Opinion filed by Circuit Judge BROWN.
Opinion concurring in the judgment filed by Circuit Judge TATEL.
Dissenting opinion filed by Circuit Judge KAVANAUGH.
BROWN, Circuit Judge.The FTC sought a preliminary injunction, under 15 U.S.C. § 53(b), to block the merger of Whole Foods and Wild Oats. It appeals the district court’s denial of the injunction. I conclude the district court should be reversed, though I do so reluctantly, admiring the thoughtful opinion the district court produced under trying circumstances in which the defendants were rushing to a financing deadline and the FTC presented, at best, poorly explained evidence. Nevertheless, the district court committed legal error in assuming market definition must depend on marginal consumers; consequently, it underestimated the FTC’s likelihood of success on the merits.
I
Whole Foods Market, Inc. (“Whole Foods”) and Wild Oats Markets, Inc. (“Wild Oats”) operate 194 and 110 grocery stores, respectively, primarily in the United States. In February 2007, they announced that Whole Foods would acquire Wild Oats in a transaction closing before August 31, 2007. They notified the FTC, as the Hart-ScottARodino Act required for the $565 million merger, and the FTC investigated the merger through a series of hearings and document requests. On June 6, 2007, the FTC sought a temporary restraining order and preliminary injunction to block the merger temporarily while the FTC conducted an administrative proceeding to decide whether to block it permanently under § 7 of the Clayton Act. The parties conducted expedited discovery, and the district court held a hearing on July 31 and August 1, 2007.
The FTC contended Whole Foods and Wild Oats are the two largest operators of what it called premium, natural, and organic supermarkets (“PNOS”). Such stores “focus on high-quality perishables, specialty and natural organic produce, prepared foods, meat, fish[,] and bakery goods; generally have high levels of customer services; generally target affluent and well educated customers [and] ... are mission driven with an emphasis on social and environmental responsibility.” FTC v. Whole Foods Market, Inc., 502 F.Supp.2d 1, 28 (D.D.C.2007). In eighteen cities, asserted the FTC, the merger would create monopolies because Whole Foods and Wild Oats are the only PNOS. To support this claim, the FTC relied on emails Whole Foods’s CEO John Mackey sent to other Whole Foods executives and directors, suggesting the purpose of the merger was to eliminate a competitor. In addition the FTC produced pseudonymous blog postings in which Mr. Mackey touted Whole *1033Foods and denigrated other supermarkets as unable to compete. The FTC’s expert economist, Dr. Kevin Murphy, analyzed sales data from the companies to show how entry by various supermarkets into a local market affected sales at a Whole Foods or Wild Oats store.
On the other hand, the defendants’ expert, Dr. David Scheffman, focused on whether a hypothetical monopolist owning both Whole Foods and Wild Oats would actually have power over a distinct market. He used various third-party market studies to predict that such an owner could not raise prices without driving customers to other supermarkets. In addition, deposition testimony from other supermarkets indicated they regarded Whole Foods and Wild Oats as critical competition. Internal documents from the two defendants reflected their extensive monitoring of other supermarkets’ prices as well as each other’s.
The district court concluded that PNOS was not a distinct market and that Whole Foods and Wild Oats compete within the broader market of grocery stores and supermarkets. Believing such a basic failure doomed any chance of the FTC’s success, the court denied the preliminary injunction without considering the balance of the equities.
On August 17, the FTC filed an emergency motion for an injunction pending appeal, which this court denied on August 23. FTC v. Whole Foods Market, Inc., No. 07-5276 (D.C.Cir. Aug. 23, 2007). Freed to proceed, Whole Foods and Wild Oats consummated their merger on August 28. The dissent argues that a reversal today contradicts this earlier decision, but our standard of review then was very different, requiring the FTC to show “such a substantial indication of probable success” that there would be “justification for the court’s intrusion into the ordinary processes of ... judicial review.” Wash. Metro. Area Transit Comm’n v. Holiday Tours, Inc., 559 F.2d 841, 843 (D.C.Cir.1977). It is hardly remarkable that the FTC could fail to meet such a stringent standard and yet persuasively show the district court erred in applying the much less demanding § 53(b) preliminary injunction standard.
