The parties to this appeal both chose a gorilla as the face of their respective brands. Litigation ensued. PlayNation Play Systems, Inc. sued Velex Corporation for trademark infringement. The district court conducted a bench trial and entered judgment for PlayNation. Velex now appeals, arguing that the district court erred in finding trademark infringement, awarding damages to PlayNation, and cancelling Velex's trademark registration. After careful review and with the benefit of oral argument, we affirm the district court's decisions on Velex's liability for trademark infringement and cancellation of Velex's trademark. We vacate and remand, however, on the district court's determination of damages based on Velex's willful infringement.
I. Factual Background
PlayNation has sold children's outdoor playground equipment, including swing sets and attachable swings, ropes, and rings, under its "Gorilla Playsets" brand since 2002. PlayNation owns three registered trademarks for Gorilla Playsets, two registered in 2004 and another in 2014.1 Velex sells "Gorilla Gym" doorway pull-up bars and various attachable accessories for children, including swings, ropes, and rings, under its "Gorilla Gym" brand. Velex successfully registered a trademark for Gorilla Gym in 2014.2 Both PlayNation and *1165Velex's trademarks are registered in Class 28, the broad class of goods for games and other playthings.
II. Trademark Infringement
Velex first argues that the district court erred by concluding that consumers would likely be confused by Velex's use of the Gorilla Gym mark. In an appeal from a bench trial, we review a district court's conclusions of law de novo and its factual findings for clear error. Tartell v. S. Fla. Sinus & Allergy Ctr., Inc. , 790 F.3d 1253, 1257 (11th Cir. 2015). A factual finding is clearly erroneous if, after viewing the totality of the evidence, the court is left with a definite and firm conviction that a mistake was made. Id.
Under the Lanham Act, 15 U.S.C. § 1114(1), a defendant is liable for trademark infringement if the plaintiff shows (1) that its mark has priority and (2) that the defendant's mark is likely to cause consumer confusion. Frehling Enter., Inc. v. Int'l Select Grp., Inc. , 192 F.3d 1330, 1335 (11th Cir. 1999). Velex stipulated to the priority of PlayNation's mark. We thus need only consider whether the district court clearly erred in finding that Velex's mark was likely to cause consumer confusion. In making that determination, we weigh seven factors: (1) the strength of the plaintiff's mark; (2) the similarity of the marks; (3) the similarity of the products the marks represent; (4) the similarity of the parties' retail outlets and customers; (5) the similarity of advertising media; (6) the defendant's intent; and (7) actual confusion. Lone Star Steakhouse & Saloon, Inc. v. Longhorn Steaks, Inc. , 122 F.3d 1379, 1382 (11th Cir. 1997).
We review for clear error both a district court's ultimate decision on likelihood of confusion and its findings on each factor. Frehling , 192 F.3d at 1335. Of these factors, the strength of the mark and actual confusion are the most probative. Id. The district court held that each factor except the defendant's intent weighed in favor of confusion. Velex challenges the district court's determination on the other factors and argues that the district court should have adopted two new factors.
A. Strength of PlayNation's Mark
"Classifying the type of mark Plaintiff has determines whether it is strong or weak. The stronger the mark, the greater the scope of protection accorded it, the weaker the mark, the less trademark protection it receives. There are four categories of marks: (1) generic, (2) descriptive, (3) suggestive, and (4) arbitrary." Frehling , 192 F.3d at 1335 (internal citations omitted). Suggestive marks are stronger and accordingly receive greater protection. Id. "Suggestive marks subtly connote something about the service so that a customer could use his or her imagination and determine the nature of the service." Freedom Sav. Loan Ass'n v. Way , 757 F.2d 1176, 1182 n.5 (11th Cir. 1985). The district court determined that PlayNation's Gorilla Playsets mark was "descriptive or suggestive." The products *1166PlayNation sells under the mark are, of course, not designed for or used by gorillas. At a minimum, the connection between the mark and the product requires some imagination. See Caliber Auto. Liquidators, Inc. v. Premier Chrysler, Jeep, Dodge, LLC , 605 F.3d 931, 938 (11th Cir. 2010) ("A suggestive mark refers to some characteristic of the goods, but requires a leap of the imagination to get from the mark to the product [e.g., Penguin Refrigerators].") (quotation marks omitted).
