More than four years ago, we remanded this case with the instruction that appellant Sun Life Assurance Co. reconsider its rejection of Diahann Gross’s claim for disability benefits based on chronic and severe pain. See Gross v. Sun-Life Assurance Co. of Can. (Gross I),'734 F.3d 1, 4 (1st Cir. 2013). Although we found at that time that Gross’s medical evidence supported a finding of total disability, we concluded that, “[a]s the record now stands, we are unable to resolve the debate between the parties on the significance of ... surveillance evidence” obtained by Sun Life. Id. at 27. After additional administrative proceedings, the company again denied her claim and Gross again challenged the denial in federal court. On cross-motions for summary judgment, the- district court ruled in Gross’s favor, finding that Sun Life should have awarded Gross benefits because the surveillance record as developed does not undermine this court’s prior assessment of the medical evidence.

In this appellate sequel, Sun Life challenges the district court’s view of the expanded administrative record. It argues that Gross failed to adduce medical evidence in the renewed proceedings to offset the contradictory surveillance—and thus did not meet her burden to prove that she is totally disabled. Sun Life also claims the district court abused its discretion in failing to impose sanctions on one of Gross’s attorneys. In a cross-appeal, Gross assigns error to the district court’s calculations of prejudgment interest and attorney’s fees.

After careful review of the record and the law, we affirm the district court’s rulings on the disability claim and sanctions. However, we vacate the prejudgment interest award and remand for consideration of the appropriate rate of interest. We affirm the district court’s attorney’s fee calculation in all but two respects, concluding that two components of the award must be increased.

I. Background

A. The First Appeal

" Until she was placed on disability leave in August 2006, at age 34, Gross worked as an optician and office manager at Pinnacle Eye Care LLC in Lexington, Kentucky. In our prior decision, we described in great detail the facts then in the administrative record concerning Gross’s condition and medical evaluations. See id. at 17-21. Here, we begin with a summary description of *5the original record and briefly review that prior decision to remand. We then describe the new evidence obtained in the second round of administrative proceedings. We elaborate below on both sets of facts where pertinent to our analysis. .

1. The Original Medical Evidence

Multiple medical professionals who examined Gross between 2005 and 2007 reported that she was experiencing a variety of debilitating symptoms, including “chronic pain, inability to sit or stand for extended periods of time, severely diminished functional capacity in her right arm, and inability to bend, kneel, or crouch.” Gross v.. Sun Life Assurance Co. of Can. (Gross Remand Op.), No. 09-11678-RWZ, slip op. at 2 (D. Mass. June 24, 2016). Gross’s treating physician, Dr. Rita Egan, a rheu-matologist, opined that Gross was incapable of performing even sedentary activity, and she concluded that Gross suffered from reflex sympathetic dystrophy (“RSD”), fibromyalgia, widespread pain, and chronic fatigue. Gross I, 784 F.3d at 17. In two reports completed in late 2006, Egan noted, with some variation between the statements, that Gross could not sit in one place for more than an hour to ninety minutes, drive for more than ninety minutes, use her right hand, or lift more than ten pounds. Id. at 17 & n.19.1

Other practitioners echoed Egan’s diagnoses, noting, inter alia, abnormalities in the appearance of, and the way Gross positioned, her right hand. See id. at 18, 21, 22.2 A physical therapist who performed a functional capacity evaluation (“FCE”) in early 2007 reported “a number of ‘key limitations’ in Gross’s physical abilities, including lack of functional use of her right arm, poor standing balance, inability to perform sustained overhead activity, need for assistance or a handrail to negotiate stairs, and inability to crouch, kneel, squat or crawl.” Id. at 18-19. The physical therapist, Chris Kaczmarek, suspected that she suffered from RSD or an equivalent condition known as complex regional pain syndrome (“CRPS”), or fibromyalgia. Id. at 18. The FCE concluded that Gross “does not present at a functional level that could maintain sustained work activity.” Id. at 19.

Significantly, Kaczmarek stated that Gross was cooperative and “willing to work to maximum abilities” when performing tasks for the FCE, Id. at 23. Additional evidence of Gross’s good-faith in describing her symptoms and limitations came from her co-workers and employers, who submitted letters “describing her persistence in continuing to work despite obvious pain and compromised physical capacity.” Id. Her boss observed that “[s]he wasn’t going to give in until she absolutely had to,” id at 23 n.29, and Pinnacle’s owner, Paul Wedge, stated that “[w]e stopped her from working when we received her doctor orders that she was not fit to work,” id at 23 (alteration in original).

The medical evidence, however, did not uniformly support Gross’s disability claim. All of her diagnostic tests were negative, and several doctors speculated that psychological factors might be contributing to *6the severity of her symptoms. Id. at 24. Despite recommendations from multiple physicians that she obtain counseling • or behavioral treatment, she never did so. Id. Most puzzling was the evidence resulting from an investigator’s surveillance of Gross on nine days between November 2006 and February 2007. On most of those days, the surveillance .revealed little activity by. Gross, including multiple days when she either did not leave the house or was out briefly in unremarkable circumstances. Our-prior decision highlighted three exceptions: .

[O]n November 9, 2006, shortly after dropping off a teenager believed to be her stepdaughter at school, Gross was observed driving for about an hour and a half to her mother’s home, with a brief stop at a rest area along the way. Second, during the evening of January 11, 2007, Gross drove a short distance with her stepdaughter to a Kmart, where she was observed bending down toward lower-level shelves, extending her arms above her head to retrieve items, and kneeling to examine other items.[3] Third, on February 21, after receiving a phone call that her mother had been admitted to the hospital with chest pain, Gross drove to a gas station, pumped gas using her right hand, and then drove for two hours to the hospital, with a brief stop halfway through the trip. About two hours later, she left the hospital and drove home.

Id. at 19. •

In support of its original denial of benefits, Sun Life also had procured opinions from two medical consultants who conducted paper reviews of Gross’s medical records. In the first records review, Dr. James Sarni noted, that “the documentation does not strongly support a diagnosis of [RSD or CRPS].” Id. at 19 n.24. He suggested an evaluation by a neurologist, which Dr. Rukmaiah • Bhupalam subsequently performed on February 22, 2007, the day after Gross' had made the trip to the hospital. Although Bhupalam initially concluded that Gross was “totally disabled even for sedentary work on a part time basis,” he changed- his assessment after viewing the surveillance videotapes. Id. at 20. He observed that “she can function quite well and probably will be able to return to her previous occupation,” although he also noted that “a re-evaluation might'be beneficial.” Id.

After Bhupalam’s examination, the second non-examining consultant, Dr. William Hall, reviewed Gross’s medical records and concluded that “the surveillance videos undermined [her] subjective reports of pain and functional limitations.” Id; A third consultant performed a paper review after Gross appealed the initial benefits denial. That physician; Dr. Alan Neuren, noted the inconsistencies between Gross’s condition as reported by healthcare providers and her appearance under surveillance, and he stated that “‘[t]he only reasonable conclusion’ to be drawn ⅛ that she has deliberately embellished her symptoms to her providers for secondary gain.’ ” Id. at 21.

2. The Remand Rationale and Directive

Given the well documented history of pain and other symptoms recorded by the medical professionals who examined her, and the buttressing observations of her coworkers, we had “no difficulty” concluding that Gross had submitted adequate medical evidence to prove her entitlement to *7disability benefits. Id, at 22.4 .We pointed out that, even though many of Gross’s complaints were not readily susceptible to objective confirmation, the record did contain some objective evidence, “as well as the recognition by Sun Life’s own medical consultant, Dr. Hall, that Gross’s ‘musculo-skeletal symptoms, as presented by her, are credible to treating and consulting physicians.’ ” Id.