II
At the threshold, Whole Foods questions our jurisdiction to hear this appeal. The merger is a fait accompli, and Whole Foods has already closed some Wild Oats stores and sold others. In addition, Whole Foods has sold two complete lines of stores, Sun Harvest and Harvey’s, as well as some unspecified distribution facilities. Therefore, argues Whole Foods, the transaction is irreversible and the FTC’s request for an injunction blocking it is moot.
Only in a rare case would we agree a transaction is truly irreversible, for the courts are “clothed with large discretion” to create remedies “effective to redress [antitrust] violations and to restore competition.” Ford Motor Co. v. United States, 405 U.S. 562, 573, 92 S.Ct. 1142, 31 L.Ed.2d 492 (1972). Indeed, “divestiture is a common form of relief’ from unlawful mergers. United States v. Microsoft Corp., 253 F.3d 34, 105 (D.C.Cir.2001) (en banc). Further, an antitrust violator “may ... be required to do more than return the market to the status quo ante.” Ford Motor, 405 U.S. at 573 n. 8 92 S.Ct. 1142. Courts may not only order divestiture but may also order relief “designed to give the divested [firm] an opportunity to establish its competitive position.” Id. at 575, 92 S.Ct. 1142. Even remedies which “entail harsh consequences” would be appropriate to ameliorate the harm to competition from an antitrust violation. United States v. E.I. du Pont de Nemours & Co., 366 *1034U.S. 316, 327, 81 S.Ct. 1243, 6 L.Ed.2d 318 (1961).
Of course, neither court nor agency has found Whole Foods’s acquisition of Wild Oats to be unlawful. Therefore, the FTC may not yet claim the right to have any remedy necessary to undo the effects of the merger, as it could after such a determination, du Pont, 366 U.S. at 334, 81 S.Ct. 1243. But the whole point of a preliminary injunction is to avoid the need for intrusive relief later, since even with the considerable flexibility of equitable relief, the difficulty of “unscrambl[ing] merged assets” often precludes “an effective order of divestiture,” FTC v. Dean Foods Co., 384 U.S. 597, 607 n. 5, 86 S.Ct. 1738, 16 L.Ed.2d 802 (1966). Section 53(b), codifying the ability of the FTC to obtain preliminary relief, FTC v. Weyerhaeuser Co., 665 F.2d 1072, 1082 (D.C.Cir.1981), preserves the “flexibility” of traditional “equity practice,” id. at 1084. At a minimum, the courts retain the power to preserve the status quo nunc, for example by means of a hold separate order, id., and perhaps also to restore the status quo ante.
Thus, the courts have the power to grant relief on the FTC’s complaint, despite the merger’s having taken place, and this case is therefore not moot. See Byrd v. EPA 174 F.3d 239, 244 (D.C.Cir.1999) (“The availability of a partial remedy is sufficient to prevent [a] case from being moot.”). The fact that Whole Foods has sold some of Wild Oats’s assets does not change our conclusion. To be sure, we have no “authority to command return to the status quo,” Weyerhaeuser, 665 F.2d at 1077, in a literal way by forcing absent parties to sell those assets back to Whole Foods, but there is no reason to think that inability prevents us from mitigating the merger’s alleged harm to competition. The stores Whole Foods has sold are only those under the Harvey’s and Sun Harvest labels, which were never relevant to the anticompetitive harm the FTC fears. Our inability to command their return does not limit the relief available to the FTC. As to the distribution facilities, neither party has described what they are, suggested Wild Oats would not be a viable competitor without them, or explained why the district court could not order some provisional substitute. Moreover, the FTC is concerned about eighteen different local markets. If, as appears to be the situation, it remains possible to reopen or preserve a Wild Oats store in just one of those markets, such a result would at least give the FTC a chance to prevent a § 7 violation in that market.