The district court also reasoned that other factors increased the strength of PlayNation's mark. See Frehling , 192 F.3d at 1336. First, PlayNation has extensively used the mark in commerce since 2002. PlayNation engaged in significant nationwide advertising and promotion using its Gorilla Playsets mark. The district court found that this widespread use indicated consumer recognition, which enhances the strength of a mark. See John H. Harland Co. v. Clarke Checks, Inc. , 711 F.2d 966, 974 n.13 (11th Cir. 1983) (noting that a mark may, "by reason of subsequent use and promotion, acquire such distinctiveness that it can function as a significant indication of a particular producer as source of the goods with which it is used").3
Second, the parties agree that PlayNation's mark has been registered with the Patent and Trademark Office for at least five years and is thus incontestable. Frehling , 192 F.3d at 1336. An incontestable mark "is presumed to be at least descriptive with secondary meaning, and therefore a relatively strong mark." Dieter v. B & H Indus. of Southwest Fla., Inc. , 880 F.2d 322, 329 (11th Cir. 1989) ; see also Frehling , 192 F.3d at 1336.
Finally, the lack of extensive third-party use also strengthens PlayNation's mark. Frehling , 192 F.3d at 1336. Velex argues that the mark is weak because roughly thirty other registered marks in Class 28 contain the word "gorilla." But none of those gorilla marks were used on play equipment. And similar marks used by third parties in unrelated businesses or markets do not diminish the strength of a mark in a particular market. See Safeway Stores, Inc. v. Safeway Disc. Drugs, Inc. , 675 F.2d 1160, 1167 (11th Cir. 1982) (finding that twenty marks containing one of the same words as the plaintiff's mark did "not significantly diminish the strength of the [plaintiff's] mark" because the other marks were used by "businesses not closely related" to the plaintiff's business), abrogated on other grounds , Tobinick v. Novella , 884 F.3d 1110 (11th Cir. 2018) ; cf. Sovereign Military Hospitaller Order of Saint John of Jerusalem of Rhodes & of Malta v. Fla. Priory of the Knights Hospitallers of the Sovereign Order of Saint John of Jerusalem, Knights of Malta, The Ecumenical Order , 809 F.3d 1171, 1186 (11th Cir. 2015) ("That other organizations use parts of the Sovereign Order's word marks is not persuasive evidence of third-party use.").
*1167The district court did not clearly err in determining that PlayNation's Gorilla Playsets mark was strong.
B. Actual Confusion
"It is undisputed that evidence of actual confusion is the best evidence of a likelihood of confusion." Frehling , 192 F.3d at 1340. The actual confusion inquiry turns on both the number of instances of confusion and the type of person confused. Caliber Auto. , 605 F.3d at 936. While numerous instances of actual confusion add greater weight in favor of confusion, "the quantum of evidence needed to show actual confusion is relatively small." Id. at 937 (quotation marks omitted). Evidence that ultimate consumers were actually confused deserves "special attention." Id. at 936-37.
The district court determined that the actual confusion factor weighed in favor of a likelihood of confusion. PlayNation produced two ultimate purchasers of PlayNation's Gorilla Playsets swing sets who contacted Velex for customer service. The consumers believed Velex manufactured both Gorilla Gym and Gorilla Playsets products. Consumers also posted on retail websites selling Gorilla Playsets swings and trapeze bars asking if those items work "with the Gorilla Gym indoors." Velex argues this evidence was insufficient to establish actual confusion because the two consumers were careless and uninformed.4 Velex points to portions of the consumers' testimony in which they confuse the names of the two companies and misremember which name was on the packaging and website they looked at before calling Velex. Velex asserts that a prudent consumer would have examined the packaging, read the product warranty, or researched the product on the internet. Velex argues that had the consumers been so diligent, they would have discovered that PlayNation-not Velex-manufactured the product they purchased.
