WOOD, Chief Judge.

This case arises out of several class actions that were brought against Unilever United States, Inc. (Unilever USA) to recover damages from a hair-smoothing product that allegedly destroyed users' hair and burned their scalps. The lead case, Reid v. Unilever USA, was brought in the Northern District of Illinois under the court's diversity jurisdiction, see 28 U.S.C. § 1332, related actions in Kentucky and California were later transferred to Illinois and consolidated with Reid. The cases were eventually settled, but not to everyone's satisfaction. Tina Martin, a class member, objected to the settlement on numerous grounds, which we detail below. We have examined all of them and conclude that the district court acted well within its discretion when it approved the settlement. We therefore affirm its judgment.

I

The class representatives in the three suits had all purchased Unilever USA's Suave® Professionals Keratin Infusion 30 Day Smoothing Kit (the Smoothing Kit), a hair product that supposedly would smooth hair and coat it with Keratin, a protein found naturally in hair. Unfortunately, for some consumers, the Smoothing Kit was a disaster. Its active ingredient, thioglycolic acid, is extremely corrosive, and if left on long enough, can dissolve the hair and burn the scalp. Asserting claims for breach of warranty, violations of state consumer fraud and deceptive practices laws, and unjust enrichment, plaintiffs in several states filed class action lawsuits against Unilever USA and related companies. (We refer to them collectively as Unilever USA.)

Once the cases were consolidated in the Northern District of Illinois, they were stayed so that the parties could pursue mediation. They worked for a year and a half, with the help of retired District Court Judge Wayne Andersen, and ultimately succeeded in reaching a settlement agreement on February 7, 2014. That settlement was presented to the district court as required by Federal Rule of Civil Procedure 23(e). Chief Judge Rubén Castillo entered an order on February 12, 2014, granting preliminary approval of the settlement and directing notice to the settlement class. After a final approval hearing held on July 9, 2014, he entered an order granting final approval on July 29, 2014. Objector Martin has appealed from the final order.

The settlement class is defined as "[a]ll persons who purchased or used the Smoothing Kit in the United States, before February 17, 2014, excluding [those who did not purchase for personal use, those who signed a release for consideration, and certain interested parties]." The settlement provides that this class would be certified, and that the class would dismiss its claims against the defendants in exchange for specified compensation. In particular, Unilever USA agreed to create two settlement funds: a Reimbursement Fund of $250,000, and an Injury Fund of $10,000,000, for a total of $10,250,000. The Reimbursement Fund is available to any member of the settlement class who seeks compensation, but the compensation is limited to a one-time payment of $10 per person. That payment represents reimbursement for the cost of purchasing the Smoothing Kit. The Injury Fund is designed to compensate any member of the settlement class (excluding those who opted out) who suffered bodily injury as a result of using the Smoothing Kit. Applicants must proceed under one of three options: Benefit A, which is capped at $40 per claimant, is available for class members who incurred expenses for hair treatment but who no longer have supporting receipts; Benefit B is for claimants who do have receipts, such as hairdresser or medical bills. Each claimant is eligible to receive $800. Persons who suffered significant bodily injury are eligible for Benefit C, which provides for an award up to $25,000 per claimant. A Special Master appointed by the district court will make the Benefit C determinations, and will evaluate any Benefit A or B claim that the Settlement Administrator deems insufficient. Unilever USA will bear all costs of notice, claims administration, and attorneys' fees for class counsel, along with litigation costs and expenses. Class counsel's fee is entirely separate from the $10,250,000 available for class compensation. Finally, two named plaintiffs receive incentive awards of $7,500, and Reid got $10,000.

Chief Judge Castillo appointed retired Magistrate Judge Nan R. Nolan to serve as the Special Master. But before much could happen, Objector Martin (along with Yolanda Reed, who has since been dismissed from the case) filed this appeal from the order finally approving the settlement. Martin raises eleven points in her brief:

We address a number of these together, as the basic points overlap somewhat. As we noted earlier, our review is deferential, for abuse of discretion only. Isby v. Bayh, 75 F.3d 1191, 1196-97 (7th Cir.1996).