Ill
“We review a district court order denying preliminary injunctive relief for abuse of discretion.” FTC v. H.J. Heinz Co., 246 F.3d 708, 713 (D.C.Cir.2001). However, if the district court’s decision “rests on an erroneous premise as to the pertinent law,” we will review the denial de novo “in light of the legal principles we believe proper and sound.” Id.
Despite some ambiguity, the district court applied the correct legal standard to the FTC’s request for a preliminary injunction. The FTC sought relief under 15 U.S.C. § 53(b), which allows a district court to grant preliminary relief “[u]pon a proper showing that, weighing the equities and considering the Commission’s likelihood of ultimate success, such action would be in the public interest.” The relief is temporary and must dissolve if more than twenty days pass without an FTC complaint. Id. Congress recognized the traditional four-part equity standard for obtaining an injunction was “not appropriate for the implementation of a Federal statute by an independent regulatory agency.” Heinz, 246 F.3d at 714. Therefore, to obtain a § 53(b) preliminary injunction, the FTC need not show any ir*1035reparable harm, and the “private equities” alone cannot override the FTC’s showing of likelihood of success. Weyerhaeuser, 665 F.2d at 1082-83.
In deciding the FTC’s request for a preliminary injunction blocking a merger under § 53(b), a district court must balance the likelihood of the FTC’s success against the equities, under a sliding scale. See Heinz, 246 F.3d at 727; FTC v. Elders Grain, Inc., 868 F.2d 901, 903 (7th Cir.1989). The equities will often weigh in favor of the FTC, since “the public interest in effective enforcement of the antitrust laws” was Congress’s specific “public equity consideration” in enacting the provision. Heinz, 246 F.3d at 726. Therefore, the FTC will usually be able to obtain a preliminary injunction blocking a merger by “raising] questions going to the merits so serious, substantial, difficult[,] and doubtful as to make them fair ground for thorough investigation.” Heinz, 246 F.3d at 714-15. By meeting this standard, the FTC “creates a presumption in favor of preliminary injunctive relief,” id. at 726; but the merging parties may rebut that presumption, requiring the FTC to demonstrate a greater likelihood of success, by showing equities weighing in favor of the merger, Weyerhaeuser, 665 F.2d at 1087. Conversely, a greater likelihood of the FTC’s success will militate for a preliminary injunction unless particularly strong equities favor the merging parties. See Heinz, 246 F.3d at 727; Elders Grain, 868 F.2d at 903.
In any case, a district court must not require the FTC to prove the merits, because, in a § 53(b) preliminary injunction proceeding, a court “is not authorized to determine whether the antitrust laws ... are about to be violated.” FTC v. Food Town Stores, Inc., 539 F.2d 1339, 1342 (4th Cir.1976). That responsibility lies with the FTC. Id. Not that the court may simply rubber-stamp an injunction whenever the FTC provides some threshold evidence; it must “exercise independent judgment” about the questions § 53(b) commits to it. Weyerhaeuser, 665 F.2d at 1082. Thus, the district court must evaluate the FTC’s chance of success on the basis of all the evidence before it, from the defendants as well as from the FTC. See FTC v. Beatrice Foods Co., 587 F.2d 1225, 1229-30 (D.C.Cir.1978) (App’x to Stmt, of MacKinnon & Robb, JJ.) (“[W]e are also required to consider the inroads that the appellees’ extensive showing has made ... [S]everal basic contentions of the FTC are called into serious question.”). The district court should bear in mind the FTC will be entitled to a presumption against the merger on the merits, see Elders Grain, 868 F.2d at 906, and therefore does not need detailed evidence of anti-competitive effect at this preliminary phase. Nevertheless, the merging parties are entitled to oppose a § 53(b) preliminary injunction with their own evidence, and that evidence may force the FTC to respond with a more substantial showing.
