MARTIN, Circuit Judge:

Frank Douglas, a federal inmate, appeals the District Court's dismissal of his Federal Tort Claims Act (FTCA) case. He claims that a Bureau of Prisons (BOP) official withheld wages he was owed for his work while incarcerated. The District Court held that this claim was barred by the FTCA's discretionary function exception. Mr. Douglas also filed related claims of discrimination, retaliation, and intentional infliction of emotional distress. The District Court dismissed these other claims based on the FTCA's exhaustion requirements. We reverse the District Court on the pay claim and affirm for the others.

I.

Mr. Douglas is a 56-year-old federal inmate. When he filed this lawsuit, he was incarcerated at FCC Coleman, a penitentiary in Florida, where he worked a trash shift multiple days a week. According to Mr. Douglas, this shift required him to "operate[ ] a very dangerous recycl[ing] machine for card-board." Mr. Douglas was one of two operators of this machine, the contents of which "weighed one or two tons" and had to be loaded into a semitruck three or four times a week.

The BOP assigns inmate workers to one of four grades of "performance pay," with Grade 1 workers paid the most. See 28 C.F.R. § 545.26(b). According to Mr. Douglas's complaint, "[m]e and my supervisor went over my grade and pay numerous times." Then, on March 15, 2012, Mr. Douglas and his supervisor both signed a "Work Performance Rating" form indicating that Mr. Douglas had worked 154 hours of "satisfactory work" in the past month. The form also indicated that Mr. Douglas's "Performance Pay Grade Class" was 1 and that he would be paid $91.60. However, when Mr. Douglas was paid four days later, he received $7.20. The same thing happened the next month, when Mr. Douglas's supervisor again approved $91.60 for 154 hours of Grade 1 work but Mr. Douglas was paid only $12.00. Mr. Douglas alleges that these changes to his pay were made by a prison official named Lieutenant Barker. According to the complaint, Lt. Barker's role with respect to inmate pay is entering pay data into a computer system. Mr. Douglas also alleges that Lt. Barker said that he reduced the pay because "I don't like Inmate Douglas black ass and I'm going to pay him what I want."

Based on these allegations, Mr. Douglas filed a "Small Claims for Property Loss" form with the BOP on April 9, 2012. The BOP sent Mr. Douglas a final denial of this claim on June 21, 2012. Mr. Douglas then filed this lawsuit on June 27, 2012. By the time he sued, Mr. Douglas had also filed BOP complaints alleging retaliation, racial discrimination, and intentional infliction of emotional distress, but these were not fully appealed until later. The government moved to dismiss the suit under Rule 12(b)(6), arguing that Mr. Douglas's pay claim was barred by the FTCA's discretionary function exception and that the other claims were barred by the FTCA's exhaustion requirements. The motion was styled "alternatively" as a motion for summary judgment and included excerpts from various BOP documents, as well as a declaration by the prison's Inmate Performance Pay Coordinator. The District Court granted the motion to dismiss on March 12, 2014.

II.

We first address whether Mr. Douglas's pay claim was barred by the FTCA's discretionary function exception. We review this question of law de novo. See Cohen v. United States, 151 F.3d 1338, 1340 (11th Cir.1998).

A.

The FTCA's discretionary function exception provides that the United States does not waive sovereign immunity for claims "based upon the exercise or performance or the failure to exercise or perform a discretionary function or duty on the part of a federal agency or employee of the Government." 28 U.S.C. § 2680(a). We apply this exception by answering two questions. "First, we consider the nature of the conduct and determine whether it involves `an element of judgment or choice.'" Ochran v. United States, 117 F.3d 495, 499 (11th Cir.1997) (quoting United States v. Gaubert, 499 U.S. 315, 322, 111 S.Ct. 1267, 1273, 113 L.Ed.2d 335 (1991)). "[C]onduct does not involve an element of judgment or choice, and thus is not discretionary, if `a federal statute, regulation, or policy specifically prescribes a course of action for an employee to follow, because the employee has no rightful option but to adhere to the directive.'" Id. (quoting Gaubert, 499 U.S. at 322, 111 S.Ct. at 1273). "Second, if the conduct at issue involves the exercise of judgment, we must determine whether that judgment is grounded in considerations of public policy." Id.

