EASTERBROOK, Circuit Judge.

In 1992, at the suggestion of the Federal Courts Study Committee, Congress authorized the Supreme Court to issue rules that expand the set of allowable interlocutory appeals. 28 U.S.C. § 1292(e). An earlier grant of jurisdictional rulemak-ing power-28 U.S.C. § 2072(c), which permits the Court to “define when a ruling of a district court is final for the purposes of appeal under section 1291” — had gone unused, in part because it invites the question whether a particular rule truly “defines” or instead expands appellate jurisdiction. Section 1292(e) expressly authorizes expansions. So far, it has been employed once. Last year the Supreme Court promulgated Fed.R.Civ.P. 23(f), which reads:

A court of appeals may in its discretion permit an appeal from an order of a district court granting or denying class action certification under this rule if application is made to it within ten days after entry of the order. An appeal does not stay proceedings in the district court unless the district judge or the court of appeals so orders.

This rule became effective on December 1, 1998, and we have for consideration the first application filed in this circuit (and, so far as we can tell, the nation) under the new rule. A motions panel directed the parties to file briefs discussing the standard the court should employ to decide whether to accept appeals under this rule.

The Committee Note accompanying Rule 23(f) remarks: “The court of appeals is given unfettered discretion whether to permit the appeal, akin to the discretion exercised by the Supreme Court in acting on a petition for certiorari.... Permission to appeal may be granted or denied on the basis of .any consideration that the court of appeals finds persuasive.” (The parties call this an “Advisory Committee Note,” following old usage, but its title was changed more than a decade ago to “Committee Note.” It speaks not only for the responsible advisory committee but also for the Standing Committee on Rules of Practice and Procedure, which coordinates and superintends the several bodies of federal rules.) Although Rule 10 of the Su*834preme Court’s Rules identifies some of the considerations that inform the grant of certiorari, they are “neither controlling nor fully measuring the Court’s discretion”. Likewise it would be a mistake for us to draw up a list that determines how the power under Rule 23(f) will be exercised. Neither a bright-line approach nor a catalog of factors would serve well — especially at the outset, when courts necessarily must experiment with the new class of appeals.

Instead of inventing standards, we keep in mind the reasons Rule 23(f) came into being. These are three. For some cases the denial of class status sounds the death knell of the litigation, because the representative plaintiffs claim is too small to justify the expense of litigation. Coopers & Lybrand v. Livesay, 437 U.S. 463, 98 S.Ct. 2454, 57 L.Ed.2d 351 (1978), held that an order declining to certify a class is not appealable, even if that decision dooms the suit as a practical matter. Rule 23(f) gives appellate courts discretion to entertain appeals in “death knell” cases— though we must be wary lest the mind hear a bell that is not tolling. Many class suits are prosecuted by law firms with portfolios of litigation, and these attorneys act as champions for the class even if the representative plaintiff would find it uneconomical to carry on with the case. E.g., Rand v. Monsanto Co., 926 F.2d 596 (7th Cir.1991). These law firms may carry on in the hope of prevailing for a single plaintiff and then winning class certification (and the reward of larger fees) on appeal, extending the victory to the whole class. A companion appeal, briefed in tandem with this' one, presented just such a case. After class certification was denied, the plaintiff sought permission to appeal under Rule 23(f); although the remaining plaintiff has only a small stake, counsel pursued the case in the district court while we decided whether to entertain the appeal, and before the subject could be argued here the district judge granted summary judgment for the defendant. That plaintiff now has appealed on the merits and will seek to revive the class to boot. Many other cases proceed similarly; Coopers & Lybrand did not wipe out the small-stakes class action. But when denial of class status seems likely to be fatal, and when the plaintiff has a solid argument in opposition to the district court’s decision, then a favorable exercise of appellate discretion is indicated.