We were concerned, however, about the “significant incompatibilities between Gross’s reports and her observed functional capacity” while under surveillance, particularly during the three episodes described above. Id. at 26. Yet, even faced with those contradictions, Dr. Bhupalam had suggested that a reevaluation of Gross could be helpful that’s—“an observation we understood] to suggest that the video surveillance, while damaging to Gross, did not necessarily undermine her claim.” Id. We also noted that the record did not reveal whether Bhupalam or Neuren had been told that Gross’s two-hour drive to a hospital in February 2007 was precipitated by news that her mother had suffered a medical emergency. Id. at 26-27. That seeming omission of context led us to question “whether Sun Life ha[d] -made a bona fide effort to determine Gross’s capabilities.” Id. at 27. At the same time, we noted the absence of a statement from Gross’s own doctor “refuting Sun Life’s assertion in its original denial letter that the surveillance ‘show[ed] a capacity for activity that far exceeds’ the limitations she claims.” Id, (alteration in original).5

We thus concluded that, ori the record then before us, we could not answer the “open question” necessary to resolve the parties’ debate over Gross’s entitlement to benefits: “the effect that the surveillance evidence, when viewed in context, may haye on other evidence indicating disability.” Id. at 27. Accordingly, we remanded the case, “so that the parties can further address both the significance of the video evidence in assessing Gross’s limitations and the.veracity of her self-reported and observed symptoms, ¡particularly concerning the condition of her right arm.” Id. at 27-28.

B. The Evidence Produced on Remand

In the renewed administrative proceedings following remand, Gross and Sun Life each submitted additional' opinions from two medical professionals, none of which were based on new examinations of Gross. Sun Life relied primarily oh reviews of Gross’s medical file by two neurologists, Drs. David Ross and Rajat Gupta. Ross provided a nine-page report summarizing Gross’s medical • history and concluding that “[t]he medical evidence does not support a functional impairment as of August 1, 2006”—Gross’s claimed disability date. Ross stated that “[t]here is no medical *8explanation for the discrepancy between the reported limitations and those seen during surveillance,” and he opined that Gross’s “observed activities are more consistent with her true functional status.”

Gupta similarly prepared a report reviewing Gross’s medical history, beginning in March 2004, and responded in the negative to a question asking whether he detected “any physical condition(s) supported by the clinical evidence that are functionally impairing.” Gupta questioned the diagnoses of CRPS or RSD, noting that the symptoms on which those assessments were based—including the swelling, discoloration, and temperature of Gross’s right arm and hand—“are known to be occasionally self-induced by particularly savvy individuals.” Gupta highlighted Gross’s long-sleeved clothing seen in the videos, “which typically would be avoided by sufferers of CRPS due to the extreme amount of hypersensitivity typically present,” and he noted the “consensus among most of her providers that there is a psychological component to [her] presentation.”

In an addendum to his report, Gupta stated that Gross’s mother’s medical emergency “would not explain the apparent ease and fluidity of movement that the claimant nonchalantly and effortlessly displayed]” with the use of her right arm and hand as she prepared to drive to the hospital in February 2007—although he acknowledged that “[pressing circumstances may conceivably allow an individual to perform physical feats of strength and/or endurance that would otherwise be considered ‘unachievable.’” On that issue, Sun Life also obtained a follow-up opinion from Dr. Neuren, who stated that “[i]t is not credible that going to visit her mother due to illness would result in resolution of her condition even on a temporary basis. ... CRPS is not a part time condition.”

Gross’s additional medical evidence consisted of two letters, one from a pain management specialist, Dr. James Murphy, and one from the physical therapist who had performed her functional capacity evaluation in 2007, Chris Kaczmarek. Having reviewed Gross’s medical records and the surveillance evidence, Murphy concluded that nothing in the three noted surveillance reports and videotapes “would contradict or invalidate the restrictions and limitations placed upon Ms. Gross by her physicians.” Murphy stated that the physical effects of CRPS and fibromyalgia “can vary from day to day—even minute to minute,” and that the severity of symptoms “are dependent upon numerous factors, such as medication regimen,[6] response to interventions ..., physical stress, systemic illness, and underlying precipitating condition(s).”

Kaczmarek submitted a two-page letter reporting that he had reviewed Gross’s records, the three identified surveillance reports and videos, and the FCE he had performed in 2007. He summarily concluded that Gross’s activities in the videos were consistent with his prior findings that she could neither sit “at a frequency sufficient to engage in sedentary employment” nor “ ‘exert up to 10 pounds of force’ on an ‘occasional’ basis sufficient to engage in sedentary employment.”7

*9C. The District Court’s Post-Remand Decision

After briefly surveying the evidence described above, the district court observed that its task was to decide whether the surveillance evidence “casts doubt ... sufficient to dislodge” the panel’s conclusion that the medical evidence supported Gross’s claim of total disability. Gross Remand Op. at 5. To undermine the disability assessment, the court stated, “would require the videos to show Gross performing activities that ‘directly contradict’ the self-reported limitations upon which her treating physicians have offered their diagnoses.” Id. (quoting Gross I, 734 F.3d at 25). The court found no such contradiction, nor “sufficient evidence that Gross otherwise exaggerated her symptoms to hoodwink her treating physicians.” Id.

Among other factors, the court noted that only one professional who had both personally examined Gross and reviewed the surveillance records—Bhupalam—had disagreed with the diagnoses of disabling conditions. But the court discounted Bhupalam’s view because he initially had agreed that Gross was unable to work, and his later contrary opinion—after viewing the surveillance tapes—was “tempered” by his recommendation that a follow-up evaluation be performed. Echoing our own sentiment, the court observed that “[t]his recommendation makes sense only if the surveillance records did not unequivocally contradict Bhupalam’s; initial opinion that Gross was totally disabled.” Gross Remand Op. at 6; see also Gross I, 734 F.3d at 26. The district court therefore ordered Sun Life to pay Gross benefits based on a disability date of January 30, 2007.

In a subsequent ruling, the district court determined that the applicable interest rate for Gross’s recovery of benefits is the rate set by 28 U.S.C. § 1961.8 Previously, the court had awarded Gross approximately $96,000 in attorney’s fees for work performed in connection with the proceedings leading up to, and including, the first appeal to this court. The fees determination followed this court’s holding that Gross had achieved a sufficient dégree of success on the merits in Gross I to qualify her for fees under ERISA. Gross v. Sun Life Assurance Co. of Can. (Gross II), 763 F.3d 73, 81 (1st Cir. 2014). Significantly, our decision in Gross I changed the standard of review for claims denials under policies requiring proof of disability “satisfactory” to the benefits decision-maker. See Gross II, 763 F.3d at 75-76; Gross I, 734 F.3d at 16. That ruling, which replaced the deferential arbitrary-and-capricious standard with de novo review, “strengthened] the entitlement to benefits for employees cov*10ered by such policies.” Gross II, 763 F.3d at 85.9

Both parties' have 'appealed from the district court’s judgment. In its briefs to us, Sun Life contends that the record does not support the court’s conclusion that Gross met her' burden to show -that she is totally disabled. The insurer also claims that the district court erred in failing to sanction one of Gross’s attorneys for threatening, to sue Bhupalam if he did not withdraw his revised opinion adverse,to Gross. In her cross-appeal, Gross argues that the district court abused its discretion in choosing a prejudgment interest rate that does not fully compensate her for the wrongful denial of benefits. She also asserts that the court abused its discretion in setting the amount of attorney’s fees for the pre-remand proceedings, which concluded with our decision in Gross II.

