The main issue presented by this appeal is whether the district court abused its discretion when it dismissed a complaint for forum non conveniens because it failed to consider all relevant public factors for each forum after determining that the private factors for the litigants were not in equipoise. Fresh Results, an American company, arranged bulk shipments of blueberries for ASF Holland, a Dutch company that repacks wholesale produce to sell to European customers. ASF Holland created reports about the results of its inspection of the shipments, and those reports determined the final price it paid for the blueberries. Fresh Results filed a complaint against ASF Holland in the Southern District of Florida, alleging that it had falsified the reports and fraudulently deflated the price. ASF Holland moved to dismiss the complaint on the ground that the Netherlands was a more convenient forum for the suit, and the district court agreed. After concluding that the interests of the litigants-the so-called "private factors"-were not in equipoise, the district court ruled that it need not consider all relevant "public factors" for each forum and dismissed the complaint so that the litigation could proceed in the *1047Netherlands. The district court derived the equipoise standard from dicta in our precedent, La Seguridad v. Transytur Line , 707 F.2d 1304, 1307 (11th Cir. 1983), that we have since recited in dicta in other cases. Fresh Results contends that the district court abused its discretion when it weighed the private factors in favor of dismissal and when it failed to consider the relevant public factors. Because we agree that the district court abused its discretion when it failed to consider the relevant public factors and committed two errors in its analysis of the private factors, we vacate and remand.
I. BACKGROUND
This appeal concerns a blueberry deal that soured. Fresh Results, an American company, acts as a sales agent for growers of produce in South America. In 2015, Fresh Results arranged bulk shipments of blueberries for two seasons for ASF Holland, a Dutch company that buys wholesale produce to repack and sell to customers in Europe.
To initiate a shipment, ASF Holland would request blueberries from Fresh Results at a reference price that purportedly reflected its anticipated net returns. When Fresh Results received the request, it would coordinate with the growers to fulfill the order. The growers would send the blueberries from South America directly to the Netherlands by air freight. When the shipment arrived, ASF Holland was responsible for inspecting, sorting, and repacking the blueberries in a timely manner. It would then send Fresh Results several reports on the shipment, with details of its inspection, sorting, sales prices, and expenses. Fresh Results used the reports to adjust the reference price and create an invoice for ASF Holland with the final price. ASF Holland would remit the invoice amount to Fresh Results in Florida.
During the second season, one of the growers hired an auditor to make an unannounced inspection of a blueberry shipment at ASF Holland's facility in the Netherlands. The auditor allegedly discovered that the blueberries were still in their original freight package, even though ASF Holland had reported to Fresh Results that the shipment had been inspected, sorted, and repacked. After learning of the auditor's inspection, Fresh Results demanded that ASF Holland pay the market price for each shipment of blueberries it had received, but ASF Holland refused.
Fresh Results filed a complaint, which it later amended, against ASF Holland in the Southern District of Florida. Fresh Results asserted claims of breach of contract, negligent misrepresentation, fraud, conversion, and tortious interference with its business relationship with the growers. It alleged that ASF Holland fraudulently promised a high reference price but then deflated the actual price it paid by sending false reports. According to Fresh Results, ASF Holland manipulated the price by understating the amount paid by its European customers and by falsely inflating its expenses in the reports. ASF Holland informed the district court that it would pursue counterclaims against Fresh Results for sending substandard blueberries.
ASF Holland then moved to dismiss the complaint for failure to state a claim and forum non conveniens . It argued that the Netherlands was a more convenient forum for the dispute. ASF Holland presented an affidavit from Ronald Jongbloed, its managing director, in which he asserted that the important documents and witnesses are in the Netherlands. And it presented an affidavit from Sebastiaan Moolenaar, a Dutch lawyer, in which he asserted that Fresh Results can obtain relief for all its *1048claims in the Netherlands and that the United States has no treaty with the Netherlands for the reciprocal enforcement of judgments.
