This is an interlocutory appeal from an order staying a habeas proceeding. We lack jurisdiction and dismiss the appeal.
Mr. Troy Kell sought habeas relief, but he had not exhausted two of his claims in state court. The unexhausted claims created a Catch-22 for Mr. Kell, risking a dismissal of all of his claims without an opportunity to timely refile. To relieve Mr. *451Kell of this Catch-22, the district court entered a limited stay, halting proceedings on one of the unexhausted claims while Mr. Kell returned to state court to exhaust the claim. For the remaining habeas claims, however, the district court continued with the proceedings.
In the midst of the ongoing habeas proceedings in district court, Utah appealed from the grant of a stay, arguing that the district court should have declined to grant a stay. Our threshold question involves appellate jurisdiction. To establish jurisdiction, Utah relies on the collateral-order doctrine, which allows appeals from some decisions before the entry of a final judgment. But the district court's issuance of a stay does not satisfy the collateral-order doctrine's requirements, so we dismiss the appeal for lack of appellate jurisdiction.
1. Mr. Kell timely files a habeas petition.
Mr. Kell was convicted of murder and sentenced to death in Utah, and his conviction became final roughly sixteen years ago. Mr. Kell then had one year to seek federal habeas relief, but the one-year limitations period was tolled while he pursued state post-conviction remedies. 28 U.S.C. § 2244(d). When the state post-conviction proceedings ended in 2009, Mr. Kell timely filed a federal habeas petition.
2. The district court stays the habeas case to allow Mr. Kell to exhaust a new claim.
In 2013, Mr. Kell asserted two new habeas claims: (1) that the trial court had improperly commented to the jury that Mr. Kell bore the burden in the penalty phase to show that his life should be spared and (2) that the jurors had improperly considered extraneous information. Mr. Kell had not exhausted the two new claims, so the district court needed to grapple with how to proceed. Continuing with the new habeas claims could prevent consideration of any of the claims because a federal district court must ordinarily dismiss the entire petition when one or more of the habeas claims are unexhausted. Rose v. Lundy , 455 U.S. 509, 522, 102 S.Ct. 1198, 71 L.Ed.2d 379 (1982). Given the possibility of dismissal, Mr. Kell faced a dilemma: If the district court were to dismiss the habeas petition and he later refiled in federal court, the statute of limitations might have expired on all of his claims.
To avoid this dilemma, Mr. Kell requested a stay so that he could exhaust his new habeas claims in state court. For this request, Mr. Kell invoked a procedure adopted in Rhines v. Weber , 544 U.S. 269, 125 S.Ct. 1528, 161 L.Ed.2d 440 (2005). Under Rhines , a district court may stay habeas proceedings to permit exhaustion of a claim upon satisfaction of three elements:
1. "Good cause" exists for the failure to exhaust the claim.
2. The unexhausted claim is "potentially meritorious."
3. The petitioner did not engage in "abusive litigation tactics" or intentionally delay the proceedings.
Rhines , 544 U.S. at 277-78, 125 S.Ct. 1528. The district court declined to stay the claim involving extraneous influence on the jury, concluding that this claim lacked potential merit. But the district court granted the stay on the claim involving the judge's comment to the jury, concluding that Mr. Kell had satisfied the three elements for a Rhines stay. For the remaining habeas claims, however, the district court stated that the proceedings would *452continue without interruption.1
3. Utah appeals the order granting a limited stay.
In this appeal, Utah argues that the federal district court erred in granting the stay because
• the court used the wrong test for "good cause" and misapplied that test,
• the new habeas claim lacks potential merit based on timeliness, the existence of a procedural default, and the absence of a constitutional violation, and
• Mr. Kell was dilatory by waiting over three years to assert the new habeas claim and over eight years to seek a stay based on this claim.
4. We lack jurisdiction to consider interlocutory appeals from Rhines stays.
We can consider these arguments only if Utah establishes appellate jurisdiction. See EEOC v. PJ Utah, L.L.C. , 822 F.3d 536, 542 n.7 (10th Cir. 2016) ("[T]he appellant ... bears the burden to establish appellate jurisdiction."). We typically acquire jurisdiction through the district court's entry of a final decision. 28 U.S.C. § 1291. But a stay does not ordinarily constitute a final decision. See Crystal Clear Commc'ns, Inc. v. Sw. Bell Tel. Co. , 415 F.3d 1171, 1176 (10th Cir. 2005) ("If a stay merely delays litigation and does not effectively terminate proceedings, it is not considered a final decision.").
Utah argues that we nonetheless have jurisdiction under the collateral-order doctrine. This doctrine would apply only if the district court's decision
• conclusively decided the disputed question,
• resolved an important issue separate from the merits, and
• could not be effectively reviewed on direct appeal.
Van Cauwenberghe v. Biard , 486 U.S. 517, 522, 108 S.Ct. 1945, 100 L.Ed.2d 517 (1988). The failure to satisfy any of the three elements would prevent us from applying the collateral-order doctrine. See Stubblefield v. Windsor Capital Grp. , 74 F.3d 990, 997 (10th Cir. 1996) (stating that the collateral-order doctrine "does not apply unless each of the three requirements are met").
Each element is considered stringent. E.g. , Flanagan v. United States , 465 U.S. 259, 270, 104 S.Ct. 1051, 79 L.Ed.2d 288 (1984). And "[i]n case after case in year after year, the Supreme Court has issued increasingly emphatic instructions that the class of cases capable of satisfying this 'stringent' test should be understood as 'small,' 'modest,' and 'narrow.' " United States v. Wampler , 624 F.3d 1330, 1334 (10th Cir. 2010) (Gorsuch, J.). Utah bears the burden on each element of this "stringent" test. See Los Lobos Renewable Power, L.L.C. v. Americulture, Inc. , 885 F.3d 659, 664 (10th Cir. 2018) (stating that the "party asserting jurisdiction under the collateral order doctrine" bears the burden on each element).
We assume, for the sake of argument, that an order issuing a Rhines stay conclusively determines the disputed question. But the grant of a Rhines stay involves issues that are intertwined with the merits and reviewable on direct appeal. We thus lack jurisdiction under the collateral-order doctrine. See *453Crystal Clear Commc'ns, Inc. v. Sw. Bell Tel. Co. , 415 F.3d 1171, 1178 (10th Cir. 2005) (noting that the collateral-order doctrine does not support appellate jurisdiction if any element is unsatisfied).
A. The grant of a Rhines stay is not completely separate from the merits.
The collateral-order doctrine applies only when the order involves an important issue that is not intertwined with the merits. Coopers & Lybrand v. Livesay , 437 U.S. 463, 469, 98 S.Ct. 2454, 57 L.Ed.2d 351 (1978). For this element, Utah must show that the issues bearing on the appropriateness of a Rhines stay are "completely separate" from the merits. E.g. , Cunningham v. Hamilton Cty., Ohio , 527 U.S. 198, 205, 119 S.Ct. 1915, 144 L.Ed.2d 184 (1999). Utah has not made this showing.
1. Avoidance of Piecemeal Litigation
The requirement of complete separation is designed to prevent piecemeal appellate review. Van Cauwenberghe v. Biard , 486 U.S. 517, 527, 108 S.Ct. 1945, 100 L.Ed.2d 517 (1988). Given this purpose, the collateral-order doctrine would ordinarily apply only if an appellate court would probably not need to consider the merits a second time.2
But interlocutory appeals of Rhines stays would often require federal appellate courts to consider the merits at least twice:
• once in the interlocutory appeal (when the respondent argues that a Rhines stay is improper because the petitioner's unexhausted claim lacks potential merit) and
• again after entry of the judgment (when the parties disagree over the claim's actual merit).
And if the district court enters multiple Rhines stays,3 we could face three or more appeals with overlapping issues.
This possibility is apparent here. For example, consider Utah's argument that the district court's jury instruction was correct. This argument involves a classic issue on the merits. See Gillette v. Prosper , 858 F.3d 833, 839 (3d Cir. 2017) (concluding that the order being appealed was "closely related" to the merits because "both concern[ed] alleged constitutional violations" and involved potential release from prison). If we were to recognize appellate jurisdiction at this stage and Mr. Kell were to obtain habeas relief, we would decide the "potential merit" of Mr. Kell's new claim now and the claim's "actual merit" after the entry of a final judgment. It is hard to imagine a better example of piecemeal litigation-precisely what the Supreme Court has tried to avoid by limiting the collateral-order doctrine to classes of orders involving "complete separation" from the merits.
The dissent disagrees:
Even where a district court issues multiple Rhines stays, each Rhines stay concerns different claims by a petitioner and therefore different issues. To illustrate, a district court issues a Rhines *454stay on claim x , allowing the petitioner to exhaust the claim in state court. After the petitioner exhausts claim x in state court and returns to federal court, the district court is not going to issue another Rhines stay for the purpose of allowing the petitioner to exhaust claim x . If there is a second Rhines stay, it would be issued for the petitioner to exhaust claim y . If both of these Rhines stays are appealed, and then the final judgment is appealed, the "same issues" would not be before this Court three or more times.
Dissent at 475 n.5 (emphasis in original).
We respectfully think that the dissent has misunderstood us. When reviewing a Rhines stay, we consider the "potential merit" of the unexhausted claim. For example, let's consider Utah's argument that the district court erred in granting a stay because the unexhausted habeas claim lacks potential merit. If we have appellate jurisdiction under the collateral-order doctrine, we would consider whether the district court acted within its discretion in treating the unexhausted claim as potentially meritorious.
Let's assume, for the sake of argument, that we were to uphold this determination. When the habeas case ends in district court, the parties could appeal the district court's ultimate determination of the claim's actual merit. See Alexander v. U.S. Parole Comm'n , 514 F.3d 1083, 1087 (10th Cir. 2008) (stating that a conditional writ of habeas corpus is final, creating appellate jurisdiction); Burton v. Johnson , 975 F.2d 690, 693-94 (10th Cir. 1992) (stating that a conditional writ of habeas corpus was an appealable final judgment).
The same would be true for any case involving a Rhines stay. So in a case with multiple Rhines stays and multiple interlocutory appeals, we could face the same issues after a final judgment (even if the interlocutory appeals individually involved different issues).
2. The Significance of the Relationship Between the Rhines Factor of "Potential Merit" and the Actual Merits of Mr. Kell's New Claim
Because "potential merit" is essential for Rhines stays, interlocutory review would frequently require us to consider the potential merit of the underlying habeas claims. And "potential merit" is obviously not "completely separate" from the actual merits. The dissent agrees, stating that the Rhines issue ("potential merit") "undoubtedly overlaps with the merits." Dissent at 475.4
3. Application of the Test of "Complete Separation"
To apply the element of complete separation, we must evaluate the pertinent *455issue based on the entire class of orders ( Rhines stays) rather than the particular arguments in this appeal. United States v. Bolden , 353 F.3d 870, 876 (10th Cir. 2003).5 We thus consider whether the issues underlying a Rhines stay are, "as a whole," completely separate from the merits. Id.
The Supreme Court has repeatedly cautioned that the collateral-order doctrine requires "complete separation" from the merits. E.g. , Microsoft Corp. v. Baker , --- U.S. ----, 137 S. Ct. 1702, 1708 n.3, 198 L.Ed.2d 132 (2017) ; Will v. Hallock , 546 U.S. 345, 349, 126 S.Ct. 952, 163 L.Ed.2d 836 (2006) ; Sell v. United States , 539 U.S. 166, 176, 123 S.Ct. 2174, 156 L.Ed.2d 197 (2003) ; Cunningham v. Hamilton Cty., Ohio , 527 U.S. 198, 205, 119 S.Ct. 1915, 144 L.Ed.2d 184 (1999) ; Quackenbush v. Allstate Ins. Co. , 517 U.S. 706, 712, 116 S.Ct. 1712, 135 L.Ed.2d 1 (1996) ; Johnson v. Jones , 515 U.S. 304, 310-11, 115 S.Ct. 2151, 132 L.Ed.2d 238 (1995) ; Digital Equip. Corp. v. Desktop Direct, Inc. , 511 U.S. 863, 867, 114 S.Ct. 1992, 128 L.Ed.2d 842 (1994) ; Richardson-Merrell, Inc. v. Koller , 472 U.S. 424, 431, 105 S.Ct. 2757, 86 L.Ed.2d 340 (1985) ; Coopers & Lybrand v. Livesay , 437 U.S. 463, 468-69, 98 S.Ct. 2454, 57 L.Ed.2d 351 (1978).
4. Utah's Focus on its Own Arguments (Rather than the Class of Orders)
Utah also points to its arguments involving good cause, insisting that they do not relate to the merits. And the dissent relies on Utah's arguments involving timeliness, dilatoriness, and procedural default. The focus of Utah and the dissent on Utah's particular appeal points is misguided, as it disregards
• the overlap between Utah's appellate arguments and the merits and
• the need to consider separation categorically based on the class of orders rather than the particular issues invoked by the appellant.
a. Utah's Arguments on Good Cause
Utah tries to justify application of the collateral-order doctrine based on the district court's alleged misidentification and misapplication of the test for good cause. We reject this effort.
i. Misidentifying the Test for Good Cause
Utah identifies the definition of good cause as an "important" issue. To determine an issue's importance under the collateral-order doctrine, we consider
• whether the issue is "important in a jurisprudential sense"6 and
• whether the interests that " 'would potentially go unprotected without immediate appellate review are significant relative to efficiency interests *456sought to be advanced by adherence to the final judgment rule.' "7
The definition of good cause might be considered jurisprudentially important now because we lack a precedent squarely defining the test for good cause under Rhines . Of course, if we were to undertake interlocutory review and define the test, that definition would settle the issue, rendering it jurisprudentially unimportant in the future. See 15A Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice & Procedure: Jurisdiction & Related Matters § 3911.5, at 438 (2d ed. 1992) ("Implementation of a serious and unsettled question requirement can easily lead to a situation in which a particular question is suitable for collateral order appeal the first time it is presented, but not thereafter.").
And Utah has not shown an urgency to immediately define the standard for good cause under Rhines . Indeed, Utah has offered many alternative arguments for reversal, such as expiration of the limitations period, procedural default, misapplication of the district court's own test for good cause, dilatoriness, and lack of potential merit. Given Utah's alternative arguments, we may not even need to decide the test for good cause if we were to reverse the grant of a Rhines stay.
The dissent nonetheless insists that the test for good cause is important because a court in "literally every grant of a Rhines stay, not just this particular grant of a Rhines stay, must decide what the appropriate standard for determining 'good cause' is." Dissent at 474. We see the issue differently. If the collateral-order doctrine applied, only two possibilities exist:
1. We decide now what the test is for good cause.
2. We do not decide the test now, ruling on other grounds.
Let's consider the first possibility (that we decide the test for good cause in this appeal). Once we define the test, future panels and district courts will be bound by stare decisis to apply that test in all future cases. United States v. Meyers , 200 F.3d 715, 720 (10th Cir. 2000). We thus couldn't continue to view the definition of good cause as jurisprudentially significant once we define the test for good cause.
The second possibility is that we avoid defining the test for good cause. This possibility is real because we could reverse a Rhines stay based on any of the three prongs. For example, we might reverse the grant of a Rhines stay based on dilatoriness or a lack of potential merit rather than address the issue of good cause.8 But if we don't decide the test for good cause, how could we apply the collateral-order doctrine based on the importance of deciding that test? If we were to do so, we could put ourselves in the odd position of recognizing appellate jurisdiction to decide the test for good cause without actually deciding what the test is.
*457Given these two possibilities, application of the collateral-order doctrine based on the test for good cause would paradoxically allow every respondent to appeal every grant of a Rhines stay based on an issue that we have already decided or, if not, might not even need to decide. This paradox highlights the need to consider the importance of the issues categorically based on the class of orders involved rather than the particular arguments raised in a particular case.9
ii. Misapplying the Test for Good Cause
On the merits of the stay, Utah argues that the district court misapplied the test for good cause. Utah's argument shows that application of the good cause test not only overlaps with the merits but would often require consideration of them. For example, petitioners could sometimes raise the same theory for good cause and the merits of a habeas claim. An example is a Brady claim. See Brady v. Maryland , 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963). Courts have considered withholding of information as "good cause." See Doe v. Jones , 762 F.3d 1174, 1182 (10th Cir. 2014). So when a petitioner asserts a Brady claim, the petitioner might rely on the withholding of evidence for both good cause and the merits of the habeas claim. See, e.g. , Wogenstahl v. Mitchell , 668 F.3d 307, 322 (6th Cir. 2012) (concluding that a habeas petitioner "had good cause for failing to raise the Brady issue prior to 2003, because the new information about [a prosecution witness's prior arrest and adjudication of delinquency] was not disclosed until then").
An overlap also exists between the inquiry on good cause and the merits of other habeas claims. For example, the inquiry on good cause may overlap with the merits when a petitioner alleges ineffective assistance of trial counsel. Suppose that a petitioner alleges good cause based on post-conviction counsel's failure to assert ineffective assistance on the part of trial counsel. This allegation could suffice to avoid a procedural default. Martinez v. Ryan , 566 U.S. 1, 9, 132 S.Ct. 1309, 182 L.Ed.2d 272 (2012). And Utah argues that the tests for good cause and procedural default are the same. Given this argument, we might find "good cause" based on post-conviction counsel's ineffectiveness for failure to assert trial counsel's ineffectiveness. And once the case ends in district court, we could again face the issue of trial counsel's ineffectiveness as part of our review on the merits.
The Supreme Court addressed this kind of overlap in Van Cauwenberghe v. Biard , 486 U.S. 517, 108 S.Ct. 1945, 100 L.Ed.2d 517 (1988), and Coopers & Lybrand v. Livesay , 437 U.S. 463, 98 S.Ct. 2454, 57 L.Ed.2d 351 (1978). In Van Cauwenberghe , the Court held that the denial of a motion to dismiss under the doctrine of forum non conveniens was not separate from the merits because the doctrine involves factors (like the location of the witnesses and evidence) that are enmeshed with the merits. 486 U.S. at 528-29, 108 S.Ct. 1945. And in Coopers & Lybrand , the Court held that consideration of class certification involved issues intertwined with the merits because *458class certification involves merits-related matters like the typicality of the claims, the adequacy of a representative, and the presence of common questions of law or fact. 437 U.S. at 469 & n.12, 98 S.Ct. 2454. Thus, the Court held that the collateral-order doctrine does not cover denials of class certification. Id . at 469, 98 S.Ct. 2454. Neither case involved actual consideration of the merits; but in both cases, the issues overlapped with the merits.
Like issues involving forum non conveniens and class certification, the issue of good cause will often overlap with a court's preliminary assessment of the merits, preventing application of the collateral-order doctrine. See Cunningham v. Hamilton Cty., Ohio , 527 U.S. 198, 205, 119 S.Ct. 1915, 144 L.Ed.2d 184 (1999) (holding that sanctions orders under Fed. R. Civ. P. 37(a) do not fall under the collateral-order doctrine because they "often will be inextricably intertwined with the merits of the action," "[m]uch like the orders at issue in Van Cauwenberghe and Coopers & Lybrand ").
Here, for example, Utah challenges the finding of good cause, arguing that Mr. Kell could have asserted the claim at trial and in a direct appeal. To address this argument, we would need to ask whether Mr. Kell's attorneys should have asserted the claim at trial or in the direct appeal. This inquiry would presumably overlap with Mr. Kell's habeas claims of ineffective assistance of counsel. Thus, resolution of Utah's arguments on good cause could entangle us in the substance of Mr. Kell's underlying habeas claims. In these circumstances, consideration of good cause under Rhines is not categorically separate from the merits.10
b. Timeliness, Dilatoriness, and Procedural Default
In this appeal, Utah also argues that (1) Mr. Kell was dilatory and (2) his unexhausted claim is untimely and procedurally defaulted. Utah does not suggest that these arguments would support collateral-order jurisdiction. But the dissent does, stating that timeliness, dilatoriness, and procedural default are separate from the merits.
For the sake of argument, let's assume that the dissent is right and ignore the fact that Utah (the appellant invoking appellate jurisdiction) didn't present these arguments as grounds to invoke the collateral-order doctrine. See pp. 465-66, below (citing cases for the unavailability of sua sponte arguments to support jurisdiction). As Utah argues, however, the separate issue must also be considered important in order to trigger this doctrine. See p. 452, above. Neither Utah nor the dissent offers *459any argument about the importance of Utah's arguments on timeliness, dilatoriness, or procedural default. These arguments involve garden-variety application of legal principles settled long ago. See U.S. Fidelity & Guaranty Co. v. Arch Ins. Co. , 578 F.3d 45, 59 (1st Cir. 2009) (concluding that an issue involving insufficiency of an affidavit was not "important enough ... to justify the application of the collateral order doctrine"); United States v. Billmyer , 57 F.3d 31, 35 (1st Cir. 1995) ("Ordinarily, a discovery order will meet the legal-importance test only if it presents a claim of clear-cut legal error and not merely a challenge to the district judge's factual determinations or the application of settled legal rule to the particular facts."). Thus, Utah's arguments on timeliness, dilatoriness, and procedural default would not constitute important issues, as required to trigger the collateral-order doctrine.
c. The Categorical Approach
Perhaps in some appeals of Rhines stays, the specific argument being advanced might not involve the potential merit of an unexhausted claim. But the Supreme Court has "consistently eschewed a case-by-case approach to deciding whether an order is sufficiently collateral." Cunningham v.Hamilton Cty., Ohio , 527 U.S. 198, 206, 119 S.Ct. 1915, 144 L.Ed.2d 184 (1999) ; see also pp. 454-55 & n.5 (discussing United States v. Bolden , 353 F.3d 870, 876 (10th Cir. 2003) ).