II. Total Disability Finding

A. Standard of Review

Our decision in Gross I established that Sun Life’s denial of benefits was subject to de novo review by the district court. See 734 F,3d at 16. We recently observéd, however, that the proper standard of appellate review is debatable in a case such as this— i,e., “in an ERISA benefít-denial case that is presented for decision exclusively on the record of proceedings before the plan administrator.” Stephanie C. v. Blue Cross Blue Shield of Mass. HMO Blue, Inc., 852 F.3d 105, 111-12 (1st Cir. 2017). Noting that the district court’s de novo review of the administrative' record may involve “weighting] the facts, resolv[ing] conflicts in the evidence, and drawing] reasonable inferences,” we acknowledged a plausible argument for applying the deferential clear-error standard .on appeal to the extent the district court’s decision “rests upon factual findings and inferences therefrom.” Id.

We need not reach that issue here, however. Not only do both parties assume that our review is de novo, but application of that., standard-r-more favorable to Sun Life—nonetheless leads us to uphold the district court’s judgment. We therefore review the administrative record de novo without affording deference to the district court’s assessment of the record,

B. Discussion

1. Evaluating the Post-Remand Evidence

Sun' Life insists that the record as supplemented on' remand reinforces its original determination that Gross did not prove that she is totally disabled. In sum, it contends that Gross “failed to produce any evidence related to- the[ ] ‘open questions’ ” that prompted our remand, namely, “the significance of the video evidence in assessing Gross’s limitations and the veracity of her self-reported and observed symptoms.” Gross I, 734 F.3d at 27-28. Accordingly, Sun Life asserts, its own “overwhelming evidence” that the surveillance undermines Gross’s disability claim is unrefuted and, hence, Gross failed to sustain her burdefi- of proof.

We see the record differently. Sun Life both overstates the persuasive value of its own post-remand submissions and sidesteps the fact that we previously found Gross’s medical evidence sufficient to prove her entitlement to benefits. See id. at 24-25 (“In sum, the sustained and progressive nature of Gross’s complaints, their facial credibility to the medical prac*11titioners who personally examined her, and the objective symptoms consistent with RSD ... support a finding of disability.”). In. seeking further development of the record, our objective was to learn whether, in light of the surveillance, there was reason to discredit the medical evidence we found adequate to prove total disability. See id. at 27 (describing “the open question” for remand as “the effect that the surveillance evidence, when viewed in context, [has] on other evidence indicating disability”). That is, we found that Gross had met her burden to show her total disability; but we sought additional information to help us determine if the surveillance evidence put forth by Sun Life was sufficiently probative to undermine Gross’s medical evidence. Regrettably, neither party took full advantage of the opportunity to reinforce its position. As we shall explain, Sun Life must bear the burden of that deficiency because—as the district court held—the insurer failed to show that the surveillance “casts doubt ... sufficient to dislodge” our judgment that the medical record supports Gross’s disability claim. Gross Remand Op. at 5.10

For its part, Sun Life did not have Gross reevaluated and, instead, secured opinions based on reviews of her existing records and the surveillance. These reports primarily reiterated what we already knew: Gross engaged in some activities that were inconsistent with the most severe symptoms and limitations she described to her doctors during years of treatment for pain, as well as with the severity and persistence of pain typically associated with a diagnosis of CRPS or RSD. The central theme of the new evidence, as Ross put it, is that “[t]he medical evidence does not support a functional impairment,” and, accordingly, “[t]he claimant’s observed activities are more consistent with her true functional status.”

Yet, “[t]he medical evidence” already was before us during Gross’s, first appeal, and we found adequate record support for her self-reported limitations to conclude, subject to further insight into the surveillance-evidence, that Gross had shown an entitlement to benefits. Moreover,-as noted above, Egan’s reports in the fall of 2006 set the outside range of Gross’s .abilities, as sitting in one place for two hours, driving for ninety minutes, standing, or walking for an hour, and lifting ten pounds. Gross I, 734 F.3d at 17 & n.19. Even the “particularly troubling” activities .surrounding the hospital visit in early 2007 were not far removed from those limits, id. at 26, and we observed in our prior decision that, “[i]n context, the extra driving, the hurried movements, the pumping of gas may have been at the far edge of what she could manage with the aid of medication in the face of a family crisis,” id at 27. . .

Indeed, Gross told Bhupalam that she can “function better” after changing her pain medication patch, id. at 20,11 and a month before the hospital trip she reported using numerous other medications on a daily’basis, see id. at 19. Given the inference we drew from Bhupalam’s addendum that the surveillance activities were not decisively at odds with a finding of total *12disability, Sun Life needed to show that, to the contrary, the capabilities Gross demonstrated in the videos were incompatible with the medical record of disability.

Ross and Gupta’s assessments, however, failed to evaluate the three highlighted surveillance' reports in the context of the entire surveillance investigation and the consistent perceptions of examining practitioners that her complaints of pain were genuine. In particular, Sun Life’s experts did not explain the contrast between the more ambitious surveillance activities that we highlighted and Gross’s numerous days of relative inactivity,-a noticeable gap in light of her reports that she could obtain temporary relief from pain medications.

As described above, the surveillance took place over nine days, and the investigator saw little activity by Gross on most of those days. Id. The surveillance began with three days in November 2006. On November 7, Gross left home at 7:17 AM for the first of two brief excursions, and she was seen limping when she returned home at 9:27 AM. On November 8, she was out of the house for roughly 30 minutes (3:20 PM to 3:51 PM) when she drove to a shopping center and back. On November 9, the surveillance showed more activity, albeit only for the morning: Gross took her stepdaughter to school, briefly went inside the high school, returned home for a few minutes, and then drove to her mother’s home—a trip that took a bit more than an hour and a half, including a rest stop after an hour. She arrived at 11:05 AM, and the investigator saw no further activity before he left the area at 2:15 PM. During the next surveillance period, in January 2007, Gross did not leave the house at all during two of the three days (January 10 and 12). Even on January 11, the day she was observed at Kmart, she was out only from 6:16 PM to 7:14 PM. During the final surveillance period, in February 2007, described more fully infra, Gross was'either inactive or physically compromised on two of the days (February 22 and 23), and on February 21 she made the hospital visit to her mother. '

Ross and Gupta thus appeared to treat the most extreme surveilled activities as decisive over Gross’s long history of credible-pain, without confronting her inactivity during most of the surveillance. Moreover, Sun Life’s counsel acknowledged at oral argument that there is no record evidence that Ross was told that Gross’s travel on February 21 occurred after she learned that her mother had been taken to the hospital on an emergency basis.12

.We recognize that Gupta’s addendum refers not only to the identified episodes, but also to Gross’s use of her right hand on the three successive days of surveillance in November 2006, including for the purpose of closing a car door and reaching into her purse, and he notes that she “is seen ambulating fluidly, with no limp, in all three days.” He similarly reports that, during the three days of surveillance in January 2007, Gross is seen walking “in a normal manner, using her right hand to brush her hair off of her face ... [and] adjusting] the shoulder strap of her purse with the same hand.” We consider brief actions by Gross, on days when she was largely inactive or also manifested limitations, of minimal significance. Some fluctuation in physical ability related to such factors as fatigue and the timing of medications is predictable, see id. at 26, and assessments of Gross’s activities that fail to account for such variations are neces*13sarily incomplete if not misleading.13 Indeed, the investigator also saw Gross limping on one of the November days that Gupta referenced.

Sun Life emphasizes that it offered ample evidence that the inconsistency between Gross’s complaints and the surveillance indicates that she was either embellishing or self-inducing her symptoms, or both. Neuren’s original report noted both of those possibilities, id. at 25,' and, as described above, Gupta’s post-remand report stated that many of the skin abnormalities associated with CRPS' or RSD that appear in Gross’s medical record can be self-induced. But all of the medical practitioners who actually examined Gross found her credible, and Neu-ren and Gupta’s generalized speculation does not explain how Gross could have deceived so many observers over a substantial period of time14—not only doctors, but her co-workers as well.