The district court granted the motion on the ground that forum non conveniens warranted dismissal. Although it explained that a strong presumption favors Fresh Results' choice of forum, the court ruled that the private factors weighed in favor of dismissal. It reasoned that most sources of proof needed to prove Fresh Results' claims are in the Netherlands, where the blueberries were delivered, repacked, and sold and where the reports were allegedly falsified. Although Fresh Results contended that the South American growers were willing to participate in litigation only in the United States and not in the Netherlands, the district court disregarded the growers' testimony as a source of proof because they were not parties. And it weighed in favor of dismissal the possibility for view of ASF Holland's facility in the Netherlands. Because the United States has no treaty with the Netherlands for the reciprocal enforcement of judgments, the court concluded that Fresh Results would inevitably have to litigate in the Netherlands to enforce a judgment if it obtained one, so it weighed the enforceability of a judgment in favor of dismissal.
The district court decided that it need not consider the public factors if the private factors were not "in equipoise or near equipoise," relying on King v. Cessna Aircraft Company , 562 F.3d 1374, 1382 (11th Cir. 2009) (quoting La Seguridad , 707 F.2d at 1307 ). It then concluded that all "the private interest factors align in favor of Defendant's position and are thus not [in] equipoise, [so] it need not engage in an exhaustive analysis of all public interest[ ] factors." The court also concluded that "[n]onetheless, the public interest factors favor dismissal as well," but it discussed only one public factor, choice of law, and weighed that factor in favor of dismissal because Dutch law would likely apply. The court dismissed the complaint for forum non conveniens and declined to decide the other grounds for dismissal asserted by ASF Holland.
II. STANDARD OF REVIEW
Because "[t]he forum non conveniens determination is committed to the sound discretion of the trial court," we review for abuse of discretion. Piper Aircraft Co. v. Reyno , 454 U.S. 235, 257, 102 S.Ct. 252, 70 L.Ed.2d 419 (1981). "[W]here the court has considered all relevant public and private interest factors, and where its balancing of these factors is reasonable, its decision deserves substantial deference." Id.
III. DISCUSSION
Under the doctrine of forum non conveniens , a district court may decline to exercise its jurisdiction when a foreign forum is better suited to adjudicate the dispute. See Kolawole v. Sellers , 863 F.3d 1361, 1369 (11th Cir. 2017). The "central purpose" of forum non conveniens is "to ensure that the trial is convenient." Id. (citation and internal quotation marks omitted). The doctrine should not be invoked "lightly ... because it effectively deprives the plaintiff of his favored forum," id. , and so a defendant bears the burden of justifying dismissal based on forum non conveniens , La Seguridad , 707 F.2d at 1309. To satisfy this burden, the defendant must establish that "(1) an adequate alternative forum is available, (2) the public and private factors weigh in favor of dismissal, and (3) the plaintiff can reinstate his suit in the alternative forum without undue inconvenience or prejudice." Tazoe v. Airbus S.A.S. , 631 F.3d 1321, 1330 (11th Cir. 2011) (quoting *1049Leon v. Millon Air, Inc. , 251 F.3d 1305, 1310-11 (11th Cir. 2001) ).
The second part of the forum non conveniens analysis-the balancing of the private and public factors-is a "comparative inquiry [that] requires the district court to weigh the 'relative' advantages and disadvantages of each respective forum." Id. at 1331. The private factors "pertain to the interests of the participants in the litigation." Id. One of these factors is "the relative ease of access to sources of proof," which includes the "availability of compulsory process for attendance of unwilling, and the cost of obtaining attendance of willing, witnesses." Gulf Oil Corp. v. Gilbert , 330 U.S. 501, 508, 67 S.Ct. 839, 91 L.Ed. 1055 (1947) ; see also Piper Aircraft , 454 U.S. at 241 n.6, 102 S.Ct. 252. Other factors are the "possibility of view of premises, if view would be appropriate to the action" and the enforceability of a judgment, if one is obtained, Gulf Oil , 330 U.S. at 508, 67 S.Ct. 839. And a court may consider "all other practical problems that make trial of a case easy, expeditious and inexpensive." Id. The public factors "pertain to the relative interests of the two fora." Tazoe , 631 F.3d at 1333. Among other things, the public factors consider "the administrative difficulties flowing from court congestion," "the 'local interest in having localized controversies decided at home,' " and "the unfairness of burdening citizens in an unrelated forum with jury duty." Piper Aircraft , 454 U.S. at 241 n.6, 102 S.Ct. 252 (quoting Gulf Oil , 330 U.S. at 509, 67 S.Ct. 839 ). A court may also consider what law will govern the action, including "the avoidance of unnecessary problems in conflicts of laws" and "the application of foreign law." Id.