The wisdom of the Supreme Court's approach is self-evident. For example, let's assume that Utah's appellate arguments are important and completely separate from the merits. If we were to base the collateral-order doctrine solely on the happenstance of what Utah argues in a given case, we would be allowing or disallowing interlocutory appeals of all Rhines stays based on what a single party has chosen to argue in a single case. To avoid this anomaly, the Supreme Court has required us to consider the underlying class of orders rather than the peculiarities of the arguments presented by this particular appellant. See pp. 454-55, above. So here we consider the entire category of orders. See id.
The dissent apparently agrees, stating that we must "look[ ] to the issues the class of orders ( Rhines stays) generally raise." Dissent at 473. And the dissent acknowledges that we are not to focus on "case-specific issues," such as the issues presented in this particular order granting a Rhines stay. Id.
With this acknowledgment, the dissent contends that the appellant must present "at least one issue " that is "completely separate from the merits." Id. at 472 (emphasis in original). But the Supreme Court has consistently rejected this approach, holding that even if some appeals of Rhines stays might involve appellate issues separable from the merits, the collateral-order doctrine cannot apply absent complete separation for the entire class of orders. Richardson-Merrell, Inc. v. Koller , 472 U.S. 424, 439, 105 S.Ct. 2757, 86 L.Ed.2d 340 (1985) ;11 see also *460Cunningham v. Hamilton Cty., Ohio , 527 U.S. 198, 206, 119 S.Ct. 1915, 144 L.Ed.2d 184 (1999) ("Perhaps not every discovery sanction will be inextricably intertwined with the merits, but we have consistently eschewed a case-by-case approach to deciding whether an order is collateral."). So when we assess whether the class of orders entails complete separation from the merits, we consider the issues in all grants of stays under Rhines . The presence of a single issue separate from the merits in a particular appeal would not trigger the collateral-order doctrine. See Cunningham , 527 U.S. at 205-06, 119 S.Ct. 1915 (holding that sanctions orders under Fed. R. Civ. P. 37(a) are categorically considered intertwined with the merits even though sanctions orders are sometimes separate from the merits); Koller , 472 U.S. at 439-40, 105 S.Ct. 2757 (holding that orders disqualifying counsel in civil cases are categorically considered intertwined with the merits because many are even though some aren't); see also In re Continental Investment Corp. , 637 F.2d 1, 6 (1st Cir. 1980) (holding "that denials of disqualification motions as a class raise significant legal questions too rarely to bring them within the Cohen exception").
The categorical approach is apparent in the Supreme Court's handling of qualified immunity. There the Court has held that legal issues involving a clearly established violation are completely separate from the merits. Mitchell v. Forsyth , 472 U.S. 511, 528, 105 S.Ct. 2806, 86 L.Ed.2d 411 (1985). But the Supreme Court made that determination by considering the issues arising in the entire class of orders rather than in the particular case being reviewed. See id. at 524-30, 105 S.Ct. 2806.
The Supreme Court later went further, expressly disavowing a case-by-case approach in Johnson v. Jones , 515 U.S. 304, 115 S.Ct. 2151, 132 L.Ed.2d 238 (1995). There the Court refused to extend the collateral-order doctrine to cover denials of qualified immunity based on insufficiency of the evidence. Johnson , 515 U.S. at 314-15, 115 S.Ct. 2151. The Court explained that under the categorical approach, appellate judges could not consider the particular facts to determine the applicability of the collateral-order doctrine. Id. at 315, 115 S.Ct. 2151. Instead, judges were to consider the kinds of issues likely to arise in appeals involving this class of orders. Id. at 315-17, 115 S.Ct. 2151. For example, the Supreme Court acknowledged that the facts in Johnson itself were straightforward, but stressed that many cases involving constitutional torts would entail factual disputes ill-suited for review under the collateral-order doctrine. Id. at 316, 115 S.Ct. 2151.
Beyond qualified immunity, circuit courts have followed the Supreme Court's lead by deciding the element of complete separation based on the class of orders involved rather than approaching this element based on the particular issues raised by a particular appellant. See Henry v. Lake Charles Am. Press, L.L.C. , 566 F.3d 164, 173 (5th Cir. 2009) ("Consistent with Supreme Court precedent and the general purposes of the final judgment rule, we determine whether an order is appealable as a general or categorical matter."); see also In re Carco Electronics , 536 F.3d 211, 213 (3d Cir. 2008) ("[T]he Supreme Court's statement in Cunningham that we should not apply the collateral order doctrine on a 'case-by-case' basis indicates that we should not attempt to carve out individualized, case-specific exceptions to the general rule that discovery orders are not immediately appealable."); Abelesz v. OTP Bank , 692 F.3d 638, 649 (7th Cir. 2012)
*461("The Supreme Court has applied the collateral order rule categorically, treating different sorts of defenses or issues as either covered or not covered.").
* * *
Viewed as a category, interlocutory appeals of Rhines stays would generally enmesh us in the merits. Thus, the collateral-order doctrine's second element is not met when the district court grants a Rhines stay.
B. The grant of a Rhines stay can be reviewed in the appeal from a final judgment.
Utah fails to satisfy not only the collateral-order doctrine's second element but also the third element.
Under this element, appellate jurisdiction exists only if the issue is important and could not otherwise be effectively reviewed after the entry of final judgment. Mitchell v. Forsyth , 472 U.S. 511, 525, 105 S.Ct. 2806, 86 L.Ed.2d 411 (1985). But we do not just consider unreviewability; we also consider the importance of the interest lost by deferring review until after the final judgment. See Digital Equip. Corp. v. Desktop Direct, Inc. , 511 U.S. 863, 878-79, 114 S.Ct. 1992, 128 L.Ed.2d 842 (1994). This inquiry entails "a judgment about the value of the interests that would be lost through rigorous application of the final judgment requirement." Id. On this element, Utah hasn't satisfied its burden.