Even most of Bhupalam’s addendum, in which he retreats from his original finding of disability, consists of nothing more than what he sees on the videotape. He observes, .for example,-that Gross “does not appear to be in any pain or discomfort in the video recorded on February 21.” Bhupalam does not attempt to explain, however, how to reconcile what can be seen on the videotapes with his in-person evaluation. In all. likelihood, the puzzling dissonance is why he noted that a reevaluation could be helpful.15

• We are frankly puzzled that .Sun Life did not act on Bhupalam’s suggestion of a reexamination,16 given our highlighting of both the believability.of Gross’s symptoms to medical practitioners and her co-workers’ description of; her deteriorating physical condition while she attempted to re-, main on the job. In defending its reliance solely on non-examining physicians, Sun Life emphasizes that ERISA plan decision-, makers “are not obliged to accord special deference to the opinions of treating physicians.” Black & Decker Disability Plan v. Nord, 538 U.S., 822, 825, 123 S.Ct. 1965, 155 L.Ed.2d 1034 (2003). But that is not *14the pertinent principle here. In Black & Decker, the difference of . opinion at issue was between the claimant’s treating physician and an independent medical examiner who also had personally examined the claimant. Id. at 827, 123 S.Ct. 1965, Where the determination of disability depends on an assessment of largely subjective, self-reported symptoms, those who have had in-person exposure—whether treating.physician or not—have access to information unavailable to non-examining doctors.

To be clear, we are not saying as a general matter that the views of examining doctors' are entitled to more weight than the opinion of á doctor who performs only a records review. Indeed, we have held to the contrary. See Orndorf v. Paul Revere Life Ins. Co., 404 F.3d 510, 526 (1st Cir. 2005) (“Denials of benefits may be based on review of medical records submitted by thé claimant.”); see also Richards v. Hewlett-Packard Corp., 592 F.3d 232, 240 (1st Cir. 2010). However, where the claimant’s credibility is a central factor in the disability determination—and particularly where, as here, the claimant’s in-person presentation of symptoms was credited by the independent .medical examiner, Bhupalam—the impressions of examining doctors sensibly may be given more weight than those who looked only at paper records. See, e.g., Kalish v. Liberty Mut./Liberty Life Assurance Co. of Bos., 419 F.3d 501, 510 (6th Cir. 2005) (giving little weight to credibility determination by doctor who did not physically examine claimant “and in contradiction of the ... investigator’s conclusion that plaintiff was ‘very credible’”); cf. United States v. Raddatz, 447 U.S. 667, 679, 100 S.Ct. 2406, 65 L.Ed.2d 424 (1980) (noting that “courts must always be sensitive to the problems of making credibility determinations on .the cold record”). We thus find minimal new insight in Sun.Life’s post-remand submissions, which fail to provide a “contextualized assessment, of the most , significant departures from her professed limitations.” Gross I, 734 F.3d at 27 (emphasis added).

Yet, Gross’s offerings, too, are less compelling than we would have anticipated. Neither of her two post-remand reports was based on a new medical examination, and both were summary in form. As described above, Murphy, the pain specialist, provided a one-page letter containing a number of general statements—e,g., that surveillance in general, and the surveillance of Gross in particular, is “not a reliable indicator of actual physical capacity” for someone with CRPS and fibromyalgia, and that the impact of her conditions “can vary from day to day” and “even minute to minute”—but he does not specifically address the gap between Gross’s reported symptoms and her seeming ability at times to “function quite- well” (Bhupalam’s words).17 Kaczmarek’s new submission similarly reports that Gross’s activities as seen in the surveillance videos are consistent with his prior findings that, she cannot engage in sedentary employment, but he provides no explanation for that conclusion.

Although Gross’s supplemental material responds to the questions raised in our prior decision, its persuasive; force would have been enhanced if either practitioner had provided some elaboration of his opinion that the surveillance is consistent with a finding that Gross is totally disabled. In addition, although we noted in Gross I that Gross had not “submitted] a statement from her own doctor refuting Sun Life’s assertion ... that the surveillance *15‘show[ed] a capacity for activity that far exceeds’ the limitations she claims,” .734 F.3d at 27.(second alteration in original), Gross did not supply such a response, .on remand. Sun Life asks us to infer that this omission is because Egan’s comments would be adverse to Gross. We decline to make that inference in part because it is highly speculative. More importantly, there is no necessary inconsistency- between the limitations Egan identified, in her evaluations of Gross at the end of 2006 and the surveilled activities. As we have observed, Egan contemplated the kind of limited activity seen in the videos—including driving and sitting for extended periods, and lifting ten pounds—and Gross’s occasional use of her right arm is not incompatible with Egan’s view that it was ordinarily useless. We nonetheless admonish Gross, along with Sun Life, for submissions on remand that are less than ideal.

2. Our Conclusion

We have thus gained little additional knowledge from the remand about the significance of the surveilled activities that previously gave us pause. In assessing the competing undeveloped views that have been presented, however, we find more plausible Murphy’s opinion that those activities do not contradict Gross’s medical history because her most extreme symptoms are not always present. A commonsense view of Gross’s maladies—one not dislodged by any persuasive contrary medical evidence—supports Murphy’s statement that “numerous factors” can affect the severity of her symptoms.

Indeed, the investigator’s reports and videos displayed significant variations in Gross’s capacity at times when Gross would have had no reason to’ fabricate symptoms.18 Importantly, the noteworthy departures comprised a small portion of her surveilled activities, and the surveillance on the day following her most ambitious activity-—traveling to her suddenly hospitalized mother—showed her physically depleted. Id. at 20. When exiting her residence for her appointment with Bhupa-lam, Gross walked with a limp and received assistance from her husband; when leaving the medieal center after roughly two- and one-half hours, she was in a wheelchair, and, the investigator observéd that Gross appeared unable to stand on her own. When she got out of the chair, she limped to the car and appeared to fall in or sit down quickly onto the passenger seat. That was the day Bhupalam concluded that she was “totally disabled even for sedentary work even on a part time basis.” Id. The next day, the investigator observed no activity'at all. Id. at 27.

Hence, we think it fair to conclude that' Sun Life gave undue importance to the few occasions when Gross appeared not to be disabled by her symptoms, leading to a distorted view of her capacities. When the surveillance is instead viewed in context, it belies Gross’s ¿bility to engage in! fulltime employment. Indeed, she appearéd to suffer significant consequences on the two days that followed the hospital visit (i.e., both the day she was examined by Bhupa-lam and the following day, when she apparently did not leave home). Such adverse physical impact is what one would expect when an individual with serious medical issues disregards her doctor’s guidelines; That is to say, activity restrictions do not necessarily define an individual’s maximum capacities on any particular occasion. Rather, prescribed limits are just as likely intended to protect the individual from aggravating her physical condition. From that perspective, when the three days of February surveillance are taken together, *16they reinforce, rather than undermine, Gross’s claim.

Moreover, the elusive source of Gross’s reported pain—perhaps fibromyalgia, perhaps RSD or CRPS, perhaps all three— gives particular importance to the credibility judgments of those who examined her. Neuren and Gupta emphatically rejected the possibility that the pain associated with CRPS could be sufficiently diminished to allow normal activity, but CRPS is not Gross’s only diagnosis. Gupta noted that, with respect to fibromyalgia, determining individuals’ “degree of impairment ... often rests on their credibility.” He went on to state that “[tjhere has not been enough credibility established for this claimant to allow for a determination of impairment to be based purely on her subjective complaints and allegations.” We disagree with his assessment of the evidence; while Gross’s credibility does find support in the record, the insinuations of fabrication do not.

To be sure, the record reflects some exaggeration by Gross, or perhaps selective reporting of her worst-case experiences. She evidently told Egan in the fall of 2006 that she “could not ... use her right hand,” id. at 17, and Kaczmarek also reported that she lacked “functional use of her right arm,” id. at 18. In January 2007, Gross told Kaczmarek that “she tolerates short bouts of activity for less than a few minutes” and has “difficulty walking with frequent falls.” Id. at 26. In November 2007, Egan stated that Gross “can hardly raise her arm.” Id. at 21. As we have seen, Gross plainly can, at times, use her right hand and arm, and—whether because of medication or otherwise—she can episodically “function quite well.” Overall, however, the medical record depicts an individual afflicted with chronic severe pain that routinely affects her ability to walk, bend and sit; whose ability to use her right arm is compromised; whose pain and fatigue restrict her day-to-day functioning; and who credibly manifested these conditions during physical examinations conducted by multiple medical professionals.19

We thus reiterate our conclusion that the medical record “support[s] a finding of total disability.” Id. at 25. Lacking persuasive evidence that that record inaccurately portrays Gross’s ability to work, we affirm the district court’s ruling that Gross is “entitled to disability benefits from Sun Life.” Gross Remand Op. at 5.

III. Sanctions

Sun Life argues that the district court abused its discretion by failing to impose sanctions on one of Gross’s attorneys for interfering with Sun Life’s post-remand investigation of Gross’s claim. We begin by describing the conduct underlying Sun Life’s contention.

A. Factual Background

In November 2014, while the renewed administrative proceedings were ongoing, *17attorney Michael Grabhorn wrote to Bhupalam and asked him to complete an addendum, provided with the letter, that would in effect override Bhupalam’s previous addendum and confirm the doctor’s original opinion that Gross’s physical limitations rendered her totally disabled.20 Grabhorn’s letter assumed that, at the time of Bhupalam’s examination of Gross, the doctor had not been provided with some of Gross’s medical records from her treating physicians or given the FCE conducted by Kaczmarek.21 Grabhorn also stated that Bhupalam had not been “made aware of the context of Ms. Gross’ physical activities observed in the' surveillance videos,” including the fact that her travel oh February 21, 2007 was in response to the news that her mother had been taken to the emergency room with chest pains. Grabhorn included with his letter Kaczma-rek’s post-remand report confirming his earlier findings.

Bhupalam did not respond to the November letter, which was re-sent to him via fax on December 1. In early February 2015, Grabhorn sent him another letter. The attorney reiterated his incorrect assertion that Bhupalam had not been provided with all of Gross’s medical records or the FCE, -noted that these materials had been sent with his previous letter, and stated that Bhupalam had failed to comply with the request that he “amend [his] prior medical opinion so as to accurately confirm [that] Mrs. Gross’ physical restrictions and limitations precluded her from engaging in active full-time employment of any kind.” In other words, Grabhorn more directly asked Bhupalam in this letter to withdraw the April 2007 addendum in which he had changed his opinion of Gross’s ability to work based on the video surveillance. Grabhorn threatened legal action if Bhupa-lam failed to “correet[]” his medical opinions, and he included a draft complaint alleging claims of negligence, defamation, and fraud, and seeking punitive damages.

Sun Life learned of these communications several weeks later when it received a letter from Bhupalam’s attorney explaining that the doctor would not be responding to Sun Life’s request for follow-up comment on Gross’s “functionality back in February 2007.” After reporting that Grabhorn had threatened to sue Bhupalam if he failed to retract his addendum, the attorney’s letter continued as follows:

As you can imagine, Dr. Bhupalam does not wish to be further involved in any way in the ongoing litigation between Ms. Gross and Sun Life Financial. Therefore, he respectfully declines to render any additional opinions regarding Ms. Gross’ condition and would stand by his addendum report.

B. Discussion

Sun Life argues on appeal that the district court should have addressed Grab-horn’s “unacceptable” actions “in some manner,” and it asserts that it was harmed because Grabhorn’s threat of legal action kept Bhupalam from responding to Sun Life’s request iior clarification of his opinion. In particular, Sun Life states that it was unable to include in its final decision letter the fact that Bhupalam had reaffirmed his addendum opining that Gross was capable of sedentary employment. In urging the need for sanctions to deter Grab-horn’s “bad behavior,” Sun Life points to an unrelated disability case in which Grab-*18horn was sanctioned for similar conduct that -another court labeled “inexcusable,” namely, making “thinly veiled threats designed tu silence the adverse opinion of an opposing party’s witness.” Graves v. Standard Ins. Co., No. 3:14-cv-558-DJH, 2015 WL 5613198, at *2 (W.D. Ky. Sept. 24, 2015). Sun Life also cites two other instances in which Grabhorn was sanctioned with an assessment of attorney’s fees for vexatious conduct. during discovery. See Graves v. Standard Ins. Co., No. 3:14-cv-558-CRS-DW, 2016 WL 6824403, at *1-3 (W.D. Ky. Nov. 17, 2016); Pogue v. Nw. Mut. Life Ins. Co., No. 3:14-cv-00598-CRS, 2016 WL 3748519, at *2-3 (W.D. Ky. July 8, 2016).

Grabhom’s threat of litigation against a potentially adverse expert is troubling, particularly given that it was not an aberration. Sun Life, however, provided no assistance to the district court in evaluating the sanctions question. It merely argued in its brief in support of its motion for judgment that Grabhorn’s “tampering with a witness ... should ... be sanctioned in a manner deemed appropriate” by the court. Gross v. Sun Life Assurance Co. of Can.; No. 1:09-cv-11678-RWZ, Docket 107, at 24 (filed Mar. 4, 2016): Sun Life neither addressed the source for the court’s authority to discipline Grabhorn nor provided examples of measures that would -be within that authority.22 Even on appeal, Sun Life does not suggest what sanctions might be “appropriate.”

The decision to impose sanctions is not to be made lightly. “The. Supreme Court has admonished courts to be cautious in using their inherent power to -sanction, explaining that -‘[b]ecause of. their very -potency, inherent powers must be exercised with restraint and discretion.’ ” United States v. Romero-Lopez, .661 F.3d 106, 108 (1st Cir. 2011) (quoting Chambers v. NASCO, Inc., 501 U.S. 32, 44, 111 S.Ct. 2123, 115 L.Ed.2d 27 (1991)). Although harm to an opposing party or counsel is not a prerequisite, see generally United States v. Kouri-Perez, 187 F.3d 1, 8-9 (1st Cir. 1999) (discussing “non-contempt punitive sanctions,” including for “harassment of opposing counsel”), the absence of harm from an attorney’s misbehavior reasonably may play a role in the district court’s decision on whether to undertake the process for imposing sanctions, see, e.g., Media Duplication Servs., Ltd. v. HDG Software, Inc., 928 F.2d 1228, 1238 (1st Cir. 1991) (“In general, a higher standard of due process protection is required where ... the sanction is a fine designed to go beyond' compensation and punish an attorney.”).

Here, notwithstanding its contention to the contrary, Sun Life has not demonstrated that it was disadvantaged by Grab-horn’s conduct. The follow-up information that Sun Life sought from Bhupalam was nonetheless added to the record. Sun Life asked Bhupalam to comment only on whether knowledge of the reason for Gross’s travel to the medical center “changes your opinion regarding her func*19tionality on that date.” Although Sun Life complains that Grabhorn’s interference resulted in belated notification that Bhupa-lam had reaffirmed his addendum, nothing turns on that delay. Grabhorn’s letters and Bhupalam’s response have been available to the district court and on appeal.23

Under these circumstances, and without condoning Grabhorn’s actions, we see no reason to second-guess the district court’s judgment not to award sanctions. Yet Grabhorn’s conduct on appeal makes this a closer issue than it might otherwise be. In the appellate brief that Grabhom signed, Gross represents that the threatening letter was a reaction to multiple unsuccessful attempts to obtain medical records, from Bhupalam. The brief also states that, “[a]s indicated in her letter, upon Dr. Bhupalam providing a complete copy of his chart, Ms. Gross agreed to forgo legal action.” Neither of the letters described above, however, referenced a request for records. To the contrary, the threat of legal action was linked to the demand that Bhupalam “correct” his medical opinions.24

In sum, while we uphold the district court’s exercise of discretion on the matter of sanctions, we Consider Grabhorn’s threat of litigation to Bhupalam, and his misrepresentations in defense of that conduct on appeal, worthy of reproach. Hence, in our mandate, we will direct the Clerk of Court to send a copy of this opinion to fhe Kentucky Office of Bar Counsel for whatever action, if any, it deems appropriate. See, e.g., Punzalan v. Holder, 575 F.3d 107, 112 (1st Cir. 2009) (directing the Clerk to send copies of the opinion to California bar disciplinary authorities); Aversa v. United States, 99 F.3d 1200, 1216 (1st Cir. 1996) (referring the matter of attorney conduct to, inter alia, the New .Hampshire .Supreme Court’s Professional Conduct Committee).

IY. Prejudgment Interest

In her cross-appeal, Gross argues that the district court . abused its discretion .’in awarding her prejudgment interest at the federal statutory rate. See 28 U.S.C. § 1961(a). We describe the applicable law before turning to Gross’s contentions.

A. Availability of Prejudgment Interest

ERISA does not explicitly provide for prejudgment interest, arid whether to grant such a remedy is thus within the discretion of the district court. Cottrill v. Sparrow, Johnson & Ursillo, Inc., 100 F.3d 220, 223 (1st Cir. 1996), abrogated on other grounds by Hardt v. Reliance Standard Life Ins. Co., 560 U.S. 242, 130 S.Ct. 2149, 176 L.Ed.2d 998 (2010). The court’s discretion also extends to the rate bf-interest to be applied, with the choice to be guided by equitable factors. Id.; see also Enos v. Union Stone, Inc., 732 F.3d 45, 50 (1st Cir. 2013).

We previously have identified two primary considerations when a,court decides to award prejudgment interest. First, ERISA’s remedial objectives are served by *20making the plan participant “whole for the period during which the fiduciary withholds money legally due.” Cottrill, 100 F.3d at 224; see also, e.g., Schumacher v. AK Steel Corp. Ret. Accumulation Pension Plan, 711 F.3d 675, 686 (6th Cir. 2013) (“An award that fails to make the plaintiff whole due to an inadequate compensation for her lost use of money frustrates the purpose of ERISA’s remedial scheme.”). Second, courts should endeavor to prevent unjust enrichment. Cottrill, 100. F.3d at 224. Awarding interest at a rate that does not recapture the lost value of the money during the period it was withheld “would create a perverse incentive” for a defendant to delay payments while it earned interest on those funds. Pacific Ins. Co. v. Eaton Vance Mgmt., 369 F.3d 584, 590 n.8 (1st Cir. 2004);25 see also Christianson v. Poly-America, Inc. Med. Benefit Plan, 412 F.3d 935, 941 (8th. Cir. 2005) (“A common thread throughout the prejudgment interest cases is unjust enrichment—the wrongdoer should not be allowed to use the withheld benefits or retain interest earned on the funds during the time of the.dispute.” (quoting Kerr v. Charles F. Vatterott & Co., 184 F.3d 938, 946 (8th Cir. 1999))); Rybarczyk v. TRW, Inc., 235 F.3d 975, 986 (6th Cir. 2000) (“To allow the Fund to retain the interest it earned on funds wrongfully withheld would be to approve of unjust enrichment.” (alteration omitted) (quoting Sweet v. Consol. Aluminum Corp., 913 F.2d 268, 270 (6th Cir. 1990))). At the same time, however, the rate should not be so high that it “impose[s] a punitive measure.” Schumacher, 711 F.3d at 686.

In Cottrill, we endorsed the district court’s use of the rate prescribed by § 1961(a), noting that “this rate promotes uniformity in ERISA cases” and was “especially appropriate ,.. because the Plan’s funds were initially invested in Treasury bills.” 100 F.3d at 225. We emphasized, however, that courts have “broad discretion” to select the rate, id., “and they may look to outside sources, including state law, for guidance,” id. at 224-25. In fact, courts have used various benchmarks to accomplish the dual objectives of making an ERISA plaintiff whole and avoiding unjust enrichment. See, e.g., Enos, 732 F.3d at 50 (upholding district court’s choice of “an interest rate set out in the parties’ own agreement”); Rybarczyk, 235 F.3d at 981, 985-87 (upholding district court’s award of the higher of the § 1961(a) rate or “the rate of return actually earned on the principal amount of the underpayment during the prejudgment period”); Frommert v. Lawrence Becker Xerox Corp. Plan Adm’rs, 216 F.Supp.3d 309, 316 (W.D.N.Y. 2016) (applying the federal prime rate because the state and federal statutory rates “could result in a 'windfall for one side or the other”); Gallagher v. Park West & Trust Co., 951 F.Supp. 10, 14 (D. Mass. 1997) (applying 12% state law rate).

One complexity in selecting an appropriate rate is the ever-changing relationship between statutory interest rates and the actual cost of money. In Schumacher, for example, the Sixth Circuit held that the *21district court abused its discretion by-awarding prejudgment interest at the federal statutory rate, which at that time was 0.12%. See 711 F.3d at 685. The court cited, inter alia, the then-current -annual rate of inflation (2.75%), the defendant’s borrowing costs (7.75%), and the defendant’s rate of return on its investments (6.55%) in concluding that the § 1961(a) rate was unfairly low, and it directed the district court on remand to “fashion an award that considers and balances the interests involved.” Id. at 686-87; see also Frommert, 216 F.Supp,3d at 315 (concluding that the 9% state rate was too high, and the federal rate at that time, 0.66%, was too low). By contrast, when we endorsed use of the § 1961(a) rate in Cottrill over Rhode Island’s 12% rate, the federal rate was 4.12%. See 100 F.3d at 224-25; https://www.treasury.gov/resource-center/ data-chart-center/interest-rates/Pages/ TextView.aspx?data=yieldYear&year= 1991 (“Treasury website”) (for Dee. 31, 1991 accrual date).

In sum, when a district court has concluded that a plaintiff should be awarded prejudgment interest, its task in selecting the rate is to identify, in the particular case, a fair percentage reflecting “both the rationale of full compensation and ERISA’s underlying goals.” Cottrill, 100 F.3d at 225.

B. Discussion

Gross asked for a prejudgment interest amount that would “reflect the actual interest earned -by Sun Life on Ms. Gross! withheld past due LTD benefits” or “interest calculated at her borrowing rate (e.g. the prime interest rate adjusted for risk of default).” Gross v. Sun Life Assurance Co. of Can., No. l:09-cv-11678-RWZ,-Docket No. 109, at 15 (filed Mar. 4, 2016). She asserted that interest should accrue from January 2007, the date of her benefits eligibility. The district court, without explanation, awarded prejudgment interest “from the date of the filing of the complaint in this action, calculated according to the method specified in 28 U.S.C. §' 1961.” Id. at Docket No. 123.

Gross argues that the court abused its discretion in selecting the federal rate because it is too low to make her whole “and by extension unjustly enrichjes] Sun Life at Mrs. Gross’ expense.”26 She now asserts that the court should have employed the greater of Massachusetts’s interest rate for contractual obligations (12%), see Mass. Gen. Laws ch. 231, § 6C, or Sun Life’s earnings percentage for the time period at issue. According to Gross, Sun Life’s public filings place the latter above 12% for at least a portion of the covered period. By comparison, the § 1961(a) rate in early October 2009, when Gross’s complaint was filed in federal court, was just 0.37%. See Treasury website (2009).27 At the time of the district court’s judgment in July 2016, it was 0.51%. Id. (2016).