Second, just as a denial of class status can doom the plaintiff, so a grant of class status can 'put considerable pressure on the defendant to settle, even when the plaintiffs probability of success on the merits is slight. Many corporate executives are unwilling to bet their company that they are in the right in big-stakes litigation, and a grant of class status can propel the stakes of a case into the stratosphere. In re Rhone-Poulenc Rorer Inc., 51 F.3d 1293 (7th Cir.1995), observes not only that class actions can have this effect on risk-averse corporate executives (and corporate counsel) but also that some plaintiffs or even some district judges may be tempted to use the class device to.wring settlements from defendants whose legal positions are justified but unpopular. Empirical studies of securities class actions imply that this is common. Janet Cooper Alexander, Do the Merits Matter? A Study of Settlements in Securities Class Actions, 43 Stan.L.Rev. 497 (1991); Reinier Kraakman, Hyun Park & Steven Sha-vell, When are Shareholder Suits in Shareholder Interests?, 82 Geo. L.J. 1733 (1994); Roberta Romano, The Shareholder Suit: Litigation Without Foundation?, 7 J.L. Econ. & Org. 55 (1991).. Class certifications also have induced judges to remake some substantive doctrine in order to render the litigation manageable. See Hal S. Scott, The Impact of Class Actions on Rule 10b-5, 38 U.Chi.L.Rev. 337 (1971). This interaction of procedure with the merits justifies an earlier appellate look. By the end of the case it will be too late— if indeed the case has an ending that is subject to appellate review.

*835So, in a mirror image of the death-knell situation, when the stakes are large and the risk of a settlement or other disposition that does not reflect the merits of the claim is substantial, an appeal under Rule 23(f) is in order. Again the appellant must demonstrate that the district court’s ruling on class certification is questionable — and must do this taking into account the discretion the district judge possesses in implementing Rule 23, and the correspondingly deferential standard of appellate review. However dramatic the effect of the grant or denial of class status in undercutting the plaintiffs claim or inducing the defendant to capitulate, if the ruling is impervious to revision there’s no point to an interlocutory appeal.

Third, an appeal may facilitate the development of the law. Because a large proportion of class actions settles or is resolved in a way that overtakes procedural matters, some fundamental issues about class actions are poorly developed. Recent proposals to amend Rule 23 were designed in part to clear up some of these questions. Instead, the Advisory Committee and the Standing Committee elected to wait, anticipating that appeals under Rule 23(f) would resolve some questions and illuminate others. When an appellant can establish that such an issue is presented, Rule 23(f) permits the court of appeals to intervene. When the justification for interlocutory review is contributing to development of the law, it is less important to show that the district judge’s decision is shaky. Law may develop through affir-mances as well as through reversals. Some questions have not received appellate treatment because ■ they are trivial; these are poor candidates for the use of Rule 23(f). But the more fundamental the question and the greater the likelihood that it will escape effective disposition at the end of the case, the more appropriate is an appeal under Rule 23(f). More than this it is impossible to say.

Judges have been stingy in accepting interlocutory appeals by certification under 28 U.S.C. § 1292(b), because that procedure interrupts the progress of a case and prolongs its disposition. , -That bogey is a principal reason why interlocutory appeals are so disfavored in the federal system. Disputes about class certification cannot be divorced from the merits — indeed, one of the fundamental unanswered questions is whether judges should be influenced by their tentative view of the merits when déciding whether to certify a class — and so this argument against interlocutory appeals carries some weight under Rule 23(f). But it has less weight than under § 1292(b), because Rule 23(f) is drafted to avoid d'elay. Filing a request for permission to appeal does not stop the litigation unless the district court or the court of appeals issues a stay — and a stay would depend on a demonstration that the probability of error in the class certification decision is high enough that the costs of pressing ahead in the district court exceed the costs of waiting. (This is the same kind of question that a court asks when deciding whether to issue a preliminary injunction or a stay of an administrative decision. See Illinois Bell Telephone Co. v. WorldCom Technologies, Inc., 157 F.3d 500 (7th Cir.1998); American Hospital Supply Corp. v. Hospital Products Ltd., 780 F.2d 589, 593-94 (7th Cir.1986).) We did not stay either of the two cases in which permission to appeal was sought; both continued in the district court and, as we related above, one already has been decided on the merits. Because stays will be infrequent, interlocutory appeals under Rule 23(f) should not unduly retard the pace of litigation.