Utah argues that the grant of a Rhines stay is unreviewable after a final judgment because (1) the loss of time can never be remedied and (2) the grant of a stay becomes moot upon entry of a final judgment. We reject both arguments.
As Utah points out, the delay itself is unreviewable because a court can't restore Utah's lost time. But we do not ordinarily regard the loss of time as sufficiently important to trigger the collateral-order doctrine. See United States v. Section 17 Tp. 23 North, Range 22 East of IBM, Delaware Cty., Okla. , 40 F.3d 320, 322 (10th Cir. 1994) (holding that an interest in defending a forfeiture action now, as opposed to later, "is not the type of 'important' right which the Supreme Court contemplated as requiring immediate review in [ Digital Equipment Corp. v. Desktop Direct, Inc. , 511 U.S. 863, 114 S.Ct. 1992, 128 L.Ed.2d 842 (1994) ]"). After all, no pretrial decision can ever restore a party's lost time, and we routinely disallow interlocutory appeals for most pretrial decisions. See Mesa Oil, Inc. v. United States , 467 F.3d 1252, 1255 (10th Cir. 2006) ("The costs of unnecessary litigation caused by what eventually turns out to be an error by the district court is insufficient to warrant an interlocutory appeal."); see also In re Kozeny , 236 F.3d 615, 619-20 (10th Cir. 2000) (two-judge motions panel)12 ("If the stay merely delays the federal litigation, courts have generally held the stay orders not to be appealable."); accord Kershaw v. Shalala , 9 F.3d 11, 14 (5th Cir. 1993) ("Absent a Moses Cone situation, stay orders ... are usually not reviewable as collateral orders.");13 see also Hon. Edith H. Jones, *462Appeals of Arbitration Orders-Coming Out of the Serbonian Bog , 31 S. Tex. L. Rev. 361, 372 (1990) ("Absent exceptional circumstances, it is unlikely that an order granting a motion to stay proceedings should be appealable as a collateral order.").
Time is a precious commodity in habeas proceedings, particularly when the petitioner faces a death sentence (as Mr. Kell does). And until now, the case has lingered in state and federal courts for roughly a quarter of a century. But we cannot single out particular cases to decide the extent of the interest lost by deferring review. Instead, we must consider the public interest based on the class of orders ( Rhines stays). See Mohawk Indus., Inc. v. Carpenter , 558 U.S. 100, 107, 130 S.Ct. 599, 175 L.Ed.2d 458 (2009) ;14 In re Motor FuelTemperature Sales Practices Litig ., 641 F.3d 470, 482 (10th Cir. 2011) ("When analyzing the third Cohen requirement, we do not consider whether the circumstances of the particular case before us warrant our immediate review; rather, we examine whether the entire category of rulings to which the claim belongs can be adequately vindicated on review of a final judgment or by other means."). This approach requires us to decide whether the value that we place on immediate review would override Congress's policy against piecemeal review.
To answer, we must consider the many important interests that exist, often in tension with one another. For example, the state has important interests in comity and enforcement of its own criminal judgments, which Congress recognized in the Antiterrorism and Effective Death Penalty Act of 1996. See Williams v. Taylor , 529 U.S. 420, 436, 120 S.Ct. 1479, 146 L.Ed.2d 435 (2000) (stating that the Act was designed to advance "comity, finality, and federalism"). But habeas review inherently creates tension with comity, as federal courts review decisions by a state's highest court. See Duckworth v. Eagan , 492 U.S. 195, 211, 109 S.Ct. 2875, 106 L.Ed.2d 166 (1989) (discussing the tension of federal-state relations because of habeas review, where "lower federal courts often sit in 'review' of the judgments of the highest courts of a state judicial system").
To ease this tension, Congress has required habeas petitioners to exhaust state-court remedies. 28 U.S.C. § 2254(b)(1)(A) ; see also Rose v. Lundy , 455 U.S. 509, 518, 102 S.Ct. 1198, 71 L.Ed.2d 379 (1982) ("The exhaustion doctrine is principally designed to protect the state courts' role in the enforcement of federal law and prevent disruption of state judicial proceedings."). Exhaustion, however, can slow the state's ability to enforce criminal judgments, especially *463those involving capital sentences. See Ira P. Robbins, Toward a More Just & Effective System of Review in State Death Penalty Cases , 40 Am. U. L. Rev. 1, 131 (1990) ("[R]equiring exhaustion [in habeas review of death-penalty cases] usually necessitates further delay, even if the unexhausted claim is an obviously frivolous one."). This risk is particularly significant in habeas proceedings, where swift action is essential. See Rhodes v. Hannigan , 12 F.3d 989, 992 (10th Cir. 1993) ("Habeas corpus 'is a speedy remedy, entitled by statute to special, preferential consideration to insure expeditious hearing and determination.' " (quoting Van Buskirk v. Wilkinson , 216 F.2d 735, 737-38 (9th Cir. 1954) )); Johnson v. Rogers , 917 F.2d 1283, 1284 (10th Cir. 1990) ("[W]rits of habeas corpus are intended to afford a 'swift and imperative remedy in all cases of illegal restraint or confinement.' " (quoting Fay v. Noia , 372 U.S. 391, 400, 83 S.Ct. 822, 9 L.Ed.2d 837 (1963) )).