Notwithstanding the district court’s considerable discretion in choosing the prejudgment interest rate, its decision must permit some scrutiny. Here, however, we are unable to .evaluate the court’s judgment call because it did not explain its reasoning, and its rationale is not apparent from the record. Cf. Enos, 732 F.3d at 50. *22(rejecting defendant’s complaint about lack of explanation for prejudgment interest award where “it, is apparent from the record that the amount was extrapolated from the rate stipulated in the GBA and recommended by the [plaintiffs]”). This is not a case, like Cottrill, where thé §, 1961(a) rate could.be expected to “approximate[ ] the likely return on the funds withheld.” 100 F.3d at 226 (describing similar holding in Algie v. RGA Global Communications, Inc., 891 F.Supp., 875, 899 (S.D.N.Y. 1994)); Moreover, the federal statutory rate is markedly lowér than' when we decided Cottrill.28 Also of importance'is" the lengthy delay in the benefits payment's to Cross— approaching eleven years since they should have commenced.

In these circumstances, mechanical adoption of the § 1961(a) rate would be an abuse of discretion. Because we cannot discern whether the court had supportable reasons for choosing that rate based on the equities and ERISA’s goals, we must vacate the award of prejudgment interest and remand to the district court for reassessment or explanation, of its interest-rate determination.

V. Attorney’s Fees

ERISA’s attorney’s fee provision allows a court in its discretion to award reasonable attorney’s fees in benefits proceedings. See 29 U.S.C. § 1132(g)(1). Following our decision in Gross I, Gross filed a motion in this court seeking fees and costs incurred thus far in the -case. Given the uncertainty about Gross’s entitlement to attorney’s fees under ERISA case law, we ordered the parties to- submit supplemental briefs addressing that issue. A split panel subsequently decided, in Gross II, that a fee award was appropriate based on Gross’s success in Gross I, where we held, inter alia, that “our circuit should no, longer apply the highly deferential ‘arbitrary and capricious’ standard of review to certain benefits decisions.” Gross II, 763 F.3d at 75. Significantly, the panel majority in Gross- II. concluded that Gross had achieved the degree of-success required for fees eligibility, see Hardt v. Reliance Standard Life Ins. Co., 560 U.S. 242, 245, 130 S.Ct. 2149, 176 L.Ed.2d 998 (2010), even though she had not yet been found entitled to disability benefits. See Gross II, 763 F.3d at 79-80. We further held that her fee request was ripe for adjudication because our “remand for reconsideration of her entitlement to benefits, in combination with a less deferential standard of review, means’-that Gross álrehdy ha[d] achieved the success that makes her eligible for fees.” Id. at 81.

We did . not ourselves perform “[t]he heavily fact-dependent lodestar analysis” that ordinarily is used to calculate fee awards, and instead instructed the district court to do so. Id., at 86. The. lodestar approach involves both an assessment of the lawyer hours reasonably spent on behalf of the prevailing party and a determination of the reasonable hourly rate for each attorney. See Matalón v. Hynnes, 806 F.3d 627, 638 (1st Cir. 2015). “Multiplying the results of these two inquiries yields the lodestar amount,” which may then be'adjusted “based on factors not captured in the lodestar calculation.” Id.

Utilizing the lodestar analysis, the district court ordered Sun Life to pay Gross $96,243.50 to- cover counsel fees through her first appeal, including for work on the post-judgment" fed petition that led to our decision in Gross II. See Gross v. Sun Life *23Assurance Co. of Can., 105 F.Supp.3d. 130, 140 (D. Mass. 2015). Gross now argues that the fee. award—a reduction of more than $188,000 from the amount requested—is unreasonably low, and she specifically challenges multiple cuts that she claims are unjustified. She contests, for example, the court’s reduction in the hourly rate of compensation for one of her attorneys, rejection of fees for work spent on discovery motions, and the one-third reduction in the time allowed for her prior appeal of the benefits denial (i.e., Gross I), and the two-thirds cut in the time for preparing her fees petition (the matter addressed in Gross II).

We review a district court’s ruling on a fee request for abuse of-discretion. Cent. Pension Fund of the Int’l Union of Operating Eng’rs & Participating Emp’rs v. Ray Haluch Gravel Co., 745 F.3d 1, 4 (1st Cir. 2014). “This standard is highly deferential, and ‘we will set aside a fee award only if it clearly appears that the trial court ignored a factor deserving significant weight, relied upon an improper factor, or evaluated all the proper factors (and no improper ones) but made a serious mistake in weighing them.’ ” Id. (quoting Gay Officers Action League v. Puerto Rico, 247 F.3d 288, 292-93 (1st Cir. 2001)).

The district court wrote a thorough opinion based on “a detailed analysis of the submitted billing records.” Gross, 105 F.Supp.3d at 133; see also- id. at 136 (noting the court’s “line-by-line review of the billing records”). The court explained its reasoning for the reductions, and we have carefully reviewed those judgment calls. For certain of Gross’s complaints, it suffices to say that, given “the latitude ceded to district courts in making fee awards and the flexibility inherent in the lodestar approach,” we find no basis for disturbing the court’s decision. Matalon, 806 F.3d at 638.

Although we also find no abuse of discretion in the court’s selected hourly rate for Grabhorn, we nonetheless think it useful to review our precedent on the choice of an appropriate rate. We also explain below why two adjustments in the fee award are necessary.

A, Grabhorn’s Hourly Rate

Both Grabhorn, a Kentucky-based ERISA lawyer, and Jonathan Feigenbaum, a Boston-based ERISA lawyer, sought to recover fees at a rate of $500 per hour. The district court found that rate reasonable for Feigenbaum, noting that Boston “hourly legal fees are among the highest in the country.” Gross, 105 F.Supp.3d at 135. However, the court concluded that Grab-horn’s 'compensation should reflect his “normal hourly rate” in Kentucky, and it therefore awarded him $375 per hour. On appeal, Gross maintains that both attorneys should have been paid based on the higher legal fees prevailing in the jurisdiction where the case was heard, i.e., Boston.

Our court has endorsed the proposition that “reasonable hourly rates should be set by reference to rates in the court’s vici-nage rather than in the lawyer’s region of origin.” Gay Officers Action League, 247 F.3d at 296 (citing Adcock-Ladd v. Sec’y of Treas., 227 F.3d 343, 350 (6th Cir. 2000)); see also, e.g„ United States v. One Star Class Sloop Sailboat, 546, F.3d 26,. 38 (1st Cir. 2008) (“Reasonable hourly rates will vary depending on the nature of the work, the locality in which it is performed, the qualifications of the lawyers, and other criteria.” (emphasis added)). We also have held, however, that a court may properly conclude that the prevailing rate in the court’s locale is not the appropriate benchmark in particular circumstances. See One Star Class Sloop Sailboat, 546 F.3d at 40 (“When a party recruits counsel from outside the vicinage of the forum court, that *24court may deem the relevant community to be the community in which the lawyer maintains his or her principal office.” (internal quotation marks omitted)); id. (noting that a court may look to an attorney’s “actual billing practices to determine the relevant rate”). Accordingly, our precedent allows a court to choose “counsel’s standard rate, or the prevailing market rate in the forum, or a reasonable rate in between.” Id. at 41.

Given this flexibility, we cannot say the district court exceeded its authority in determining that Grabhorn’s hours should be compensated at a lower rate than Feigenbaum’s. The court expressly recognized that it could properly award Grabhorn the Boston hourly rate. See Gross, 105 F.Supp.3d at 136 (noting that “Grabhorn’s out-of-state status does not, in itself, weigh in favor of reducing his fee request”). As a matter of discretion, however, the court concluded that Grabhorn is more appropriately compensated based on the prevailing rate where he maintains his office and was hired by Gross. On the record before us, we find no basis on which to disturb that judgment.