Fresh Results challenges the balancing of the private and public factors by the district court. Fresh Results argues that the district court abused its discretion both when it weighed the private factors in favor of dismissal and when it failed to consider all the relevant public factors. We agree.
We divide our discussion in two parts. First, we explain that the district court abused its discretion when it failed to consider the public factors after determining that the private factors were not in equipoise. Second, we explain that the district court committed two errors in its analysis of the private factors.
A. The District Court Abused Its Discretion when It Failed to Consider the Relevant Public Factors After Determining that the Private Factors Were Not in Equipoise.
Fresh Results argues that the district court abused its discretion when it failed to consider all relevant public factors after concluding that the private factors were not in equipoise. The equipoise standard employed by the district court comes from dicta in our caselaw. Although our holdings are precedential, our dicta are not. See United States v. Caraballo-Martinez , 866 F.3d 1233, 1244 (11th Cir. 2017). Dicta refer to "those portions of an opinion that are not necessary to deciding the case then before us." Id. (citation and quotation marks omitted). In contrast, our holdings "constitute the precedent, as a point necessarily decided" in that case. Bryan A. Garner et al., The Law of Judicial Precedent § 4, at 44 (2016) (emphasis omitted); accord Powell v. Thomas , 643 F.3d 1300, 1304-05 (11th Cir. 2011) (explaining that "a holding is comprised both of the result of the case and those portions of the opinion necessary to that result by which we are bound" (citation and internal quotation marks omitted)). And we have explained that, "regardless of what a court says in its opinion, the decision can hold nothing beyond the facts of that case."
*1050Edwards v. Prime, Inc. , 602 F.3d 1276, 1298 (11th Cir. 2010).
Our first mention of the equipoise standard occurred in dicta in La Seguridad v. Transytur Line , 707 F.2d 1304, 1307 (11th Cir. 1983). In that decision, we quoted at length from a general description of the forum non conveniens analysis from a decision of the District of Columbia Circuit, Pain v. United Technologies Corporation , 637 F.2d 775, 784-85 (D.C. Cir. 1980). One sentence in our block quotation from Pain stated that "[i]f the trial judge finds [the] balance of private interests to be in equipoise or near equipoise, he must then determine whether or not factors of public interest tip the balance in favor of a trial in a foreign forum." La Seguridad , 707 F.2d at 1307 (quoting Pain , 637 F.2d at 784-85 ). We never drew particular attention to that sentence, and the equipoise standard was of no particular relevance in the context of our discussion. But, in the next paragraph, we explained that "controlling weight cannot be given to any one factor in the balancing process or the doctrine would lose much of the flexibility that is its essence." Id. And we made no further mention of the equipoise standard anywhere else in our decision.
Nor was the equipoise standard a point necessarily decided. Because the parties in La Seguridad kept shifting their theories of liability and defenses, we held that "the case [was not in] a posture in which the forum non conveniens motion could be resolved." Id. at 1309. So "there [was] simply no basis for a forum non conveniens determination other than sheer speculation." Id. Because we vacated and remanded on that basis, we had no occasion to adopt Pain 's equipoise standard as a holding.