The district court did not apply the sliding scale, instead declining to consider the equities. To be consistent with the § 53(b) standard, this decision must have rested on a conviction the FTC entirely failed to show a likelihood of success. Indeed, the court concluded “the relevant product market in this case is not premium natural and organic supermarkets ... as argued by the FTC but ... at least all supermarkets.” Whole Foods, 502 F.Supp.2d at 34. It also observed that several supermarkets “have already repositioned themselves to compete vigorously with Whole Foods and Wild Oats for the consumers’ premium natural and organic food business.” Id. at 48. Thus, considering the defendants’ evidence as well as the FTC’s, as it was obligated to do, the court was in no doubt that this merger would not substantially lessen competition, because it found the evidence proved Whole Foods *1036and Wild Oats compete among supermarkets generally. If, and only if, the district court’s certainty was justified, it was appropriate for the court not to balance the likelihood of the FTC’s success against the equities.
rv
However, the court’s conclusion was in error. The FTC contends the district court abused its discretion in two ways: first, by treating market definition as a threshold issue; and second, by ignoring the FTC’s main evidence. The district court acted reasonably in focusing on the market definition, but it analyzed the product market incorrectly.
A
First, the FTC complains the district court improperly focused on whether Whole Foods and Wild Oats operate within a PNOS market. However, this was not an abuse of discretion given that the district court was simply following the FTC’s outline of the case.
Inexplicably, the FTC now asserts a market definition is not necessary in a § 7 case, Appellant’s Br. 37-38, in contravention of the statute itself, see 15 U.S.C. § 18 (barring an acquisition “where in any line of commerce ... the effect of such acquisition may be substantially to lessen competition”); see also Brown Shoe Co. v. United States, 370 U.S. 294, 324, 82 S.Ct. 1502, 8 L.Ed.2d 510 (1962) (interpreting “any line of commerce” to require a “determination of the relevant market” to find “a violation of the Clayton Act”); Elders Grain, 868 F.2d at 906 (“[A]ll this assumes a properly defined market.”). The FTC suggests “market definition ... is a means to an end — to enable some measurement of market power — not an end in itself.” Appellant’s Br. 38 n. 26. But measuring market power is not the only purpose of a market definition; only “examination of the particular market — its structure, history[,] and probable future — can provide the appropriate setting for judging the probable anticompetitive effect of the merger.” Brown Shoe, 370 U.S. at 322 n. 38, 82 S.Ct. 1502.
That is not to say market definition will always be crucial to the FTC’s likelihood of success on the merits. Nor does the FTC necessarily need to settle on a market definition at this preliminary stage. Although the framework we have developed for a -prima facie § 7 case rests on defining a market and showing undue concentration in that market, United States v. Baker Hughes Inc., 908 F.2d 981, 982-83 (D.C.Cir.1990), this analytical structure does not exhaust the possible ways to prove a § 7 violation on the merits, see, e.g., United States v. El Paso Natural Gas Co., 376 U.S. 651, 660, 84 S.Ct. 1044, 12 L.Ed.2d 12 (1964), much less the ways to demonstrate a likelihood of success on the merits in a preliminary proceeding. Section 53(b) preliminary injunctions are meant to be readily available to preserve the status quo while the FTC develops its ultimate case, and it is quite conceivable that the FTC might need to seek such relief before it has settled on the scope of the product or geographic markets implicated by a merger. For example, the FTC may have alternate theories of the merger’s anticompetitive harm, depending on inconsistent market definitions. While on the merits, the FTC would have to proceed with only one of those theories, at this preliminary phase it just has to raise substantial doubts about a transaction. One may have such doubts without knowing exactly what arguments will eventually prevail.1 Therefore, a district court’s assessment of the FTC’s chances will not *1037depend, in every ease, on a threshold matter of market definition.