II

The general principles that guide the court's evaluation of a proposed class settlement agreement include (1) the strength of the class's case, (2) the complexity and expense of further litigation, (3) the amount of opposition, (4) the reaction of class members to the settlement, (5) the opinion of competent counsel, and (6) the stage of the proceedings and the amount of discovery that was completed. Wong v. Accretive Health, Inc., 773 F.3d 859, 863 (7th Cir.2014). Of these, we have suggested that the first consideration is the most important. Id. at 864. That said, the likelihood of the class's success after full litigation affects many of the points Martin has raised. More than that, it is generally for the parties to decide how much litigation risk they wish to take, and courts should be hesitant to second-guess them.

Martin's first four points deal in one way or another with the accuracy of the data on which the district court relied. Relying on an estimate that the Food and Drug Administration put on its website about Unilever USA's May 2012 recall of the product, she argues that the class size is probably larger than the negotiating parties assumed. She also notes that some Smoothing Kits are still being sold after the recall — a point that she believes adds to the indeterminacy of the class. If the class is significantly larger than the parties assumed, then the dollar amounts provided for the settlement are more likely to be inadequate.

Martin adds that the court also did not know enough about the number of people who suffered serious injuries, or the value of those injuries, to make an informed decision. As of the July hearing, a little more than 500 injury claims had been filed. It is true that very little was said about them, and that more information might have been helpful. But Martin herself does not seem to be sure about what point she is making here. On the one hand, she urges that the low number of claims proves that no one will benefit from this settlement and too much money will be returned to Unilever USA; on the other hand, she says that the number of claims will balloon and there will not be enough money to cover them all.

The district court addressed these points and concluded that it had enough data for an informed decision and that the dollar amounts were within a reasonable range. The judge noted that he did not "think the $10 million that has been set aside is going to be fully used and ... a chunk of that money, which I estimate as much as a third or maybe even 40 percent, might revert back to Unilever. But the attorneys' fees in this case are not coming from that fund, and even if 40 percent reverts back to Unilever, that still results in a significant recovery for the remaining class members, somewhere in the neighborhood of $6 million." Martin's speculation that there might be many more claims filed is also unsupported. The hearing occurred two-thirds of the way through the claim period. By that time, the administrator had received 2,294 claims (including 458 Benefit A claims, 67 Benefit B claims, and 131 Benefit C claims). These numbers do not suggest that there was a horde of people who were likely to swamp the monies available in the remaining six months. Nor is there any reason to think that a trial would have produced substantially different results for the class. Compare Reynolds v. Beneficial Nat'l Bank, 288 F.3d 277, 284-85 (7th Cir.2002) (estimate of trial recovery important where settlement is problematic). We see no abuse of discretion in the district court's decision to approve the amount of the benefits.

Martin's next arguments (points 5 and 6) focus on the way the settlement treated the personal injury and tort claims. She criticizes what she sees as the lumping together of quite different personal injury claims in one "adjudicatory scheme." It is unclear why she believes that she personally would be better off under a different arrangement, but we set that point aside and address what seems to be her main argument: that the settlement's value was too low because it failed to recognize that there are a number of different applicable laws.

A common question of law (or fact), see Rule 23(a)(2), is necessary for class certification. And the Supreme Court has held that settlement classes are subject to the same certification criteria as litigation classes. See Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 619-22, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997); see also Ortiz v. Fibreboard Corp., 527 U.S. 815, 846-47, 864-65, 119 S.Ct. 2295, 144 L.Ed.2d 715 (1999). Martin does not seem to be arguing that the district court erred in certifying this class; she appears to be saying only that a different structure for the class (perhaps subclasses) would have been more advantageous to the members. An exchange at the fairness hearing sheds some light on her point. Local counsel for Martin said that "personal injury cases are inherently difficult to manage on a class-wide basis because of the differences in injuries amongst the plaintiffs, differences in state laws, et cetera." Chief Judge Castillo then asked, "So are you saying that there should not be class actions for product liability cases?" Counsel responded, "No, I'm not taking it that far." Nor has Martin taken it "that far" in this court.