Indeed, our instructions to the district court on remand made clear that we could not have adopted the equipoise standard as a holding. Despite our suggestion that a court need not always consider the public factors, we instructed the district court to consider on remand "the interest, if any, of the forum in having a United States court applying controlling United States statutes to shipping contracts that generate much local revenue and employment"-a public factor. Id. at 1310. And we instructed the district court that it "must weigh the advantages of the United States forum in the balance, such as its familiarity with the law that will govern the suit "-another public factor. Id. (emphasis added); see also Gulf Oil , 330 U.S. at 509, 67 S.Ct. 839. Because we "still [did] not know the underlying nature of" the claims and defenses and were "left only to speculate as to what witnesses and documents might be relevant and where they might be located," we could not have known whether the private factors were or were not in equipoise when we issued these instructions about the public factors. La Seguridad , 707 F.2d at 1308. The equipoise standard was "not essential to the reasoning behind that decision." Garner et al., supra § 4, at 44.
After La Seguridad , we clarified that the public factors do not "enter the equation only when the private interest factors are at or near equipoise." Leon , 251 F.3d at 1311 (internal quotation marks omitted). Although "the private factors are generally considered more important than the public factors," we explained that the public factors are not superfluous, even when the private factors are far from equipoise. Id. And we opined that "the better rule is to consider both factors in all cases," which "has been our approach in recent cases." Id. ; see also SME Racks, Inc. v. Sistemas Mecanicos Para Electronica, S.A. , 382 F.3d 1097, 1100 n.5 (11th Cir. 2004) ("We have clarified that ... courts should consider both public and private factors in all *1051cases." (citation and internal quotation marks omitted)).
The equipoise standard is hard to square with Supreme Court precedent. As one commentator has observed, "there is no sound basis in either Gulf Oil or the Supreme Court's later decision in Piper Aircraft Company v. Reyno for such an approach." 14D Charles Alan Wright et al., Federal Practice and Procedure § 3828.4 (4th ed. 2009). In Gulf Oil , the Supreme Court set forth the balancing test and provided a nonexhaustive list of both private and public factors, albeit without mention of how the factors fit together. 330 U.S. at 508-09, 67 S.Ct. 839. Then, in Piper Aircraft , the Court explained that we should not disturb a district court's forum non conveniens decision when it "considered all relevant public and private interest factors , and where its balancing of these factors [was] reasonable." 454 U.S. at 257, 102 S.Ct. 252 (emphasis added). The Court also explained that a district court may dismiss for forum non conveniens "when trial in the chosen forum would 'establish oppressiveness and vexation to a defendant out of all proportion to plaintiff's convenience' or when the 'chosen forum is inappropriate because of considerations affecting the court's own administrative and legal problems.' " Id. at 241, 102 S.Ct. 252 (alterations adopted) (emphasis added) (quoting Koster v. Lumbermens Mut. Cas. Co. , 330 U.S. 518, 524, 67 S.Ct. 828, 91 L.Ed. 1067 (1947) ). That is, a court may dismiss because of either the private factors, which pertain to the interests of the litigants, or the public factors, which pertain to the interests of the fora. See id. So the Court has at least suggested that the public factors may warrant dismissal on their own accord. See id.
We now expressly disavow the equipoise standard. In the light of Piper Aircraft , the District of Columbia Circuit, which we quoted in our first mention of the standard, has since abandoned it. See Nemariam v. Fed. Democratic Republic of Ethiopia , 315 F.3d 390, 393 (D.C. Cir. 2003) (explaining that Piper Aircraft "overrul[ed] the third part of the Pain test"-that is, the equipoise standard). And we too recognize that our dicta about that standard are inconsistent with PiperAircraft and hold that a district court must consider all relevant public factors when conducting a forum non conveniens analysis.
We acknowledge that after ruling that the private factors were not in equipoise and so no "exhaustive analysis of the public interest factors" was required, the district court stated that "[n]onetheless, the public interest factors favor[ed] dismissal as well," but we cannot conclude based on this conclusory statement that the district court considered all the relevant public factors. To be sure, the district court discussed the public factor of choice of law, which also touches on aspects of other public factors, see Piper Aircraft , 454 U.S. at 241 n.6, 102 S.Ct. 252, but it did not mention any other public factor. And we cannot be sure that this factor was the only one the district court would have considered relevant had it not been misled by the equipoise standard.