So much for abstractions; what of this case? Equifax Check Services, which supplies a check-verification service to merchants, also attempts to collect fees imposed on dishonored checks. After we held that checks create “debts” within the meaning of the Fair Debt Collection Practices Act, 15 U.S.C. §§ 1692-1692o, see Bass v. Stolper, Koritzinsky, Brewster & Neider, S.C., 111 F.3d 1322 (7th Cir.1997), *836many of Equifax’s practices came under challenge. Until recently Equifax used a letter implying that it would refuse to verify checks written by anyone who had not paid all outstanding checks. Beverly Blair and Letressa Wilbon filed suits contending that this letter violated § 1692g of the Act because it did not adequately inform the recipients that they had 30 days within which to demand that Equifax obtain a verification of the debt from the merchant. Blair sought to represent a class of shoppers at Champs, and Wilbon a class of persons who had shopped at T.J. Maxx. The suits were consolidated and, after it became apparent that Equifax had sent the same letter to every person situated similarly to the plaintiffs,.the district judge certified the case as a class action under Fed.R.Civ.P. 23(b)(3), defining the class as: “all Illinois residents (i) who were sent a demand letter by [Equifax] on or after a date one year prior to the filing of this action, (ii) in the form represented by Exhibit A ..., (iii) in connection with an attempt to collect a check written to Champs or TJ Maxx for personal, family, or household purposes, where (iv) the letter was not returned by the Postal Service.” 1999 U.S. Dist. Lexis 2536 *21-22 (N.D.Ill.1999). The court also certified a subclass of persons who received a particular follow-up letter less than 30 days after Equifax sent the first. Because plaintiffs sought only statutory penalties, the difficulties of proving individual loss did not block class treatment. Cf. Keele v. Wexler, 149 F.3d 589 (7th Cir.1998). Equifax all but concedes that class certification was proper if the case is viewed in isolation, but it insists that what happened in another case requires a different outcome.

Several class actions against Equifax are pending in the Northern District of Illinois. On the same day Judge Plunkett certified the class in Blair, the plaintiffs in Crawford v. Equifax Check Services, Inc., No. 97 C 4240, which is pending before Magistrate Judge Schenkier, reached a settlement with Equifax. The class certified in Crawford is a superset of the class certified in Blair, and Equifax contends that as a result the terms of the Crawford settlement control here. A peculiar settlement it is. Equifax agreed to change the letters it sends in the future. Crawford personally receives $2,000. Members of the Crawford class get no relief for the letters sent to them, though Equifax agreed to donate $5,500 to Northwestern Law School’s Legal Aid Clinic and (natch) the lawyers for the class receive fees for their work. According to the settlement, none' of the class members will receive individual notice, and none will be offered the opportunity to opt out. The theory behind this is that the class was certified under Fed.R.Civ.P. 23(b)(2), even though it began as an action seeking damages. Finally, the settlement provides that all class members’ claims for compensatory or statutory damages pass through the litigation unaffected and may be asserted elsewhere — but only in individual suits. The settlement forbids prosecution of any other case as a class action. It is this final feature of the Crawford settlement that Equifax contends should have led Judge Plunkett to decertify the Blair-Wilbon class. Maintaining Blair as a class action creates at least a possibility of inconsistent outcomes.

Judge Plunkett was not amused. He was piqued at Equifax’s failure to ask the district court to consolidate Crawford with Blair, if indeed one comprises the other. He also concluded that the settlement in Crawford could not affect another pending suit. Because he deemed the Crawford settlement irrelevant,. Judge Plunkett denied Equifax’s motion for reconsideration of the class certification. This is the order Equifax wants to appeal under Rule 23(f). Before turning to that appeal, however, we need to describe additional proceedings before Magistrate Judge Schenkier.

Attorneys representing Blair and Wil-bon were invited to a settlement conference in Crawford and there learned — for the first time, they say — that the Craioford *837class includes the Blair class. Counsel opposed the Crawford settlement as inadequate but did not persuade either Magistrate Judge Sehenkier or Crawford’s lawyers. Blair and Wilbon then sought to intervene in Crawford so that they would be able to appeal from final approval of the settlement, if their objections at the hearing under Rule 23(e) should be rejected. Magistrate Judge Sehenkier denied the motion to intervene, concluding that counsel should have found out about the overlap and acted earlier. That decision is the subject of a separate appeal.

According to Blair and Wilbon, Equi-fax’s request for leave to appeal from Judge Plunkett’s decision is untimely. Rule 23(f) permits an application to be made “within ten days after entry of the order.” Judge Plunkett certified the class on February 25, 1999, but Equifax did not file its Rule 23(f) application until March 22. Plaintiffs insist that an order denying reconsideration is not the kind of “order” to which Rule 23(f) refers. Only the order “granting or denying class action certification under this rule” is subject to appeal, and on this view Equifax waited too long.-

