114 at 5.

Since the CPCS scores are central to this protest, we set out their basis in further detail here. As discussed above, the CPCS evaluation, performed every 4 months, measures the relative performance of each PCA contractor on all accounts transferred. The current contract provides that the CPCS evaluation will gauge the performance of each contractor under four performance indicators. [⁷] The most important indicator, net back recovery, is the net amount of money recovered by the government when defaulted loans are placed in repayment. The indicator is measured as a ratio of the net dollars collected over the average current inventory balance--the average dollar value of the accounts--held by the contractor. The remaining three indicators are, respectively, the number of accounts placed for administrative wage garnishments, the number of litigation packages prepared, and the number of accounts prepared and returned for non-cash, administrative resolution. Each of these indicators is measured as a ratio of the number of accounts resolved (by administrative wage garnishment, litigation, or administrative resolution, respectively) over the average current inventory of accounts--the average number of accounts--held by the contractor. For each 4-month period, the contractor rated highest under each indicator receives the maximum number of points assigned to that indicator and the remaining contractors receive points in proportion to their standing relative to the leading contractor. [⁸] Each contractor's periodic CPCS score is the sum of its scores for all four performance indicators for that period. To obtain the overall ("cumulative") score, the agency simply added together each contractor's scores for each period.

The record shows that these cumulative scores were virtually the only performance data the agency used in evaluating the performance under the current contracts by the offerors who are incumbent contractors. The individual evaluation documents for each firm contain virtually identical narrative regarding the CPCS portion of the evaluation, and little of that narrative concerns the actual quality of the individual offeror's past performance. Instead, the narrative assigns each contractor to a first or second tier based upon its CPCS scores; characterizes the 8 "first tier" offerors as "consistent top performer[s];" and characterizes the 5 "second tier" offerors as having overall performance of average or below. The SEB's award recommendation memorandum repeats these general statements, merely adding notations regarding whether some firms' most recent CPCS rankings trended upward or downward and whether this had an impact on their overall CPCS rankings. The SSA's decision document adopts the SEB's recommendations without further substantive analysis. [⁹]

In his post-protest affidavit, the SEB chairperson states that the SEB "knew how the CPCS rankings were derived and . . . the meaning ‘behind the numbers,'" considered the data generated by the CPCS system, and considered the CPCS data for each of eight ranking periods. SEB Chairperson's Affidavit ¶¶ 3, 6. The contemporaneous evaluation documentation does not support these statements. Rather, the contemporaneous evaluation documentation shows that the CPCS aspect of the agency's past performance evaluation contained virtually no analysis of individual offerors' past performance, and that the agency limited its consideration of the performance data on hand to ranking the incumbents based upon the arithmetic total of each firm's seven periodic CPCS scores and reviewing whether some firms' eighth periodic CPCS score reflected an upward or downward trend in performance. To the extent the agency performed any qualitative analysis, it is not documented. For our Office to perform a meaningful review of an agency's selection determination, however, the agency must have adequate documentation to support its evaluation of proposals and its selection decision. Green Valley Transp., Inc., supra, at 8; Future-Tec Management Sys., Inc.; Computer & Hi-Tech Management, Inc., supra.

Here, we are faced with a lack of contemporaneous evidence that the agency considered performance data relevant to the offerors' performance under the current contracts other than the cumulative CPCS numbers, while the solicitation committed to offerors that, "[f]or those companies with a current contract, the Department will use performance data that we have on hand such as the CPCS scores." It is true that the solicitation stated that competitive ranking information (such as the CPCS scores) was extremely relevant, but the agency had other relevant performance data it could have considered. Given the relevance of that data and the limitations of the cumulative CPCS scores, we agree with the protester that the agency's near-exclusive reliance on the cumulative CPCS scores was unreasonable.

Our principal concern stems from the agency's overly mechanical application of the cumulative CPCS scores to evaluate the past performance of incumbent contractors. We recognize that the agency constructed the CPCS system in a way intended to capture various aspects of contractors' performance through the CPCS scores. Point scores can, however, only be aids in decision-making, and they must be used in a defensible way. In this case, the nature of the underlying data meant that the cumulative CPCS scores could not be relied on mechanically to assess offerors' past performance.

The CPCS methodology set forth in the current contracts envisioned not a cumulative evaluation over the life of the contract, but a series of discrete periodic evaluations of performance. In defending the use of cumulative scores, the SSA states that the agency needs PCAs that will perform well consistently over the life of the contract, not just during a 4-month period. SSA's Affidavit ¶ 6.

We agree that considering one CPCS period in isolation would not give an actual picture of the overall performance of the contractors overtime. In fact, review of the CPCS scores for the individual periods demonstrates another flaw in the agency's reliance on cumulative scores. The data shows that the performance of most PCA contractors, including those in the SEB's first tier, varied, often widely, from period to period. For example, the rankings of Pioneer over the eight CPCS periods were 9th, 7th, 1st, 1st, 10th, 11th, 2nd and 1st; the rankings of FAM were 3rd, 11th, 10th, 13th, 11th, 2nd, 4th and 10th; and the rankings of Aman were 5th, 6th, 14th, 15th, 5th, 1st, 6th and 14th. [¹⁰] These disparities demonstrate the unreasonableness of the SEB's characterization of these first tier contractors as "consistent top performers," and reinforce the protester's point that an examination of each of the CPCS periodic scores, and the underlying data, should have been a component of the past performance evaluation.