First, nothing in the actual confusion analysis requires the plaintiff to produce the most careful consumer, who diligently researched and meticulously examined the inner and outer packing to determine the provenance of the product she purchased. It only requires that the consumers be reasonably prudent. The district court found the consumers' testimony probative. That the district court made that determination even though the two witnesses confused the names of the two brands or failed to examine the packing before mistakenly calling Velex for help does not amount to clear error.
Second, the number of instances and the type of consumer also support the district court's actual confusion finding. The customers PlayNation identified are ultimate *1168consumers of the product, to which we give special weight. And the number of instances need not be large to be probative of confusion. See Safeway Stores , 675 F.2d at 1167 (finding one instance of actual customer confusion sufficient and holding that "[a]lthough the number of instances is small, the people confused are precisely those whose confusion is most significant").
The district court thus did not clearly err in holding that the actual confusion factor weighed in favor of a likelihood of confusion.
C. Similarity of Marks
To determine if two marks are similar, the court "considers the overall impressions that the marks create, including the sound, appearance, and manner in which they are used." Frehling , 192 F.3d at 1337. The dominant portion of both marks is an image of a large gorilla and the word "gorilla." The other words, "playset" in one and "gym" in the other, are very similar and do not contribute to distinctiveness. See id. (holding that "a mark may be surrounded by additional words of lesser importance and not have its strength diluted"). The colors and fonts are different, and the gorillas are in different areas and positions in the marks. But even so, given the similarity in the type of gorilla, size of gorilla, and the almost indistinguishable words used in the two marks, those differences do not make the district court's decision clearly erroneous.
D. Similarity of Products, Retail Outlets, and Advertising
The district court did not clearly err in finding that the products, retail outlets, and advertising of PlayNation and Velex were sufficiently similar to weigh in favor of confusion. First, the products are highly similar. The test for similar products is "whether the products are the kind that the public attributes to a single source, not whether or not the purchasing public can readily distinguish between the products of the respective parties." Frehling , 192 F.3d at 1338. Velex argues that it sells only indoor products, while PlayNation sells only outdoor products. But the question is not whether there is a distinguishing characteristic, but whether the goods are so related that a consumer would believe that a single producer made both. See E. Remy Martin & Co., S.A. v. Shaw-Ross Int'l Imports, Inc. , 756 F.2d 1525, 1530 (11th Cir. 1985) (holding that it was not unreasonable for consumers to conclude that a maker of wine could also produce brandy). Both PlayNation and Velex sold a core unit with separate attachable accessories for children, including swings, ropes, and rings. These are sufficiently similar to support the district court's finding.
Second, the retail outlets are similar. "This factor takes into consideration where, how, and to whom the parties' products are sold." Frehling , 192 F.3d at 1339. But "[d]irect competition between the parties is not required for this factor to weigh in favor of likelihood of confusion." Id. Both PlayNation and Velex sold their products through substantially similar retail outlets, including big box stores, their own websites, and other online retailers like Amazon and Walmart. Velex argues that because PlayNation also sells its products through dealers, their retail outlets are dissimilar. But here again, exact parallels are not required, and one distinct retail outlet does not amount to clear error.
Third, PlayNation and Velex engaged in similar advertising. "This factor looks to each party's method of advertising." Id. at 1339. "[T]he standard is whether there is likely to be significant enough overlap in the readership of the publications in which the parties advertise that a *1169possibility of confusion could result." Id. at 1340. PlayNation and Velex both advertised using their own websites, television, and the internet-including Google and Amazon keyword advertising, Facebook, and YouTube.
While there may be some evidence to support dissimilar products, retail outlets, and advertising, they are not enough to compel an alternative finding under clear error review.
These factors are not exclusive. