Wood, Chief Judge.

The backdrop for this appeal is the world of automotive dealerships. But we are not concerned with the cars themselves. Instead, the controversy before us has to do with the management of the data those dealerships need in order to run an efficient business. In an effort to keep track of such vital business matters as accounting, payroll, inventory, sales, parts, service, finance, and insurance, the dealerships use computerized dealer-management systems. The dealers in our case did not write their own software or create their own hardware, however. Instead, some licensed a dealer-management system from CDK Global LLC, and others licensed a system from Reynolds and Reynolds Company.

Some dealer-management systems use open architecture, under which third parties have some access to dealer-originated data that has been plugged into the system. Others use closed architecture, under which that type of data scraping is forbidden under the license. The present litigation arose when CDK decided to change from an open system to a closed system, and CDK and its competitor Reynolds entered into agreements designed to ease the transition. Authenticom, a company that had been in the business of collecting data from the dealer-management systems and selling or using it for various applications ("apps"), sued under section 1 of the Sherman Act, 15 U.S.C. § 1, claiming that those agreements violated the Act. Because Authenticom's loss of access to the data was imperiling its survival as a company, it asked for and received a preliminary injunction from the district court in support of its suit.

CDK and Reynolds took an interlocutory appeal from the grant of the preliminary injunction, as is their right. See 28 U.S.C. § 1292(a)(1). We heard oral argument, and we now conclude that the injunction must be set aside. It goes well beyond the scope of the alleged violation, and in so doing fails to adhere to the lessons of Verizon Communications Inc. v. Law Offices of Curtis v. Trinko, 540 U.S. 398, 124 S.Ct. 872, 157 L.Ed.2d 823 (2004), and Pacific Bell Telephone Co. v. Linkline Communications, Inc., 555 U.S. 438, 129 S.Ct. 1109, 172 L.Ed.2d 836 (2009). In light of Authenticom's representations about its need for a quick resolution of this matter, however, we urge the district court to do what it can to expedite its final judgment.

I

Companies such as CDK and Reynolds that furnish dealer-management systems to automotive dealerships are in the business of providing a service to their customers. So are the companies that collect data for apps, or that design apps using data either uploaded by the dealer to the systems or generated by the systems. A glimpse at Authenticom's website indicates that it is in the data collection business. See http://www.authenticom.com/why.html (last visited Nov. 2, 2017). (It is unclear whether it also creates apps, but that detail does not matter for present purposes.) Neither the integration of data nor the development of apps competes with the systems. Instead, the collection and integration of data creates an intermediate downstream product; that data set can then be organized and used as an input for apps, either developed by outside vendors or by Authenticom itself. Functionally, a data integrator such as Authenticom is no different from an intermediary in any industry, whether sheet steel, uncut fabrics, or anything else. The fact that some of the dealer-management system providers are, or have been, vertically integrated such that they also sell integrated data or produce apps does not change the fundamental competitive analysis. That was one of the main points in the Supreme Court's Linkline decision, which dealt with independent internet service providers that competed with AT & T in the retail market, while at the same time they leased transport service lines from AT & T.

Authenticom has been around since 2002, when it was founded by Steve Cottrell. It describes itself as a third-party data integrator, meaning that it collects data from dealerships, organizes the data for various purposes, and sells this service to third-party app vendors, who use Authenticom's integration services to ensure that their apps are compatible with a dealer's management system. Critically, however, Authenticom does not obtain its data directly from raw dealership records. Instead, it "scrapes" (or collects) data from the management system that the dealer uses. CDK and Reynolds are the nation's largest system providers, although their relative market shares have changed over time. Until 2015, the system furnished by CDK placed no restrictions on data harvesting by third-party integrators; Reynolds, in contrast, has always forbidden that practice in its system licenses. (The record indicates that the restriction in the Reynolds licenses did not stop Authenticom from persuading some Reynolds users to permit it to scrape data from them as well, in violation of their agreements with Reynolds.) Apparently dealers liked the open approach. At one point Reynolds had about 40% of the market and CDK had less, but by the time the complaint in this case was filed, their relative positions had flipped, with CDK holding over 40% of the market and Reynolds around 30%. The remaining quarter of the market is occupied by one significant fringe firm, DealerTrack (now owned by Cox Automotive, with roughly 17% of the market), and numerous other smaller players. Despite the apparent preference of the dealers for the open model, CDK decided in 2015 to switch to a closed system.

Both CDK and Reynolds provide some data-integration services in-house (i.e. they are vertically integrated to some degree). The CDK product is called 3PA, and the Reynolds product is RCI. During its period as an open system provider, CDK also had two subsidiaries, Digital Motorworks and IntegraLink, which operated in the same way as Authenticom. An interested dealer would provide its log-in credentials to the integrator, which would then be able to pull data from the dealer's management system and provide it to an app vendor. The app market is highly competitive, populated not only by Reynolds and CDK, but also by such well-known firms as Carfax, AutoLoop, and Kelly Blue Book.

Until CDK switched to a closed system, Authenticom seems to have been satisfied with its place in the market. Reynolds had always used the closed model, meaning that it blocked (or tried to block) third-party access to data generated by its system. CDK, on the other hand, did not prevent third-party integrators, including Authenticom, from accessing and scraping data from its system, with dealer permission. The reason CDK gave for changing its model from open to closed in 2014 was the need to respond to a series of "well-publicized security breaches" that worried its cybersecurity team. It also appears to have been motivated by a desire to squeeze more value out of its internal integration program, 3PA. Either way, after CDK made the switch to a closed system, DealerTrack was left as the only major open dealer-management system provider.

CDK's change to a closed system left its data-integration subsidiaries, Digital Motorworks and IntegraLink, in an awkward position, because they had been following the same business model as Authenticom and had, it seems, scraped data from dealers with the Reynolds system despite the prohibitions in the dealers' license agreements. In order to wean CDK's subsidiaries from their dependence on the ability to scrape data, CDK and Reynolds entered into a series of written agreements. The first was the "Data Exchange Agreement," in which CDK agreed to a five-year wind-down of the two subsidiaries. During the wind-down period, CDK agreed that the subsidiaries would not scrape data from Reynolds users without Reynolds's permission. In turn, Reynolds agreed that it would not block Digital Motorworks and IntegraLink from obtaining access to its system during the wind-down. CDK also promised that it would help its subsidiaries' clients (that is, the third-party app providers who used the subsidiaries to collect data from the Reynolds system) to move away from using Digital Motorworks and IntegraLink's services, over to Reynolds's in-house integrator, RCI. Finally, the Data Exchange Agreement committed both Reynolds and CDK not to help anyone else gain access to the other company's data management system without permission.

The other two agreements between CDK and Reynolds were called the 3PA and RCI Agreements. In them, each company promised to allow the other company's proprietary integration software to have access to its data management system. Thus, for instance, CDK's 3PA software was entitled to take data from the Reynolds system. The parties agreed that neither would use any unauthorized or unsecured methods of obtaining access to the other's data management system. Notably, nothing in any of the three agreements required either CDK or Reynolds to block third-party access to its own data management system.

In the wake of CDK's switch to a closed system and its three agreements with Reynolds, Authenticom's business took a dive. Without the ability to scrape data from the big data management systems, it was faced with the unattractive choice of going straight to the dealers and creating a new data base, or giving up the chance to serve the great majority of dealerships. The district court found that the measures to block third-party integrators Reynolds had adopted in 2013 caused Authenticom to lose roughly a quarter of its business. The primary source of this injury appears to have been CDK's shift from an open system to a closed system, even though some harm may also have been attributable to the 2015 agreements. Some customers left Authenticom for Digital Motorworks and IntegraLink, because they had access to both CDK's system (as subsidiaries) and Reynolds's system (because of the Data Exchange Agreement).

II

That is where matters stood when, in May 2017, Authenticom sued Reynolds and CDK under the antitrust laws. It asserted that the three 2015 agreements are per se unlawful under section 1 of the Sherman Act, because they eliminate competition in the data-integration market. It characterizes the 2015 agreements as agreements not to compete in the data-integration market, and further alleges that these agreements were designed to drive independent providers, such as Authenticom, from the market. It also attacks CDK and Reynolds's alleged exclusive dealing provisions in their dealer licenses, under which dealers are forbidden to provide their log-in credentials to anyone else.

There is evidence that prices have increased considerably since the 2015 agreements took effect. Alan Andreu, a manager who works for the third-party application vendor Dominion Dealer Solutions, testified that in 2011 his company purchased data integration from Authenticom and another independent company for $30 to $35 per month, per dealer. At that time, Reynolds charged $247 for similar services, a figure that has now risen to a whopping $893 in the aftermath of its agreements with CDK. Another vendor, AutoLoop, had a similar tale to tell: before the 2015 agreement, CDK was charging it $160 per month, and now it charges $735 per month for the identical service. App vendors pass along these bloated fees to dealers, who are already paying large fees for their data management system. Authenticom also represents that it is hanging on by a thread, and that without relief it is likely to be forced to shutter its business altogether. Such a move would be harmful overall to the data-integration market, it claims.

III

The merits of this lawsuit have yet to be tried, and so nothing we say should be taken as presaging the eventual outcome of the case. The only matter before us is the propriety of the preliminary injunction that the district court issued — a step it took out of concern for preserving Authenticom as a functioning company until the suit can be resolved one way or the other. We begin with a brief description of the injunction, which the court explained in an order issued contemporaneously with the preliminary injunctions against each defendant. See Authenticom, Inc. v. CDK Global, LLC, et al., No. 17-cv-318-jdp, 2017 WL 3206943 (W.D. Wis. July 28, 2017).

The injunctions (one per defendant) do not focus on the 2015 agreements that are the focal point of Authenticom's lawsuit. Instead, they address the measures that the court believed were necessary to "extend a lifeline to Authenticom, to maintain its viability until this case is finally decided on the merits." Id. at *1. The key provisions (simplified somewhat) are as follows:

There is more, but this is enough to give the flavor of the injunctions. The defendants argue both that no preliminary injunction should have issued, and that in any event there is no warrant for these injunctions.

Perhaps the only common ground among the parties is the legal standard that governs preliminary injunctions:

Winter v. Natural Resources Defense Council, Inc., 555 U.S. 7, 20, 129 S.Ct. 365, 172 L.Ed.2d 249 (2008). See also Whitaker v. Kenosha Unified Schl. Dist. No. 1 Bd. of Educ., 858 F.3d 1034, 1044 (7th Cir. 2017) (citation omitted), citing Turnell v. CentiMark Corp., 796 F.3d 656, 661-62 (7th Cir. 2015).

Harm cannot be considered irreparable if it can be fully rectified in a final judgment. Whitaker, 858 F.3d at 1045. The district court thought that business failure by Authenticom would be irreparable, in this sense. But that is not clear: we do not know, at this stage, whether insolvency would drive it to reorganization or dissolution; we do not know if a treble damage award (to which a winning antitrust plaintiff is entitled, see 15 U.S.C. § 15(a)) might be enough to resuscitate the company; nor do we know whether Authenticom might be able to survive with a combination of the dealers who are not licensees of CDK or Reynolds and those who prefer open systems and thus abandon the defendants. (If licenses last five years, it is likely that 20% of them come up for renewal each year. It thus would not take too long for dealers who resent the closed system, as some apparently do, to free themselves of CDK or Reynolds and take their data management system business to DealerTrack or one of the fringe firms.)

We need not explore this question further. For present purposes, we may assume (generously to Authenticom) that it can show the first three requirements for a preliminary injunction: irreparable harm, inadequate remedy at law, and at least some likelihood of success on the merits. That takes us to the balancing issue: how does Authenticom's harm from the denial of a preliminary injunction compare to CDK's and Reynolds's harm from the grant of an injunction?

We already have discussed Authenticom's theory of harm. But it does not stand in a vacuum; to the contrary, any injunction, and especially this one, has costs for CDK and Reynolds. Reynolds argued to the district court, and argues here, that it has always had a closed system, and it has been willing to pay a price in customer loyalty because it believes that the closed system delivers a higher quality product to the dealers. That is because it believes that it has better control of system security when it knows what data is going in, who can manipulate that data, and what is coming back out. Extracting data from the system involves at least some access to the data base (how much is disputed, as is the extent to which the data scraped by data integrators is limited to dealer information versus additional content added by Reynolds). Reynolds believes too that it would be perverse to punish it under the antitrust laws for doing exactly the same thing it always has done. It sees the three 2015 agreements as benign, arguing that CDK's promise not to help a third party assist in the breach of Reynolds's license with its dealers is neutral from a competitive standpoint. Even monopolists are almost never required to assist their competitors, see Trinko, 540 U.S. at 408, 124 S.Ct. 872, and neither Reynolds nor CDK is a monopolist. See also Olympia Equip. Leasing Co. v. Western Union Telegraph Co., 797 F.2d 370 (7th Cir. 1986). Finally, to the extent that Authenticom asserts that Reynolds and CDK were engaged in a group boycott to force it from the market, Reynolds argues that the facts do not fit the theory. Reynolds was indifferent to Authenticom's market presence, save for its interest in preventing Authenticom from seeking free, unauthorized access to its work product.