Fed.R.App.P. 4(a)(4) provides that a timely motion for reconsideration suspends the finality of a judgment and thus extends the time for appeal until after the district court has acted on the motion, but this does not assist Equifax because it deals only with final decisions. For example, Rule 4(a)(4)(A)(iv) says that a motion “to. alter or amend the judgment under Rule 59” (emphasis added) tolls the time for appeal. An order certifying or, declining to certify a class is not a “judgment,” and the other subsections of Rule 4(a)(4) likewise refer to the kind of motions that follow entry of a final decision. Perhaps Rule 4(a)(4) could be read (rewritten?)‘so that “judgment” comes to mean “any order from which an appeal lies,” but this linguistic exercise is unnecessary. Rule 4(a)(4) just restates an accepted rule of practice: federal courts long have held that a motion for reconsideration tolls the time for appeal, provided that the motion is made within the time for appeal. United States v. Dieter, 429 U.S. 6, 97 S.Ct. 18, 50 L.Ed.2d 8 (1976); United States v. Healy, 376 U.S. 75, 84 S.Ct. 553, 11 L.Ed.2d 527 (1964); In re X-Cel, Inc., 823 F.2d 192 (7th Cir.1987). ‘ The practice is independent of Rule 4(a)(4), or any other rule.

Healy, for example, holds that a motion by a criminal prosecutor asking the district court to reconsider an order dismissing the indictment suspends the time for appeal, even though Fed.R.App.P. 4(b)(3), the parallel to Rule 4(a)(4) for criminal cases, gives this effect only to motions by the “defendant.” In re X-Cel similarly holds that post-decision motions in bankruptcy cases defer the time for appeal from the bankruptcy judge to the district judge. Dieter concluded that “the wisdom of giving district courts the opportunity promptly to correct their own alleged errors” (429 U.S. at 8, 97 S.Ct. 18) is all the justification needed for this practice. District judges should have no less opportunity to reconsider their orders before appeals under Rule 23(f). Thus we hold that a motion for reconsideration filed within ten days of “an order of a district court granting or denying class action certification” defers the time for appeal until after the district judge has disposed of the motion. Moreover, because Rule 23(f) is part of the civil’rules, the ten-day period does not include weekends or holidays. Fed.R.Civ.P. 6(a). Equifax therefore had until March 11 to seek reconsideration (it filed the motion on March 8), and because the district court reaffirmed its order on March 11 Equifax had until March 25 to seek permission to appeal (it applied ■ on Mhrch 22). Equifax took each step in time, so the appeal is within our:jurisdic-tiori — if we choose to accept it.

We do accept it. This situation fits our third category of appropriate interlocutory appeals. Equifax contends that it is entitled to be rid of multiple overlapping class actions. Questions concerning the *838relation among multiple suits may evade review at the end- of the ease, for by then the issue will be ‘the relation among (potentially inconsistent) judgments, and not the management of pending litigation. That neither side can point to any precedent in support of its position implies that this is one -of the issues that has evaded appellate resolution, and the issue is important enough to justify review now.

Because both sides favored us with their view of the merits of the appeal, as well as the question whether we should entertain it, we can bring matters to a swift conclusion. That the issue has evaded appellate resolution does not imply that it is difficult. Far from it. Judge Plunk-ett is plainly right—though not altogether for the reason he gave. We do not see any reason in principle why the disposition of the Crawford, litigation cannot b.e conclusive'on the plaintiffs in Blair. All members of the class certified in Blair also are members of the class certified in Crawford; a judgment binding on members of the Crawford class therefore will bind all members of the Blair class. See Tice v. American Airlines, Inc., 162 F.3d 966 (7th Cir.1998). If the judgment binds them not to pursue class actions, then the class in Blair must be decertified. But it does not yet have this effect, and the district judge was not required to jump the gun just to avoid all possibility of inconsistent outcomes.

Parallel cases often seek the same relief. There’s nothing peculiar about class actions. Sometimes the same plaintiff will file in two courts; sometimes different plaintiffs will seek equivalent relief in the same court. Our situation has a little of each, since Blair, Wilbon, and Crawford are not the same person, but they are in the same class. No mechanical rule governs the handling of overlapping cases. Judges sometimes stay proceedings in the more recently filed case to allow the first to proceed; sometimes a stay permits the more comprehensive of the actions to go forward. Cf. Colorado River Water Conservation District v. United States, 424 U.S. 800, 96 S.Ct. 1236, 47 L.Ed.2d 483 (1976). But the judge hearing the second-filed case may conclude that it is a superior vehicle and may press forward. When the cases proceed in parallel, the‘first to reach judgment controls the other, through claim preclusion (res judicata). Davis v. Chicago, 63 F.3d 801 (7th Cir.1995); Rogers v. Desiderio, 58 F.3d 299 (7th Cir.1995). Crawford has yet to produce a final and binding decision, however, so Judge Plunkett was entitled to proceed with Blair in the interim.