In this case, however, the FTC itself made market definition key. It claimed “[t]he operation of premium natural and organic supermarkets is a distinct ‘line of commerce’ within the meaning of Section 7,” and its theory of anticompetitive effect was that the merger would “substantially increase concentration in the operation of [PNOS].” Compl. ¶¶ 34, 43. Throughout its briefs, the FTC presented a straightforward § 7 case in which “whether the transaction creates an appreciable danger of anticompetitive effects ... depends upon ... [the] relevant product ... [and] geographic market ... and the transaction’s probable effect on competition in the product and geographic markets.” FTC’s Br. Mot. Prelim. Inj. 11-12. It purported to show “undue concentration in the relevant market,” as the mainstay of its case. Id. at 12. Because of the concentration in the supposed PNOS market, the FTC urged the district court to hold the merger “presumptively unlawful,” and this was its sole reason for blocking the merger. FTC’s Proposed Conclusions of Law ¶¶ 57-63, 99-108. At oral argument, the FTC’s counsel suggested it had other ideas about the anticompetitive effect of the merger even if its PNOS market definition is wrong; but the FTC never offered those ideas to the district court. It is incumbent on the parties to shape a case, and it was hardly an abuse of discretion for the district court to focus on the questions as the FTC presented them.
B
Thus, the FTC assumed the burden of raising some question of whether PNOS is a well-defined market. As the FTC presented its case, success turned on whether there exist core customers, committed to PNOS, for whom one should consider PNOS a relevant market. The district court assumed “the ‘marginal’ consumer, not the so-called ‘core’ or ‘committed’ consumer, must be the focus of any antitrust analysis.” Whole Foods, 502 F.Supp.2d at 17 (citing Horizontal Merger Guidelines, 57 Fed.Reg. 41,552 (1992)). To the contrary, core consumers can, in appropriate circumstances, be worthy of antitrust protection. See Horizontal Merger Guidelines § 1.12, 57 Fed.Reg. at 41,555 (explaining the possibility of price discrimination for “targeted buyers”). The district court’s error of law led it to ignore FTC evidence that strongly suggested Whole Foods and Wild Oats compete for core consumers within a PNOS market, even if they also compete on individual products for marginal consumers in the broader market. See, e.g., Appellant’s Br. 50, 53.
A market “must include all products reasonably interchangeable by consumers for the same purposes.” Microsoft, 253 F.3d at 52. Whether one product is reasonably interchangeable for another depends not only on the ease and speed with which customers can substitute it and the desirability of doing so, see id. at 53-54, but also on the cost of substitution, which depends most sensitively on the price of the products. A broad market may also contain relevant submarkets which themselves “constitute product markets for antitrust purposes.” Brown Shoe, 370 U.S. at 325, 82 S.Ct. 1502. “The boundaries of such a submarket may be determined by examining such practical *1038indicia as industry or public recognition of the submarket as a separate economic entity, the product’s peculiar characteristics and uses, unique production facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors.” Id.
To facilitate this analysis, the Department of Justice and the FTC developed a technique called the SSNIP (“small but significant non-transitory increase in price”) test, which both Dr. Murphy and Dr. Scheffman used. In the SSNIP method, one asks whether a hypothetical monopolist controlling all suppliers in the proposed market could profit from a small price increase. Horizontal Merger Guidelines § 1. 11, 57 Fed.Reg. at 41,560-61. If a small price increase would drive consumers to an alternative product, then that product must be reasonably substitutable for those in the proposed market and must therefore be part of the market, properly defined. Id.