ASF Holland contends that district courts should not be required to address all public factors in all cases because it would include "numerous and unnecessary public factors that will have no bearing on the court's ultimate determination," but this argument misses the mark. Under the correct forum non conveniens analysis, a district court must consider all relevant public factors, not all public factors, as ASF Holland suggests. See id. at 257, 102 S.Ct. 252 ; see also Van Cauwenberghe v. Biard , 486 U.S. 517, 528-29, 108 S.Ct. 1945, 100 L.Ed.2d 517 (1988) (explaining *1052that "some factors may not be relevant in the context of a particular case"). Our rejection of the equipoise standard means that a district court may not bypass the public factors, even when the private factors are not at or near equipoise. We vacate the dismissal of the complaint and remand for the district court to consider all relevant private and public factors.
B. The District Court Must Correct Two Errors when It Reweighs the Private Factors on Remand.
Fresh Results also argues that the district court abused its discretion when it weighed the private factors in favor of dismissal. Because we vacate that dismissal, the district court should also reweigh the private factors on remand. Although it may reach the same result, the district court must correct two errors that it made in its analysis of the private factors.
First, when considering the factor of relative ease of access to sources of proof, the district court disregarded the testimony of the South American growers because they are not parties to this suit, but the relative ease of access to sources of proof is not limited to evidence from the parties. See, e.g. , Ford v. Brown , 319 F.3d 1302, 1308 n.18 (11th Cir. 2003) (considering access to governmental investigations in a suit where the government was not a party). Indeed, the description of the private factors in Piper Aircraft expressly contemplated access to nonparty witnesses, including the "availability of compulsory process for attendance of unwilling ... witnesses" and "the cost of obtaining attendance of willing[ ] witnesses." 454 U.S. at 251 n.6, 102 S.Ct. 252 (quoting Gulf Oil , 330 U.S. at 508, 67 S.Ct. 839 ).
Although Fresh Results presented an affidavit asserting that the South American growers were willing to testify in the United States but not in the Netherlands, the district court disregarded the importance of their testimony because "the Growers are not parties to this lawsuit." That the growers are nonparties is an inadequate basis for ignoring them as a source of proof. To be sure, the district court may find on remand that the importance of the growers' testimony is insubstantial compared to the location of other sources of proof accessible in the Netherlands. See Van Cauwenberghe , 486 U.S. at 528, 108 S.Ct. 1945 (explaining that a district court may consider whether pieces of evidence "are critical, or even relevant," and weigh them accordingly). But it cannot categorically disregard their testimony solely because they are nonparties.
Second, the district court was distracted by a red herring when it reasoned that the enforceability of a possible judgment favored dismissal because no treaty exists between the United States and the Netherlands that governs the reciprocal enforcement of judgments. "There is no bilateral treaty or multilateral convention in force between the United States and any other country on reciprocal recognition and enforcement of judgments." Enforcement of Judgments , U.S. Dep't of State, https://travel.state.gov/content/travel/en/legal/travel-legal-considerations/internl-judicial-asst/Enforcement-of-Judges.html (last visited Apr. 22, 2019) (emphasis added). Relying solely on the absence of a treaty-when no such treaty exists for the United States-was an erroneous basis to weigh this factor in favor of dismissal. And the one-sided reasoning that Fresh Results "would be required to commence new proceedings in Holland to enforce its judgment" overlooked that the enforcement of a judgment may require a separate proceeding no matter in which forum the suit proceeds. For example, if the suit proceeds in the Netherlands and ASF Holland obtains a judgment for its *1053counterclaims, a proceeding in the United States might still be required to enforce that judgment against Fresh Results.
Although the district court must correct these two errors when it reweighs the private factors on remand, it may well conclude that the balance of the private-and public-factors still weighs in favor of the Netherlands as the more convenient forum for this dispute. We express no opinion on the matter. We commit that determination "to the sound discretion of the trial court" in the light of our instructions. Piper Aircraft , 454 U.S. at 257, 102 S.Ct. 252.
IV. CONCLUSION
We VACATE the dismissal of Fresh Results' complaint and REMAND for proceedings consistent with this opinion.