The Supreme Court sought to balance these competing interests in Rhines by giving discretion to federal district courts to stay habeas proceedings while a petitioner exhausts state-court remedies. Rhines v. Weber , 544 U.S. 269, 276-78, 125 S.Ct. 1528, 161 L.Ed.2d 440 (2005). But discretion can be abused. The question presented here is when an appellate court should review the district court's exercise of discretion.15 Given the conflicting interests, we must consider whether interlocutory review of Rhines stays would speed review or slow it through piecemeal review.
For this inquiry, we start with Congress's policy against piecemeal review. In general, Congress has tried to avoid piecemeal review by confining appellate review to final orders. See Abney v. United States , 431 U.S. 651, 656, 97 S.Ct. 2034, 52 L.Ed.2d 651 (1977) ("[T]here has been a firm congressional policy against interlocutory or 'piecemeal' appeals and courts have consistently given effect to that policy."). And Congress has expressed its judgment "that the district judge has the primary responsibility to police the prejudgment tactics of litigants," reasoning "that the district judge can better exercise that responsibility if the appellate courts do not repeatedly intervene to second-guess prejudgment rulings." Richardson-Merrell, Inc. v. Koller , 472 U.S. 424, 436, 105 S.Ct. 2757, 86 L.Ed.2d 340 (1985) (emphasis in original).
The Supreme Court has respected this congressional policy choice and restricted prejudgment review because it inefficiently fosters piecemeal appeals. See Mohawk Indus., Inc. v. Carpenter , 558 U.S. 100, 106, 130 S.Ct. 599, 175 L.Ed.2d 458 (2009) ("Permitting piecemeal, prejudgment appeals, we have recognized, undermines 'efficient judicial administration' and encroaches upon the prerogatives of district court judges, who play a 'special role' in managing ongoing litigation."); Richardson-Merrell, Inc. v. Koller , 472 U.S. 424, 430, 105 S.Ct. 2757, 86 L.Ed.2d 340 (1985) (stating that the final judgment rule prevents delays from frequent appellate interruptions *464in the litigation); Abney , 431 U.S. at 657, 97 S.Ct. 2034 (stating that 28 U.S.C. § 1291, which restricts appellate review to final decisions, is designed to prevent delays and disruptions through intermediate appeals).
Given Congress's preference against interlocutory review, we must consider whether Rhines stays involve aberrant circumstances justifying our intrusion into Congress's effort to avoid piecemeal review. We think not. In Rhines itself, the Supreme Court balanced the compelling interests by devising a procedure to accommodate
• the state's interest in comity,
• the congressional requirement of exhaustion of state-court remedies,
• the congressional determination that piecemeal review generally slows the litigation,
• the universal recognition that habeas cases should proceed expeditiously, and
• the congressional objective in the Antiterrorism and Effective Death Penalty Act to streamline habeas proceedings.
Rhines v. Weber , 544 U.S. 269, 276-78, 125 S.Ct. 1528, 161 L.Ed.2d 440 (2005) ; see Howard v. Norris , 616 F.3d 799, 802-03 (8th Cir. 2010) (stating that " Rhines effectively balances the state's interest in avoiding delay in a habeas proceeding, with the petitioner's (and the state's) interest in having habeas claims addressed first in state court"); see also Grace v. Vannoy , 826 F.3d 813, 819 (5th Cir. 2016) (stating that in Rhines , "the Supreme Court implicitly de-emphasized the importance of the state's interests in finality and speedy resolution of mixed federal petitions").
If we were to intervene after a district court granted a stay, would our intervention quicken or slow the litigation? Here, the district court stayed the proceedings for only a single claim and stated that the proceedings would continue on all of the other habeas claims. If we were to interject ourselves now, we could inadvertently trigger simultaneous litigation of the same case in three courts:
1. the state district court or the state appellate court,
2. the federal district court, and
3. our court.
We question the efficiency of duplicative litigation in three courts. See Swanson v. DeSantis , 606 F.3d 829, 833 (6th Cir. 2010) ("Claim-by-claim litigation in the district and appellate courts 'undermines [Congress'] goal of streamlining federal habeas proceedings....' " (emphasis and alteration in original) (quoting Rhines v. Weber , 544 U.S. 269, 277, 125 S.Ct. 1528, 161 L.Ed.2d 440 (2005) )).
But double or triple litigation tracks could create not only inefficiency but also more delay. For example, if we were to affirm the grant of a stay, we could be needlessly slowing the habeas litigation for the time that we take to decide the appeal. And, of course, district courts aren't limited to the number of Rhines stays in a single case. Here, for example, the district court has issued two Rhines stays. If those stays had triggered the collateral-order doctrine, we could have slowed the litigation twice already with the possibility of a third delay when the case ends in district court. In these circumstances, we decline to single out Rhines stays as a class of orders that would be resolved more quickly by authorizing piecemeal appeals.16
*465Our dissenting colleague disagrees, stating that this is not the relevant question. Dissent at 477-78. But we are simply addressing the arguments presented by Utah and our dissenting colleague. Both argue that interlocutory review is essential to prevent delays in enforcing the State's criminal judgment. Indeed, while challenging our characterization of the question, the dissent insists that "[t]he delay-which is, in itself, a win for Petitioner-is exactly what harms the State's interest." Id. at 478. In short, the arguments by Utah and our dissenting colleague require us to consider whether piecemeal review would truly promote the State's interest in expeditious enforcement of its criminal judgments or cause more delay.