Experts for the two sides disagreed about how to do the SSNIP of the proposed PNOS market. Dr. Scheffman used a method called critical loss analysis, in which he predicted the loss that would result when marginal customers shifted purchases to conventional supermarkets in response to a SSNIP.2 Whole Foods, 502 F.Supp.2d at 18. He concluded a hypothetical monopolist could not profit from a SSNIP, so that conventional supermarkets must be within the same market as PNOS. In contrast, Dr. Murphy disapproved of critical loss analysis generally, preferring a method called critical diversion that asked how many customers would be diverted to Whole Foods and how many to conventional supermarkets if a nearby Wild Oats closed. Whole Foods’s internal planning documents indicated at least a majority of these customers would switch to Whole Foods, thus making the closure profitable for a hypothetical PNOS monopolist. One crucial difference between these approaches was that Dr. Seheffman’s analysis depended only on the marginal loss of sales, while Dr. Murphy’s used the average loss of customers. Dr. Murphy explained that focusing on the average behavior of customers was appropriate because a core of committed customers would continue to shop at PNOS stores despite a SSNIP.
In appropriate circumstances, core customers can be a proper subject of antitrust concern. In particular, when one or a few firms differentiate themselves by offering a particular package of goods or services, it is quite possible for there to be a central group of customers for whom “only [that package] will do.” United States v. Grinnell Corp., 384 U.S. 563, 574, 86 S.Ct. 1698, 16 L.Ed.2d 778 (1966); see also United States v. Phillipsburg Nat’l Bank & Trust Co., 399 U.S. 350, 360, 90 S.Ct. 2035, 26 L.Ed.2d 658 (1970) (“[I]t is the cluster of products and services ... that as a matter of trade reality makes commercial banking a distinct” market.). What motivates antitrust concern for such customers is the possibility that “fringe competition” for individual products within a package may not protect customers who need the whole package from market power exercised by a sole supplier of the package. Grinnell, 384 U.S. at 574, 86 S.Ct. 1698.
Such customers may be captive to the sole supplier, which can then, by means of price discrimination, extract monopoly profits from them while competing for the business of marginal customers. Cf. Md. People’s Counsel v. FERC, 761 F.2d 780, 786-87 (D.C.Cir.1985) (allowing natural gas pipelines to charge higher prices to *1039captive customers would be anticompeti-tive). Not that prices that segregate core from marginal consumers are in themselves anticompetitive; such pricing simply indicates the existence of a submarket of core customers, operating in parallel with the broader market but featuring a different demand curve. See United States v. Rockford Mem’l Corp., 898 F.2d 1278, 1284 (7th Cir.1990). Sometimes, for some customers a package provides “access to certain products or services that would otherwise be unavailable to them.” Phillipsburg Nat’l Bank & Trust, 399 U.S. at 360, 90 S.Ct. 2035. Because the core customers require the whole package, they respond differently to price increases from marginal customers who may obtain portions of the package elsewhere. Of course, core customers may constitute a submarket even without such an extreme difference in demand elasticity. After all, market definition focuses on what products are reasonably substitutable; what is reasonable must ultimately be determined by “settled consumer preference.” United States v. Phila. Nat’l Bank, 374 U.S. 321, 357, 83 S.Ct. 1715, 10 L.Ed.2d 915 (1963).
In short, a core group of particularly dedicated, “distinct customers,” paying “distinct prices,” may constitute a recognizable submarket, Brown Shoe, 370 U.S. at 325, 82 S.Ct. 1502, whether they are dedicated because they need a complete “cluster of products,” Phila. Nat’l Bank, 374 U.S. at 356, 83 S.Ct. 1715, because their particular circumstances dictate that a product “is the only realistic choice,” SuperTurf, Inc. v. Monsanto Co., 660 F.2d 1275, 1278 (8th Cir.1981), or because they find a particular product “uniquely attractive,” Nat’l Collegiate Athletic Ass’n v. Bd. of Regents of the Univ. of Okla., 468 U.S. 85, 112, 104 S.Ct. 2948, 82 L.Ed.2d 70 (1984). For example, the existence of core customers dedicated to office supply superstores, with their “unique combination of size, selection, depth[,] and breadth of inventory,” was an important factor distinguishing that submarket. FTC v. Staples, Inc., 970 F.Supp. 1066, 1078-79 (D.D.C.1997). As always in defining a market, we must “take into account the realities of competition.” Weiss v. York Hosp., 745 F.2d 786, 826 (3d Cir. 1984). We look to the Brown Shoe indicia, among which the economic criteria are primary, see Rothery Storage & Van Co. v. Atlas Van Lines, Inc., 792 F.2d 210, 219 n. 4 (D.C.Cir.1986).