B. The Fee Petition

In calculating the attorneys’ compensation for litigating Gross’s fee petition following our decision in Gross I, the district court trimmed Gross’s request by two-thirds, awarding fees for only 22.4 of the 67.3 hours claimed. Gross, 105 F.Supp.3d at 137. The court explained this decision with the conclusory observation that “[a] fee petition in an ERISA case should be a straightforward exercise, particularly for experienced ERISA practitioners like plaintiffs counsel.” Id. In the court’s view, the fee petition “could have reasonably been completed in a third of the time billed by plaintiffs counsel.” Id

The court’s characterization of ERISA fee petitions as “straightforward” may be apt for the ordinary case. See Matalón, 806 F.3d at 639 (“[W]e have indicated that certain components of fee awards (such as work performed in preparing and litigating fee petitions) may be calculated at discounted rates due to the comparative simplicity of the task.”). It does not apply, however, to the petition here. Indeed, as described above, we requested supplemental briefs on the question whether Gross was entitled to fees for the proceedings leading up to, and including, our decision in Gross I—as well as the proper timing for any such award—because of the complexity of those issues. The panel divided in the decision. See Gross II, 763 F.3d at 86 (dissenting opinion). Then, once we remanded the case to the district court for the factbound lodestar analysis, Gross needed to submit additional materials to that court. In these circumstances, we conclude that the court erred in treating Gross’s fee petition as run-of-the-mill and, hence, abused its discretion in finding that only 22.4 hours were reasonably expended on that aspect of the litigation.

' Having reviewed the billing records ourselves, we are satisfied that the 67.3 hours billed—37.6 by one attorney, 27.7 by another, plus two paralegal hours—reflect a reasonable expenditure of time in light of both the difficulty of the legal questions and the multiple phases of the fee proceedings. Indeed, the 67.3 total appears to be both an accurate accounting of the time spent and an appropriate allocation of resources. On remand, the district court should adjust its calculation of compensa-ble hours to include the full 67.3 hours for the work on the fees petition.

C. Summary Judgment

Also problematic in the court’s lodestar analysis is its 50% reduction in *25the attorney hours allowed for summary judgment work. The court appeared to adjust the compensable time downward, in part, because of hours “spent on plaintiffs alternative arguments, many of which were not successful.” Gross, 105 F.Supp.3d at 138. Gross’s summary judgment briefing, however, primarily challenged Sun Life’s benefits decision, a position on which she ultimately prevailed. Although her memorandum in support of judgment included some off-the-mark arguments about Sun Life’s experts, it consisted for the most part of ordinary advocacy for her view of the record. Gross also needed to respond to Sun Life’s cross-motion for judgment. Hence, to the extent the court discounted the time spent on summary judgment for lack of success, we conclude that it erred.

However, the court also expressed the view that the total hours devoted to the summary judgment motions—105.5 attorney hours and 5.5 paralegal hours—was unreasonable, and we find no abuse of discretion in that judgment. Accordingly, we conclude that a 25% downward adjustment in the hours sought would more accurately reflect both the success Gross achieved on her claim for benefits and the district court’s permissible view that the total of summary judgment hours was excessive.

We also wish to briefly comment on the district court’s 33% downward adjustment for the hours spent on Gross’s first appeal. That reduction was among those made to - account for “time spent pursuing unsuccessful claims and to reflect the quality of the plaintiffs victories.” Id. at 137. Noting Gross’s -mixed results in Gross I,29 the district court concluded that it was not reasonable to award the full amount of fees sought—162.4 hours. Id. at 138-39. This judgment is within the bounds of the court’s discretion. We concluded in Gross II that “the relative merits of th[e] action do not line up solely on Gross’s side of the calculus.” 763 F.3d at 85. Gross had not at that point established a right to benefits, and we rejected one of her primary contentions. See id. In other words, Gross achieved only partial success in the first round of litigation, Against this backdrop, we cannot say the district court made “a serious mistake” in determining the allowable hours for the appeals work. Gay Officers Action League, 247 F.3d at 293.30

YI. Summary

We uphold the district court’s determinations on both of the issues appealed by Sun Life,' affirming the award of disability benefits to Gross and leaving intact the court’s judgment declining to impose sanctions on attorney Michael Grabhorn. On Gross’s cross-appeal, we remand to the district court the question of the appropriate rate of prejudgment' interest. We affirm in part and vacate in part the district court’s attorney’s fee calculation. As explained above, we direct the court to recal-*26enlate the fee award with an additional 44.9-hours for Gross’s attorneys’ work on her fee petition and with a 25%, rather than 50%, downward -adjustment in the time for work on the summary judgment motions.

■ Affirmed in part, vacated in part, and remanded for further proceedings consistent with this opinion. The Clerk of Court is directed to send a copy of this opinion to the Kentucky OfSce of Bar Counsel. Costs to appellee/cross-appellant.

(Concurring)

The central merits issue in this case is whether Ms. Gross was physically-disabled within the- meaning of Sun Life’s group disability policy. The law is quite clear that Gross bore the burden of proof on that issue. See Orndorf, 404 F.3d.at 518-19 (characterizing the conclusion that claimant bears the burden .of proving disability as a “guiding principle” in the appellate court’s analysis). It is also quite clear that in resolving such an issue courts generally review and -weigh .the administrative record as a whole. See Scibelli v. Prudential Ins. Co. of Am., 666 F.3d 32, 40 (1st Cir. 2012). Were we . to so proceed on this appeal, I would find that .Gross failed to carry her burden of showing that the evidence, as. a whole established her claimed disability. Only one doctor, Bhupalam, saw Gross and viewed the video. A straight-shooter, Bhupalam initially deemed Gross disabled. He then reversed his opinion when shown the video evidence, finding it too incompatible with Gross’s subjectively supported symptoms and limitations. As best the record shows, Gross either never showed the video to her own doctors or, if she did, she was unable to get. them to confirm their opinions once they saw the video. Instead, her lawyer set to trying to muzzle Bhupalam. On such a record, it requires no undue speculation to figure out what is likely going on. At the very least, Í would find Giross’s inability to parry Bhupqlam’s post-video opinion with an opinion from any doctor who saw. or treated her to be a dispositive failure in her effort to carry her burden.

My colleagues, though,.-read Gross I as having bifurcated the usual ERISA merits inquiry. They conclude that some of the evidence (i.e., the evidence excluding both the video and conclusions to be drawn from the video) weighs in favor of finding of disability, and then treat this appeal as a proceeding in which Sun Life bears the burden of upsetting that conclusion. This bifurcated parsing of the evidence provides a potent tool for burden shifting. Here, for example, Gross I found the evidence as a whole did not justify the entry of judgment for Gross, where the only doctor who saw both Gross and the video recanted his opinion after he saw the video. Common sense would suggest that the case might therefore turn on what Gross’s, numerous treating physicians had to say in response, i.e., did.they stand by their opinions once shown the video? Instead, when the treating physicians remained mum, my colleagues (treating the Bhupalam opinion as old news) now find that Sun Life needed to do more “to show that ... the capabilities Gross demonstrated in the videos were incompatible ‘with the medical record of disability.” While'I'doubt that this is a proper way to proceed, I acknowledge that one could reasonably read Gross I as setting up such a burden-shifting inquiry. And while I would normally eschew such a reading in the absence of more express direction (and perhaps some support in the case law), I defer here to my colleagues, both of whom were on the Gross I panel and seem to regard such a reading as plainly manifest (or at least intended). For this reason alone, I concur.