That point is given further support when the impact of shifting account inventories on CPCS scores is considered. Contractors' periodic CPCS scores were based upon variable numbers of accounts, and those numbers also varied, often dramatically, from period to period. It was not unusual for the average number of accounts held by contractors to vary by more than 100,000 within a CPCS period, or for the average number of accounts held by one contractor to fluctuate by as many as 50,000 accounts from period to period.

In this regard, contractors began the first CPCS period on relatively equal footing: each contractor was transferred 20,000 accounts with roughly the same average balances. However, one purpose of the CPCS evaluation was to determine the number of accounts contractors might receive in the next CPCS period; the higher a contractor's CPCS score in one period, the more accounts might be subsequently transferred to that contractor. The agency acknowledged during a bidder's conference that this mechanism meant that it could be difficult for top performers to remain on top because they might have large transfers of accounts during the next period. First Bidder's Conference Transcript at 62-63. Conversely, accounts that remained unresolved after 9 months were required to be returned to the agency for reassignment to other contractors, and contractors could return unresolved accounts earlier than that provided they had engaged in a minimal level of effort to resolve the accounts. This minimal level of effort was defined differently in each PCA contractor's contract, which adds a further difficulty in comparing contractors' scores. As the agency acknowledged, contractors could "manipulate" the size of their inventory to move up in the rankings. Id. at 62.

The record contains additional evidence of a relationship between the average number of accounts held by a contractor and its CPCS scores. Our review of the data shows a frequent correlation between a decreased volume of accounts and an increased percentage of account resolutions, and vice versa. For example, between the 5th and 6th CPCS periods, Pioneer's average number of accounts increased by approximately 37,000 and all of its resolution rates decreased. Between the 6th and 7th CPCS periods, the firm's average number of accounts decreased by approximately 76,000 and all of its resolution rates increased substantially. Again, the CPCS methodology provides that higher account resolution rates generally result in higher CPCS scores, and lower account resolution rates generally result in lower CPCS scores.

The protester does not contend, as the agency states, that the CPCS rankings are unfair because there are differences in the number of accounts transferred to contractors during a period or that the SEB should have adjusted or disregarded CPCS scores as a result. Rather, the protester asserts, and we agree, that the differing workloads of contractors ought to have been considered when assessing their actual past performance. As an example of what might be considered, the protester points out that its having received a large number of accounts in the second CPCS period as a result of its success in the first CPCS period presented particular challenges associated with the need to provide staffing early in performance to handle those newly assigned accounts.

While the indicators used in calculating CPCS scores for a specific period factored in the differences in inventory size for that period, we conclude that it was unreasonable for the agency to rely as exclusively as it did on the cumulative CPSC scores. Given the significantly different account inventories held by the contractors both within one CPCS period and from period to period, we view it as irrational to focus only on the CPCS scores without examining the circumstances and the performance data in the agency's possession to reach a considered judgment regarding the quality of an offeror's performance and to make comparisons between offerors' performance. [¹¹] Green Valley Transp., Inc., supra, at 6-7. We recognize that reducing the past performance of contractors to a single score might result in a more streamlined evaluation, but the use of such a technique is no substitute for the reasoned judgment of evaluators in examining and comparing the actual past performance of offerors. [¹²]

Supplemental Protest

In its supplemental protest, OSI contends that the agency awarded at least one task order contract to a non-incumbent offeror whose evaluation was, in every comparable category, inferior to OSI's. Specifically, the protester correctly asserts that the proposals of both OSI and Maximus received high technical ratings; that OSI had far more years of experience than Maximus; and that OSI was rated higher than Maximus under the key personnel component of the technical evaluation factor, the D&B ratings, and the reference ratings. The only factor the firms did not have in common was incumbent performance. OSI was an incumbent contractor with CPCS scores and Maximus had no experience with PCA contracts. OSI contends that its incumbent performance should have been a positive factor in its favor but that, notwithstanding its proposal's status as equal or superior to that of Maximus, the record contains no indication why Maximus was selected over OSI. [¹³]

The agency's only response to this allegation is to repeat the information contained in the individual evaluations of these offerors and to assert that OSI is merely disagreeing with those individual evaluations. However, OSI is not challenging these individual evaluations (save its own CPCS evaluation) but, rather, is correctly asserting that the record is devoid of any comparison between OSI's proposal and the proposals of the non-incumbent awardees, notwithstanding the fact that the evaluation documentation shows that its proposal is equal or superior to at least one of these proposals. Again, an agency that fails to adequately document its source selection decision in sufficient detail to show that it is not arbitrary bears the risk that our Office will not conclude that the agency had a reasonable basis for its determinations. Future-Tec Management Sys., Inc.; Computer & Hi-Tech Management, Inc., supra, at 7. The record here leads us to conclude that the source selection in this regard was unsupported and unreasonable.