When a plaintiff challenges the actions of an individual employee who is working within a broader administrative scheme, "a court must first consider whether the action is a matter of choice for the acting employee." Berkovitz v. United States, 486 U.S. 531, 536, 108 S.Ct. 1954, 1958, 100 L.Ed.2d 531 (1988). "This inquiry is mandated by the language of the exception," which "protects the discretion of the executive or the administrator to act according to one's judgment of the best course." Id. (quotation omitted).1 "For a complaint to survive a motion to dismiss, it must allege facts which would support a finding that the challenged actions are not the kind of conduct that can be said to be grounded in the policy of the regulatory regime." Gaubert, 499 U.S. at 324-25, 111 S.Ct. at 1274-75.

B.

Before getting to the details of the government's discretionary function challenge, we must address a preliminary question: whether our review is limited to the allegations in Mr. Douglas's complaint, or whether we should consider the extrinsic evidence that the government filed in the District Court. The government attached several documents to its motion to dismiss, which was styled "alternatively" as a summary judgment motion. The District Court dismissed the case without considering this evidence or allowing Mr. Douglas to conduct discovery to respond to it.

When reviewing a discretionary function decision "entered on a motion to dismiss," the standard of review is usually straightforward: we "accept all of the factual allegations in [the] complaint as true and ask whether the allegations state a claim sufficient to survive a motion to dismiss." Gaubert, 499 U.S. at 327, 111 S.Ct. at 1276 (quotation omitted); see also Mesa v. United States, 123 F.3d 1435, 1437 (11th Cir.1997). The government says this rule doesn't apply here because "this Court has long treated a section 2680 bar as jurisdictional." According to the government, this means its attack on Mr. Douglas's complaint is based on Rule 12(b)(1), which deals with subject matter jurisdiction, rather than Rule 12(b)(6), which asks if the plaintiff properly stated a claim. Whereas 12(b)(6) challenges are "facial" (meaning we accept the allegations in a complaint as true), 12(b)(1) challenges can be "factual" (meaning we can use extrinsic evidence). See Carmichael v. Kellogg, Brown & Root Servs., Inc., 572 F.3d 1271, 1279 (11th Cir. 2009).

The problem with this argument is that the government expressly moved to dismiss based on Rule 12(b)(6). The District Court then granted the government's motion based solely on the allegations in Mr. Douglas's complaint. That means the dismissal was facial. See McElmurray v. Consol. Gov't of Augusta-Richmond Cty., 501 F.3d 1244, 1251 (11th Cir.2007) (deciding that a dismissal for lack of subject matter jurisdiction was facial because "[t]he district court did not decide any issues of disputed fact"); Stalley ex rel. United States v. Orlando Reg'l Healthcare Sys., Inc., 524 F.3d 1229, 1233 (11th Cir. 2008) (per curiam) (deciding that a subject matter jurisdiction challenge was "a facial attack on the complaint" because "the record establishes that the district court considered only the complaint and the attached exhibits"). If the District Court had decided a factual challenge, this case would have come to us in a very different form. That is because "in a factual challenge the district court must give the plaintiff an opportunity for discovery and for a hearing that is appropriate to the nature of the motion to dismiss." McElmurray, 501 F.3d at 1251 (citing Williamson v. Tucker, 645 F.2d 404, 414 (5th Cir. 1981)). Mr. Douglas was not given these opportunities here. No matter the reason behind a dismissal, we can't convert a facial dismissal into a factual one on appeal. If we did, we would be finding facts for the first time on appeal.

There is another reason to treat the government's challenge as facial: the challenge turns on the merits of the overall case. "We have cautioned [ ] that the district court should only rely on Rule 12(b)(1) if the facts necessary to sustain jurisdiction do not implicate the merits of plaintiff's cause of action. If a jurisdictional challenge does implicate the merits of the underlying claim then: `[T]he proper course of action for the district court. . . is to find that jurisdiction exists and deal with the objection as a direct attack on the merits of the plaintiff's case.'" Morrison v. Amway Corp., 323 F.3d 920, 925 (11th Cir.2003) (quoting Garcia v. Copenhaver, Bell & Assocs., 104 F.3d 1256, 1261 (11th Cir.1997)). This serves to protect "the plaintiff who in truth is facing a challenge to the validity of his claim: the defendant is forced to proceed under Rule 12(b)(6) . . . or Rule 56 . . . both of which place great restrictions on the district court's discretion." Id. (quotation omitted).2

The government's discretionary function challenge is based on factual contentions that go right to the merits of the overall case: questions like whether Mr. Douglas's pay grade was correct, who can set this grade, who can review this decision, and when exactly Mr. Douglas's pay vested. If the government wishes to press on with this challenge, "the proper course of action for the district court . . . is to find that jurisdiction exists and deal with the objection as a direct attack on the merits of the plaintiff's case." Id. (quotation omitted). The court can then address this challenge in the proper manner: after discovery and with the protections of either a trial or review on summary judgment.3 In this appeal, we "accept all of the factual allegations in [the] complaint as true and ask whether the allegations state a claim sufficient to survive a motion to dismiss." Gaubert, 499 U.S. at 327, 111 S.Ct. at 1276 (quotation omitted).