It is true that nationwide settlements of claims subject to state law are not always possible. In In re Bridgestone/Firestone, Inc., 288 F.3d 1012, 1020 (7th Cir.2002), we cautioned that "[d]ifferences across states may be costly for courts and litigants alike, but they are a fundamental part of our federal republic and must not be overridden in a quest to clear the queue in court." But this did not mean that nationwide classes are impermissible as a matter of law. In Pella Corp. v. Saltzman, 606 F.3d 391 (7th Cir. 2010), we found certification to be appropriate for just such a class. We distinguished the issue in Pella — a design defect in windows leading to wood rot — from the myriad different tire defects and recalls in Bridgestone. For what it is worth, the present case appears to resemble Pella more than Bridgestone, but there is a stronger reason for supporting the district court's decision here: the settlement agreement contained a choice-of-law clause, which specified the law of Illinois. Martin seems to have overlooked this point, when she asks in her brief, "will variations in claimants' respective state laws be considered in determining the award?" The short answer is that those variations will not make a difference, because of the choice of a single law.

Martin's seventh and eighth points relate to the absence of any injunctive relief in the settlement agreement. She would like to see a court order preventing Unilever USA from making and marketing any products containing hazardous chemicals, as well as an order requiring a more effective recall procedure that ensures that all Smoothing Kits are removed from store shelves. But the agreement carves out retailers that are still selling the product. The need for such an injunction in the settlement agreement was something the district court considered and rejected. We find no abuse of discretion in its decision.

The ninth contention Martin presses is that the settlement agreement, via the class definition, should not have carved out the people who had already signed releases. She argues that "Unilever has ... engaged in a campaign designed to obtain unconscionable and unenforceable releases from consumers injured by the Product." Under the settlement agreement, Unilever is entitled to provide the Special Master with evidence of releases in the Benefit C cases (or any Benefit A or B cases that were rejected by the Administrator) that are referred to her. The district court saw nothing wrong with this, nor do we. It will be up to the Special Master to consider the effect of any particular release. Moreover, Unilever has a point when it notes that Martin lacks standing to make this argument, because she did not sign a release.

Argument 10 is a general complaint about the documentation requirements for the Benefit C claimants. Martin argues that these requirements are so onerous that monies intended for class members will wind up going back to Unilever USA or to class counsel instead. The latter point is simply wrong, because the settlement ensures that the funds available for compensation are entirely separate from the funds that will go to counsel. As for the general claim that the procedures are too burdensome, we have no reason to disagree with the district court's assessment to the contrary. Anyone can get $10 just by signing an affidavit to the effect that she or he purchased the kit and providing any "available" documentation they might have. Claimants can also be reimbursed up to $40 for haircuts if those were necessary. The fact that better documentation is needed for those with significant injuries is hardly a surprise. In the absence of any evidence that the Special Master has been imposing unrealistic requirements on claimants, we see nothing wrong with this aspect of the settlement.

Finally, Martin raises a number of complaints about class counsel's fee and the procedures the court followed in approving it. The district court deferred consideration of the fee motion until after it had given its final approval to the settlement agreement. Martin objected that this prevented her from commenting on the fee petition, in violation of both Rule 23 and due process. Unilever responds that any claim she might have was not ripe at the time of the final approval, because the fee award had not yet occurred then. (The proceedings related to fees were completed on June 10, 2015.) More persuasively, it also notes that the attorneys' fee petition was submitted two weeks before the deadline for objections — that is, in plenty of time for input — and Martin herself filed an objection. Nothing in Rule 23 prohibits the deferral of the final fee award until after the agreement is approved, especially for an agreement structured as this one is, where the fees are kept entirely separate from the funds that will be available for compensation. This type of provision is to be encouraged, not criticized. It was certainly within the district court's discretion to approve it.

III

Martin has made some additional arguments against the settlement agreement, but we see no need to address them separately. We conclude that the district court properly approved the settlement agreement reached in this case, and we thus AFFIRM its judgment.