On occasion it will be so clear that the first-filed suit is the superior vehicle that it would be an abuse of discretion for the court in the second-filed suit to press forward. This is not such a case, however. Crawford is far from decision on the merits; it has seen negotiation, not combat. It is not clear that Crawford’s settlement will beat Blair to finality even if Blair is fully litigated. As we have recounted, Blair and Wilbon have tried to intervene in Crawford, and they have appealed , from the order denying that motion. We anticipate that they will appeal again from any order giving final approval to the Crawford settlement after the Rule 23(e) hearing. The latter appeal will of course be contingent on success in the intervention appeal, because only parties may appeal from an order settling a class action. See Felzen v. Andreas, 134 F.3d 873 (7th Cir.1998), affirmed by an equally divided Court under the name California Public Employees’ Retirement System v. Felzen, — U.S. -, 119 S.Ct. 720, 142 L.Ed.2d 766 (1999). But if Blair and Wilbon persuade us that Magistrate Judge Schenkier erred in excluding them from Crawford, or if some other class member intervenes and appeals from approval of the settlement, then this court will have to address the propriety of that disposition.

Approval cannot be called a foregone conclusion. Crawford was settled for a pittance, plus attorneys’ fees. Some cases *839settle for tiny suras because they have little chance of success; maybe Crawford, is such a case. (We have resisted all temptation to peek at its merits.) But if the class in Crawford has such a weak position, why were the debtors’ rights to compensatory and statutory damages preserved? If damages are at issue, how can Rule 23(b)(2) be used to avoid opt-outs and notice? If damages claims survive, what’s wrong with pursuing them in a separate class action? We have never heard of a class action being settled on terms that amount to: “For $7,500 plus attorneys’ fees, the class is disestablished.” When the individual claims are small, class actions are most useful. Perhaps Equifax found a plaintiff (or lawyer) willing to sell out the class — a possibility that we discussed most recently in Greisz v. Household Bank, 176 F.3d 1012 (7th Cir.1999)— and then tried to use Crawford as a way to thwart parallel actions where the class had more vigorous champions. Then again, perhaps the deal in Crawford was the best the class could obtain. We do not prejudge that issue. Enough questions have been raised, however, to show that Judge Plunkett was entitled to keep the Blair class in place until final decision in Crawford.

When overlapping suits are filed in separate courts, stays (or, rarely, transfers) are the best means of coordination. But both Craivford and Blair were ■ filed in the Northern District of Illinois. By far the best means of avoiding wasteful overlap when related suits are pending in the same court is to consolidate all before a single judge. Rules of the Northern District permit just such a process. At oral argument we asked the parties why this had not been done. Plaintiffs’ counsel replied that until shortly before they attended the settlement conference in Crawford they believed that the classes did not overlap. Counsel say that they were shocked to learn that Crawford is much the larger case and that the Blair class is its subset. Lawyers representing Equifax say that Blair’s lawyers knew this all along or should have deduced it, and Magistrate Judge Schenkier agreed. We can’t tell who is right, but surely Equifax knew from the get-go the relative sizes of, and relations among, the different class actions pending against it. Equifax could not plausibly explain at oral argument why it had not asked the district court to transfer all related actions to a single judge for decision. It is still not too late for the district court to accomplish this — although Magistrate Judge Schenkier will drop out of the picture if either case is transferred. Unanimous consent of the parties is required for a magistrate judge to enter final decision in a civil case, see 28 U.S.C. § 636(c), and it is obvious that. Blair and Wilbon won’t consent to that procedure. But both Craivford and Blair easily could be handled by the same district judge— whether Judge Plunkett, to whom Blair is assigned, or Judge Andersen, to whom Crawford was initially assigned, does not matter for this purpose.

No matter what the district court does, we will do our own part to consolidate and expedite decision. Crawford is a related case for purposes of our Operating Procedure 6(b), so that any appeal in Crawford, and any further appeal in Blair, will come to this panel. For today, it is enough to hold that, until Crawford reaches final judgment, Judge Plunkett does not abuse his discretion by handling Blair as a class action.

Affirmed.