C.

At the pleading stage, Mr. Douglas must allege a plausible claim that falls outside the discretionary function exception. Autery v. United States, 992 F.2d 1523, 1526 n. 6 (11th Cir.1993). The exception does not cover government acts that "violated a mandatory regulation or policy that allowed no judgment" or acts that, although a permissible exercise of discretion, were not "grounded in the policy of the regulatory regime." Id. at 1526-27 (quoting Gaubert, 499 U.S. at 325, 111 S.Ct. at 1275). The relevant act alleged is Lt. Barker's refusal at the payroll record data-entry stage to enter a wage amount that Mr. Douglas was entitled to receive because it was based on his pay grade and hours worked; the wages had been approved by Mr. Douglas's supervisor; and Mr. Douglas's right to the wages had vested. The record shows that BOP regulations allowed no discretion to refuse to pay the wages at that stage and that the refusal was not grounded in policy. The District Court erred in dismissing the claim on the basis of the pleading allegations.

1.

The BOP has promulgated regulations describing the process by which an inmate's pay "becomes vested." 28 C.F.R. § 545.26(h). These regulations generally give prisons discretion to decide which inmates will work and which types of work, if any, will be compensated at which pay grade. See id. § 545.25(b); id. § 545.26(a)-(b). But once an inmate's work is assigned a pay grade and evaluated by his supervisor, the regulations make clear that the inmate is entitled to the pay that the supervisor computed. The regulations state that "[a]t the end of each month the work detail/program supervisor shall compute the hours worked by the inmate and the pay to be awarded for that month." Id. § 545.26(e)(1). A supervisor is required to "rate the inmate's performance," "review the evaluation with the inmate," and "request that the inmate sign the evaluation form." Id. § 545.26(e)(3). Following this review, the regulations declare that "[a]n inmate's performance pay, once earned, becomes vested." Id. § 545.26(h).4

This process is exactly what Mr. Douglas says happened with his pay, up until Lt. Barker intervened. Mr. Douglas says his supervisor evaluated his work for both March and April 2012 and approved 154 hours of satisfactory work at Grade 1 each month. The supervisor then computed that Mr. Douglas would be paid $91.60 each month. Both the supervisor and Mr. Douglas then signed the evaluation. Assuming these allegations are true, Lt. Barker had no discretion to input a lower pay grade at this point. According to the BOP's own regulations, Mr. Douglas's pay had already vested. A government employee's conduct "is not discretionary[ ] if `a federal statute, regulation, or policy specifically prescribes a course of action for an employee to follow, because the employee has no rightful option but to adhere to the directive.'" Ochran, 117 F.3d at 499 (quoting Gaubert, 499 U.S. at 322, 111 S.Ct. at 1273). Once a supervisor evaluated Mr. Douglas's work and computed his pay, the decision to pay him the computed amount was not discretionary. It was mandated by the BOP's own regulations.

The District Court held otherwise. It reasoned that "[i]f the BOP has discretion with respect to work assignments, it follows that there is discretion with respect to the amount to pay prisoners for their work." That is true. But Mr. Douglas has challenged something narrower. He says Lt. Barker changed his pay grade after BOP officials exercised discretion to approve his pay and his supervisor approved the hours he worked, in accordance with BOP regulations setting out how and when inmate pay vests. See 28 C.F.R. § 545.26(e)-(h). If discretion to set inmate pay is the same as discretion to withhold pay an inmate already earned, then the BOP's entire regulatory scheme for inmate pay—codified through notice-and-comment rulemaking—would mean nothing. If a prison can at every point say that inmates haven't earned their pay because the prison could have paid them less in the first place, then the rule wouldn't be that "[a]n inmate's performance pay, once earned, becomes vested." 28 C.F.R. § 545.26(h). It would be: pay becomes vested only when the prison says so.