The FTC’s evidence delineated a PNOS submarket catering to a core group of customers who “have decided that natural and organic is important, lifestyle of health and ecological sustainability is important.” Whole Foods, 502 F.Supp.2d at 23 (citing Hr’g Tr. 43^4, Aug. 1, 2007). It was undisputed that Whole Foods and Wild Oats provide higher levels of customer service than conventional supermarkets, a “unique environment,” and a particular focus on the “core values” these customers espoused. Id. The FTC connected these intangible properties with concrete aspects of the PNOS model, such as a much larger selection of natural and organic products, FTC’s Proposed Findings of Fact 13-14 & ¶ 66 (noting Earth Fare, a PNOS, carries “more than 45,000 natural and organic SKUs”) and a much greater concentration of perishables than conventional supermarkets, id. 14-15 & ¶ 69-70 (“Over 60% of Wild Oats’ revenues” and “[n]early 70% of Whole Foods sales are natural or organic perishables.”). See also Whole Foods, 502 F.Supp.2d at 22-23 (citing defendants’ depositions as evidence of Whole Foods’s and Wild Oats’s focus on “high-quality perishables” and a large variety of products).
Further, the FTC documented exactly the kind of price discrimination that enables a firm to profit from core customers for whom it is the sole supplier. Dr. Murphy compared the margins of Whole Foods *1040stores in cities where they competed with Wild Oats. He found the presence of a Wild Oats depressed Whole Foods’s margins significantly. Notably, while there was no effect on Whole Foods’s margins in the product category of “groceries,” where Whole Foods and Wild Oats compete on the margins with conventional supermarkets, the effect on margins for perishables was substantial. Confirming this price discrimination, Whole Foods’s documents indicated that when it price-checked conventional supermarkets, the focus was overwhelmingly on “dry grocery,” rather than on the perishables that were 70% of Whole Foods’s business. Thus, in the high-quality perishables on which both Whole Foods and Wild Oats made most of their money, they competed directly with each other, and they competed with supermarkets only on the dry grocery items that were the fringes of their business.
Additionally, the FTC provided direct evidence that PNOS competition had a greater effect than conventional supermarkets on PNOS prices. Dr. Murphy showed the opening of a new Whole Foods in the vicinity of a Wild Oats caused Wild Oats’s prices to drop, while entry by non-PNOS stores had no such effect. Similarly, the opening of Earth Fare stores (another PNOS) near Whole Foods stores caused Whole Foods’s prices to drop immediately. The price effect continued, while decreasing, until the Earth Fare stores were forced to close.
Finally, evidence of consumer behavior supported the conclusion that PNOS serve a core consumer base. Whole Foods’s internal projections, based on market experience, suggested that if a Wild Oats near a Whole Foods were to close, the majority (in some cases nearly all) of its customers would switch to the Whole Foods rather than to conventional supermarkets. Since Whole Foods’s prices for perishables are higher than those of conventional supermarkets, such customers must not find shopping at the latter interchangeable with PNOS shopping. They are the core customers. Moreover, market research, including Dr. Scheffman’s own studies, indicated 68% of Whole Foods customers are core customers who share the Whole Foods “core values.” FTC Proposed Findings of Fact ¶ 135.
Against this conclusion the defendants posed evidence that customers “cross-shop” between PNOS and other stores and that Whole Foods and Wild Oats check the prices of conventional supermarkets. Whole Foods, 502 F.Supp.2d at 30-32. But the fact that PNOS and ordinary supermarkets “are direct competitors in some submarkets ... is not the end of the inquiry,” United States v. Conn. Nat’l Bank, 418 U.S. 656, 664 n. 3, 94 S.Ct. 2788, 41 L.Ed.2d 1016 (1974). Of course customers cross-shop; PNOS carry comprehensive inventories. The fact that a customer might buy a stick of gum at a supermarket or at a convenience store does not mean there is no definable groceries market. Here, cross-shopping is entirely consistent with the existence of a core group of PNOS customers. Indeed, Dr. Murphy explained that Whole Foods competes actively with conventional supermarkets for dry groceries sales, even though it ignores their prices for high-quality perishables.