CDK makes many of the same points. Unlike Reynolds, it does not have a history of exclusive use of closed systems. It switched from open to closed in 2015. It asserts that it did so with data security in mind. Essentially, however, its position is that it had no duty to continue dealing with data integrators such as Authenticom, that nothing in the 2015 agreements forbade CDK itself from allowing access to its own data integration system, and thus that the agreements cannot be seen as anticompetitive.

This is not a complete summary of Reynolds and CDK's arguments, but it is enough to show that they have raised serious points. As we hinted earlier, however, the resolution of this appeal does not depend on which side of the balance we favor. (Indeed, given the fact that the district court enjoys considerable discretion in preliminary injunction proceedings, simple disagreement with that court would not be enough to justify a different result.) There is a more fundamental problem with the injunction that requires us to set it aside. Authenticom's case centers on agreements — actual and alleged — between Reynolds and CDK. The actual agreements are the ones signed in 2015; the alleged agreement is one described by Authenticom's CEO, Cottrell, when he said that executives at Reynolds and CDK confessed that they were determined to block Authenticom from their systems.

The proper remedy for a section 1 violation based on an agreement to restrain trade is to set the offending agreement aside. From the standpoint of preliminary injunctive relief, that would mean ordering Reynolds and CDK not to implement their 2015 agreements or any alleged agreement collectively to bar Authenticom from their data management systems. Had the district court issued such an injunction, we would have a different case. But it did not. Instead, it ordered the defendants to enter into an entirely new arrangement with Authenticom (granted, during the pendency of the litigation), essentially forcing them to do business with Authenticom on terms to which they did not agree. Such an order is inconsistent with Trinko, which cautioned that an order to continue to do business with a firm is proper only if the case fits "within the limited exception recognized in Aspen Skiing [Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 105 S.Ct. 2847, 86 L.Ed.2d 467 (1985)]." 540 U.S. at 409, 124 S.Ct. 872. Aspen Skiing was a section 2 case, which already distinguishes it from the present one. In a section 1 case, all that is needed is to break up the allegedly troublesome agreement and return the market, for the time being, to the status quo ante. See In re Microsoft Corp. Antitrust Litig., 333 F.3d 517, 533 (4th Cir. 2003) (reversing a mandatory injunction similar to the one in this case on the ground that "relief under the antitrust laws must flow from that conduct which is proscribed by the antitrust laws") (internal quotation marks omitted; emphasis in original).

Authenticom also contends that Reynolds and CDK have agreed to tie their data management systems to data integration. We are dubious in the extreme that this amounts to tying, rather than simply participation at two levels of the market, as in Linkline. But perhaps there is more to the claim than meets the eye on this record. Even if that were so, the proper remedy would be to enjoin the tie, not to create a duty to deal.

Authenticom has not argued that either Reynolds or CDK has sufficient market power on its own to trigger either a monopolization or an attempt-to-monopolize claim under section 2 of the Sherman Act, 15 U.S.C. § 2. And as we have noted, even if it did, this case is a far cry from Aspen Skiing, which represented the high-water mark in section 2 cases for a duty-to-deal theory. Trinko and Linkline leave no doubt that such a theory cannot support relief.

IV

We appreciate the district court's concern to ensure that a potentially sound antitrust case should not disappear before its eyes because the plaintiff runs a high risk of going out of business while the litigation drags on. That does not, however, justify a preliminary injunction that goes so far beyond a measure that restores what the market would look like in the absence of the alleged violation. After trial, the court will be better able to assess the competitive significance of CDK and Reynolds's business models, of the 2015 agreements, and of any other agreement Authenticom is able to prove. In the meantime, however, the court must vacate the preliminary injunction that it entered on July 28, 2017.

The preliminary injunction is VACATED and the case is REMANDED to the district court for further proceedings consistent with this opinion.