Utah and the dissent also argue that waiting for a final judgment might prevent any appeal of an order granting a Rhines stay.17 They apparently assume that Rhines stays will always become moot even when they ultimately lead to a grant of habeas relief. But neither Utah nor the dissent says why a Rhines stay would always become moot with a grant of habeas relief.18 See Thompson v. Frank , 599 F.3d 1088, 1090 (9th Cir. 2010) (declining to apply the collateral-order doctrine to a Rhines stay because it would be reviewable after the final judgment). And we don't ordinarily construct arguments in support of appellate jurisdiction. See United States ex rel. Ramseyer v. Century Healthcare Corp. , 90 F.3d 1514, 1518 n.2 (10th Cir. 1996) ("Our duty to consider unargued obstacles to subject matter jurisdiction does not affect our discretion to decline to consider waived arguments that might have supported such jurisdiction." (emphasis in original)); see also Raley v. Hyundai Motor Co. , 642 F.3d 1271, 1275 (10th Cir. 2011) (Gorsuch, J.) ("It is the appellant's burden, not ours, to conjure up possible theories to invoke our legal authority to hear her appeal."). Given our reluctance to construct arguments for appellate jurisdiction, we cannot assume-as Utah and the dissent do-that review of a Rhines stay would become moot if Mr. Kell were to obtain habeas relief.
Even if review of the grant of a Rhines stay would eventually become moot, however, the final judgment would certainly be appealable. See *466Alexander v. U.S. Parole Comm'n , 514 F.3d 1083, 1097 (10th Cir. 2008) (stating that a conditional writ of habeas corpus is final, creating appellate jurisdiction); Burton v. Johnson , 975 F.2d 690, 693-94 (10th Cir. 1992) (stating that a conditional writ of habeas corpus was an appealable final judgment). Because the district court's ultimate rulings on the habeas claims would be reviewable after the final judgment, the collateral-order doctrine's third element would remain unsatisfied even if the grant of a Rhines stay were to become moot. See Bean v. Dormire , 10 F.3d 538, 539 (8th Cir. 1993) (per curiam) (holding that the third element of the collateral-order doctrine is unsatisfied in a case involving the stay of a prisoner's § 1983 action because the prisoner's underlying claims would be reviewable later even if the stay itself were otherwise unreviewable). The only difference is that waiting for a final judgment would postpone the appeal. But as discussed above, the delay results from a congressional policy choice. See pp. 462-65, above.
* * *
The issues in Rhines stays are not categorically separate from the merits and can be effectively reviewed in an appeal from a final judgment. So two of the three elements of the collateral-order doctrine are absent, precluding appellate jurisdiction.
C. The Supreme Court's consideration of the merits in Rhines does not support jurisdiction here.
Finally, Utah contends that appellate jurisdiction is supported by the Supreme Court's decision in Rhines to reach the merits. This contention is based on two steps:
1. Rhines v. Weber was appealed to the Eighth Circuit as a collateral order. See Rhines v. Weber , 346 F.3d 799, 800 (8th Cir. 2003) (per curiam).
2. If the Eighth Circuit had lacked jurisdiction in Rhines , the Supreme Court would not have reached the merits.
We reject this contention.
In Rhines , appellate jurisdiction was not mentioned in any of the briefs or in the Supreme Court's opinion. And "[w]hen a potential jurisdictional defect is neither noted nor discussed in a federal decision, the decision does not stand for the proposition that no defect existed." Ariz. Christian School Tuition Org. v. Winn , 563 U.S. 125, 144, 131 S.Ct. 1436, 179 L.Ed.2d 523 (2011).
But it wouldn't matter here even if the Supreme Court had implicitly blessed application of the collateral-order doctrine in Rhines . Before Rhines was appealed to the Eighth Circuit, that court's opinion in Carmichael v. White had allowed stays of mixed habeas petitions only in "truly exceptional circumstances." Carmichael v. White , 163 F.3d 1044, 1045 (8th Cir. 1998). So when Rhines was appealed to the Eighth Circuit, the court applied the test that had been used in Carmichael . See Rhines v. Weber , 346 F.3d 799, 800 (8th Cir. 2003) (per curiam) (citing Carmichael , 163 F.3d at 1045 ).
When Rhines later went to the Supreme Court, the parties disagreed over the validity of the Eighth Circuit's test and the extent of a district court's authority to stay, consider, or dismiss mixed habeas petitions. See Rhines v. Weber , 544 U.S. 269, 273, 125 S.Ct. 1528, 161 L.Ed.2d 440 (2005). The Supreme Court resolved this disagreement in Rhines by creating a new test for stays when some of the habeas claims are unexhausted. See id. at 277-78, 125 S.Ct. 1528.
Given the Supreme Court's creation of a new test in Rhines , our consideration of Utah's appeal would entail
*467• the application of the Rhines factors rather than the previous Eighth Circuit test and
• the appropriateness of a stay rather than the district court's authority to stay, consider, or dismiss mixed habeas petitions.
These issues did not exist when Rhines was appealed to the Eighth Circuit. So even if the Supreme Court had silently concluded that the collateral-order doctrine applies to stays granted under the test in Carmichael , that conclusion would not apply to an appeal addressing the appropriateness of a stay based on the Rhines factors. See Howard v. Norris , 616 F.3d 799, 802-03 (8th Cir. 2010) (noting that the Eighth Circuit's pre- Rhines rationale for applying the collateral-order doctrine to stays of habeas proceedings was "no longer applicable"). As a result, Rhines 's procedural posture supplies no meaningful guidance on our jurisdictional issue.
5. Conclusion
We lack appellate jurisdiction. A Rhines stay is not a final decision, and two elements of the collateral-order doctrine are not met. This doctrine applies only when the order conclusively decides an important question, separate from the merits, that would be effectively unreviewable in a direct appeal from a final judgment. But when Rhines stays are viewed categorically, the issues are generally inseparable from the merits and reviewable after entry of the final judgment. We thus lack appellate jurisdiction and dismiss the appeal.