The government has one more argument for why Lt. Barker had discretion to withhold Mr. Douglas's performance pay. It argues that the BOP's pay regulations reference the ability of the "Warden" or "Department Head" to approve "bonus pay" and "special bonus" pay. See 28 C.F.R. § 545.26(f)-(g).5 But the government doesn't contend that Lt. Barker is a Warden or Department Head. It also doesn't claim that anyone—let alone a Warden or Department Head—told Lt. Barker to lower Mr. Douglas's pay. Even if the government had raised either of these contentions, we would have to ignore them at the 12(b)(6) stage. Besides, even if we assume that a Warden or Department Head ordered the reduction of Mr. Douglas's pay, there's another problem: the regulations only require Warden or Department Head approval for bonus pay, not performance pay. See id. For performance pay, the regulations say that pay vests after a "work detail/program supervisor" evaluates an inmate's work and computes his pay. Id. § 545.26(e)(3).6 Mr. Douglas says that is precisely what happened here: his supervisor evaluated his work and computed his pay, so it vested.

2.

Not only has Mr. Douglas plausibly alleged an act that disregarded and violated a mandatory BOP regulation, he has alleged a violation that was not grounded in policy. And, if as Mr. Douglas claims, Lt. Barker unilaterally altered Mr. Douglas's wages out of racial animus, Lt. Barker's conduct, even if discretionary, "cannot be said to be based on the purposes that the regulatory regime seeks to accomplish." Gaubert, 499 U.S. at 325, 111 S.Ct. at 1275.

III.

The government offers a few more arguments on appeal. It says Mr. Douglas's pay claim is barred by two additional FTCA exceptions: 28 U.S.C. § 2680(h), which bars claims based on "misrepresentation, deceit, or interference with contract rights," and 28 U.S.C. § 2680(c), which bars "[a]ny claim arising in respect of . . . the detention of any . . . property by any. . . law enforcement officer." Neither bar was raised in the District Court, where the government moved to dismiss solely based on the discretionary function exception. But the government quotes our rule "that the judgment of a district court may be affirmed upon any adequate ground, even if it is other than the one on which the court actually relied." Greenberg v. Nat'l Geographic Soc'y, 533 F.3d 1244, 1272 n. 28 (11th Cir.2008) (en banc).

Mr. Douglas responds that "a factual subject-matter jurisdiction argument" is not one of the grounds on which "this Court may affirm based on arguments never presented to the district court." This is true. Again, a "factual attack" on subject matter jurisdiction "challenge[s] the existence of subject matter jurisdiction in fact, irrespective of the pleadings, and matters outside the pleadings, such as testimony and affidavits are considered." In re CP Ships Ltd. Sec. Litig., 578 F.3d 1306, 1311-12 (11th Cir.2009) (quotation omitted), abrogated on other grounds by Morrison v. Nat'l Australia Bank Ltd., 561 U.S. 247, 130 S.Ct. 2869, 177 L.Ed.2d 535 (2010). If we decide this kind of challenge for the first time on appeal, we necessarily have to act as a fact finder. Here, both the § 2680(c) and the § 2680(h) challenges require us to look beyond the facts alleged in Mr. Douglas's complaint.

First, "[t]he test in applying [the § 2680(h)] exception is whether the essence of the claim involves the government's failure to use due care in obtaining and communicating information." JBP Acquisitions, LP v. U.S. ex rel. F.D.I.C., 224 F.3d 1260, 1264 (11th Cir.2000). Mr. Douglas's complaint does not claim "failure to use due care in obtaining or communicating information" to him. He does not allege that the prison made a false statement to him, deceived him, or entered into a contract with him. Instead, he says Lt. Barker wrongfully reduced his pay after it vested. Once the case gets past the pleadings stage, the government may offer proof that Mr. Douglas was in fact set for lower pay or that his pay didn't vest until after Lt. Barker entered it. But we can't hold Mr. Douglas responsible for raising this kind of claim without going beyond his complaint.

The same is true about § 2680(c), which bars "any claim arising in respect of. . . the detention of any . . . property by any . . . law enforcement officer." Mr. Douglas alleges that a prison official wrongfully interfered with his pay. This is not a claim that someone wrongfully took custody or possession of his property, like in both of the § 2680(c) cases the government cites. See Ali v. Bureau of Prisons, 552 U.S. 214, 128 S.Ct. 831, 169 L.Ed.2d 680 (2008); Krug v. United States, 442 Fed.Appx. 950 (5th Cir.2011) (per curiam) (unpublished). From the face of the complaint, Mr. Douglas has not raised a claim about "detention" of property.

IV.