In addition, the defendants relied on Dr. Scheffman’s conclusion that there is no “clearly definable” core customer. Whole Foods, 502 F.Supp.2d at 28. However, this conclusion was inconsistent with Dr. Scheffman’s own report and testimony. Market research had found that customers who shop at Whole Foods because they share the core values it champions constituted at least a majority of its customers. Scheffman Expert Report 56-57. Moreover, Dr. Scheffman acknowledged “there *1041are core shoppers [who] will only buy organic and natural” and for that reason go to Whole Foods or Wild Oats. Hr’g Tr. 31, July 31, 2007. He contended they could be ignored because the numbers are not “substantial.” Id. Again, Dr. Scheffman’s own market data undermined this assertion.
In sum, the district court believed the antitrust laws are addressed only to marginal consumers. This was an error of law, because in some situations core consumers, demanding exclusively a particular product or package of products, distinguish a submarket. The FTC described the core PNOS customers, explained how PNOS cater to these customers, and showed these customers provided the bulk of PNOS’s business. The FTC put forward economic evidence — which the district court ignored — showing directly how PNOS discriminate on price between their core and marginal customers, thus treating the former as a distinct market. Therefore, I cannot agree with the district court that the FTC would never be able to prove a PNOS submarket. This is not to say the FTC has in fact proved such a market, which is not necessary at this point. To obtain a preliminary injunction under § 53(b), the FTC need only show a likelihood of success sufficient, using the sliding scale, to balance any equities that might weigh against the injunction.
V
It remains to address the equities, which the district court did not reach, and see whether for some reason there is a balance against the FTC that would require a greater likelihood of success. The FTC urges us to carry out the rest of this determination, but “[w]e believe the proper course of action at this point is to remand to the district court, Chaplaincy of Full Gospel Churches v. England, 454 F.3d 290, 304 (D.C.Cir.2006).” Since the district court “expressly withheld consideration,” id. at 305, of the equities, we have not had the benefit of its findings. Although the equities in a § 53(b) preliminary injunction proceeding will usually favor the FTC, Heinz, 246 F.3d at 726, the district court must independently exercise its discretion considering the circumstances of this case, including the fact that the merger has taken place. The district court should remember that a “risk that the transaction will not occur at all,” by itself, is a private consideration that cannot alone defeat the preliminary injunction. See id.; Weyerhaeuser, 665 F.2d at 1082-83.
I appreciate that the district court expedited the proceeding as a courtesy to the defendants, who wanted to consummate their merger just thirty days after the hearing, Whole Foods, 502 F.Supp.2d at 4, but the court should have taken whatever time it needed to consider the FTC’s evidence fully. For the reasons stated above, the district court’s conclusion that the FTC showed no likelihood of success in an eventual § 7 case must, be reversed and remanded for proceedings consistent with this opinion.
So ordered.
. For example, a merger between two close competitors can sometimes raise antitrust concerns due to unilateral effects in highly differentiated markets. See generally Hori*1037zontal Merger Guidelines, 57 Fed.Reg. 41,-552, 41,560-61, § 2.2 (1992). In such a situation, it might not be necessary to understand the market definition to conclude a preliminary injunction should issue. The FTC alludes to this theory on appeal, but to the district court it argued simply that the merger would result in a highly concentrated PNOS market.
. Dr. Scheffman did not actually calculate the amount of this loss. He simply predicted that because many Whole Foods and Wild Oats customers also shop at conventional supermarkets, the loss would at any rate be too large.