Prejudice and Recommendation

Our Office will not sustain a protest unless the protester demonstrates a reasonable possibility that it was prejudiced by the agency's actions, that is, unless the protester demonstrates that, but for the agency's actions, it would have had a substantial chance of receiving the award. McDonald-Bradley, B-270126, Feb. 8, 1996, 96-1 CPD ¶ 54 at 3; see Statistica, Inc. v. Christopher, 102 F.3d 1577, 1581 (Fed. Cir. 1996). We believe that OSI has met this standard.

It is undisputed that more weight was to be accorded more recent competitive ranking data and that OSI's most recent CPCS ranking trended downward, and it is undisputed that OSI's performance was erratic over the course of the contract based upon its CPCS rankings. However, the agency's reliance on these rankings to argue that OSI has not been prejudiced perpetuates the central flaw in this past performance evaluation: the agency's failure to review relevant performance data in its possession and to make reasoned judgments as to its value to arrive at an accurate assessment of each offeror's past performance. Since the agency's past performance evaluation did not comply with the stated evaluation terms; the CPCS scores alone are an inadequate indicator of overall past performance without consideration of the circumstances and underlying performance data; the agency has not adequately supported its past performance evaluation of offerors, both on an individual and a relative basis; and OSI has otherwise very high scores and a positive evaluation, we conclude that the source selection cannot be viewed as reasonable and that the protester was prejudiced by the agency's actions.

We recommend that the agency reevaluate the proposals with respect to past performance, giving appropriate consideration and weight to the performance data in its possession. If the reevaluation shows that the previously made award decisions should be revised, the agency should terminate the appropriate task order contracts and make award to OSI or other appropriate offerors. We also recommend that the protester be reimbursed the reasonable costs of filing and pursuing the

protests, including attorneys' fees. 4 C.F.R. § 21.8(d)(1) (2000). The protester should

submit its certified claim for such costs, detailing the time expended and the costs incurred, directly to the contracting agency within 60 days after receipt of this decision.

The protests are sustained.

Anthony H. Gamboa

Acting General Counsel

Notes

1. The agency evaluated each technical proposal as representing a low, moderate, or high probability of success, and evaluated small business participation proposals based upon how close offerors came to the RFTOP's specified targets.

2. OSI argues that the agency improperly failed to secure a response from its third reference, but there is generally no legal requirement that all references listed in a proposal be checked. Advanced Data Concepts, Inc., B-277801.4, June 1, 1998, 98-1 CPD ¶ 145 at 10.

3. OSI's proposal also met the RFTOP's small business participation and pricing targets.

4. "Net back recovery percentage," the most important CPCS performance indicator under the current contract, measures the net proportion of dollars collected from the contractor's inventory of accounts.

5. The sixth group was comprised of eight non-incumbent offerors not recommended for award.

6. The SSA also experimented with the CPCS data using weighted average formulas to give the most recent past performance more weight, in accordance with the RFTOP instructions. He concluded there would be an effect on the award recommendations only if he used a highly skewed weighting system, which he deemed unreasonable.

7. The current contract initially included two additional performance indicators that were eliminated by contract amendments in 1998.

8. The effect of this methodology was that a contractor's score reflected not simply its own performance, but also how well other contractors performed during the same period. Thus, a contractor could receive different scores for the same level of performance in two different periods.

9. The SSA states that the single past performance score he put together from the D&B scores, reference scores, and CPCS scores was used only as part of his post-recommendation analysis to confirm that the SEB's recommendations were appropriate. Regardless of its purpose, for the reasons discussed below in the context of the CPCS scores, the SSA's use of this technique was overly mechanical and the formula he used to arrive at his single past performance score suffered from several errors pointed out by the protester.

10. These ratings were generally based on a universe of 17 PCA contractors.

11. Since the CPCS performance indicators contemplated by this RFTOP differ from the indicators previously used, OSI contends that the agency should have considered whether certain aspects of performance should be weighted more than others in order to more accurately predict future good performance. Our review of the two sets of indicators and their methodology shows that they are not materially different, but the agency may wish to consider whether it deems certain aspects of past performance more valuable than others in reevaluating proposals.

12. The RFTOP also required the agency to evaluate "key personnel" for all offerors under both the past performance and technical factors. The record shows the SEB considered the experience of key personnel as part of the technical evaluation of all proposals, but did not consider this experience in the past performance evaluation where CPCS data was available. The agency states that if the key personnel proposed were also involved in the past performance of a PCA contract for that offeror, their relative success or failure in the performance of that contract would be reflected in the CPCS scores. To the extent the agency failed to evaluate key personnel in the past performance evaluation of all offerors, that was inconsistent with the RFTOP's terms. To the extent the agency factored key personnel into the past performance evaluations of incumbent offerors, it is not apparent from the evaluation documentation and should be addressed in the reevaluation.

13. OSI makes these same arguments with respect to the two other non-incumbent awardees. In addition, in its supplemental protest, OSI argues that the rating given to one of these firms for its references was suspect because two of the three references appeared to be from related firms. The agency's supplemental report fails to address this allegation, which appears to be well-founded, and the agency should review the matter as part of its reevaluation of past performance.