Finally, Mr. Douglas argues that the District Court erred when it found that his remaining claims of discrimination, retaliation, and intentional infliction of emotional distress were not exhausted. Before filing an FTCA lawsuit, a plaintiff must fully exhaust administrative remedies for his claims. McNeil v. United States, 508 U.S. 106, 113, 113 S.Ct. 1980, 1984, 124 L.Ed.2d 21 (1993). The BOP has promulgated regulations describing how prisoners should file administrative FTCA claims against the BOP. 28 C.F.R. § 543.30. Prisoners may file suit only after the final agency action. Id. § 543.32(g).

Mr. Douglas failed to fully exhaust the BOP's administrative remedies before filing suit. Mr. Douglas filed his claims through the BOP's Administrative Remedy Program (ARP), which handles all inmate claims relating to confinement. This appears to be separate from the BOP's procedure for inmate FTCA claims. See id. § 542.10(c); id. § 543.31. The government argues that exhausting through the ARP does not satisfy the FTCA's exhaustion requirement. We need not decide this question now because Mr. Douglas submitted his earliest ARP appeal on July 31, 2012. This was weeks after he filed this suit. We therefore agree with the District Court that Mr. Douglas failed to fully exhaust administrative remedies before filing suit.

V.

For these reasons, we reverse the District Court on Mr. Douglas's pay claim, affirm the remainder of the court's holdings, and remand for further proceedings consistent with this opinion.7

AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.

TJOFLAT, Circuit Judge, concurring:

For the reasons given in the court's considered opinion, I agree that the District Court improperly dismissed Douglas's pay-grade claim under Rule 12(b)(6) of the Federal Rules of Civil Procedure.1 I write separately to emphasize the narrowness of this holding and to explain what must happen on remand.

If this were a typical case, the path forward would be clear. Because Douglas has plausibly stated a claim upon which relief could be granted, taking as true the allegations in his complaint, the United States' motion to dismiss would be defeated and the case would proceed to discovery. See Fed.R.Civ.P. 12(b)(6); Ashcroft v. Iqbal, 556 U.S. 662, 678-79, 129 S.Ct. 1937, 1949-50, 173 L.Ed.2d 868 (2009); Bell Atl. Corp. v. Twombly, 550 U.S. 544, 554-56, 127 S.Ct. 1955, 1964-66, 167 L.Ed.2d 929 (2007). Following discovery and absent a settlement, the United States or Douglas could then move for summary judgment or, if genuine disputes of material fact remained, proceed to trial. Either way, the result would be a decision on the merits.

But this is not a typical case. Douglas sued the United States, which would normally be immune from suit because of its sovereign immunity, under the Federal Tort Claims Act ("FTCA"), 28 U.S.C. § 1346(b)(1). The FTCA waives the United States' "traditional all-encompassing immunity" in certain circumstances. Rayonier Inc. v. United States, 352 U.S. 315, 319, 77 S.Ct. 374, 377, 1 L.Ed.2d 354 (1957). Relevant here, Douglas seeks "money damages" for "loss of property"— his allegedly vested pay—that was "caused by the negligent or wrongful act or omission" of a government employee—Lt. Barker—"acting within the scope of his office or employment, under circumstances where the United States, if a private person, would be liable to the claimant in accordance with the law of the place where the act or omission occurred." 28 U.S.C. § 1346(b)(1). If Douglas can make this showing and none of the exceptions enumerated in 28 U.S.C. § 2680 apply to bar his suit, the United States has waived its immunity to suit and Douglas's case may proceed. If Douglas cannot make this showing or one of the § 2680 exceptions applies, the United States has not waived its immunity and Douglas's case must be dismissed.2

The complicating factor for cases like Douglas's is simply put: "Sovereign immunity is jurisdictional in nature." Fed. Deposit Ins. Corp. v. Meyer, 510 U.S. 471, 475, 114 S.Ct. 996, 1000, 127 L.Ed.2d 308 (1994). Federal courts are courts of limited jurisdiction. The existence of subject-matter jurisdiction as a necessary prerequisite to decide "Cases" and "Controversies" is critical to our Constitution's separation-of-powers principles and the rule of law in a system of limited government. U.S. Const. art. III, § 2. That is why litigants cannot waive a lack of subject-matter jurisdiction, and why federal courts have an independent duty to inquire into their own jurisdiction when it may be in doubt. See Commodity Futures Trading Comm'n v. Schor, 478 U.S. 833, 850-51, 106 S.Ct. 3245, 3256-57, 92 L.Ed.2d 675 (1986) ("When these Article III limitations are at issue, notions of consent and waiver cannot be dispositive because the limitations serve institutional interests that the parties cannot be expected to protect."). In line with these principles, a party may move to dismiss for lack of subject-matter jurisdiction at any time. Fed.R.Civ.P. 12(b)(1), (h)(1). And if "the court determines at any time that it lacks subject-matter jurisdiction, the court must dismiss the action." Fed.R.Civ.P. 12(h)(3).

So what happens when, as is the case here, there is a sufficient pleading of subject-matter jurisdiction under the FTCA based on waiver of sovereign immunity as a facial matter—that is, taking as true the allegations contained in the complaint—but serious doubts remain as a factual matter? The answer depends on where you bring suit. Some circuits take the approach that courts should make a threshold finding under Rule 12(b)(1), which may entail limited discovery and jurisdictional fact-finding, before ruling on the merits because it would be improper to proceed before subject-matter jurisdiction has been properly established. See CNA v. United States, 535 F.3d 132, 144-45 (3d Cir.2008); Hamm v. United States, 483 F.3d 135, 137-38 (2d Cir.2007); see also Crawford v. United States, 796 F.2d 924, 927-28 (7th Cir.1986), abrogated on other grounds by United States v. Kwai Fun Wong, 575 U.S. ___, 135 S.Ct. 1625, 191 L.Ed.2d 533 (2015). Other circuits take the approach that the finding of subject-matter jurisdiction should be made together with a merits determination under Rule 56 when, as will often be the case, the two are sufficiently "intertwined." See Kerns v. United States, 585 F.3d 187, 192-94 (4th Cir.2009); Torres-Negrón v. J & N Records, LLC, 504 F.3d 151, 162-63 (1st Cir.2007); Montez v. Dep't of the Navy, 392 F.3d 147, 149-51 (5th Cir.2004); Augustine v. United States, 704 F.2d 1074, 1077-78 (9th Cir. 1983). Our precedent puts us squarely in the second camp. See Lawrence v. Dunbar, 919 F.2d 1525, 1530-31 (11th Cir.1990) (per curiam).

There is great underlying tension motivating these divergent approaches. On the one hand, the core principles of a limited federal judiciary and respect for sovereign entities' immunity from unconsented-to suits clearly cut against proceeding when jurisdiction remains uncertain. On the other hand, the additional procedural safeguards offered by Rule 56 not present in Rule 12(b)(1)—most notably, that the court is to view all facts and make all reasonable inferences "in the light most favorable to the nonmoving party," Hill v. Cundiff, 797 F.3d 948, 967 (11th Cir. 2015)—work to ensure that potentially meritorious but factually contested claims are not prematurely dismissed. How to strike the proper balance between these competing interests, as evidenced by the split in authority, is a difficult question indeed. Until the United States Supreme Court or the Eleventh Circuit sitting en banc rules otherwise, we are bound by our precedent. We must therefore remand for additional fact-finding on the many intertwined jurisdictional-merits questions raised by the record. The District Court should then assess its jurisdiction under the Rule 56 summary-judgment standard.

In this case and others like it, however, the fact-finding on remand should be carefully limited to avoid aggravating the Article III and sovereign-immunity problems discussed above. The outcome of this appeal is driven by its posture. The United States did not raise its motion to dismiss under Rule 12(b)(1), though it did request the District Court, in the alternative, to convert its motion to dismiss into one for summary judgment. The record demonstrates that the District Court addressed Douglas's pay-grade claim under the Rule 12(b)(6) motion-to-dismiss standard as a facial matter; the District Court refused to consider any evidence extrinsic to the pleadings, including the exhibits the United States attached to its motion. Had the United States also moved to dismiss for lack of subject-matter jurisdiction under Rule 12(b)(1) or had the District Court considered the evidence presented to it and made jurisdictional fact-findings on summary judgment under Rule 56, the result may well have been different.

Going forward, the District Court should make sufficient findings to resolve the jurisdictional-merits questions identified by the court's opinion.3 These findings may, but need not necessarily, extend beyond the exhibits the United States attached to its motion to dismiss. The District Court should then decide whether it has subject-matter jurisdiction over Douglas's suit in line with our decision in Lawrence, 919 F.2d at 1530-31. Though other cases may prove trickier to resolve and thus require relatively extensive discovery and jurisdictional fact-finding, this does not appear to be so here, given the straightforward nature of Douglas's pay-grade claim.

With that said, I concur fully in the court's opinion.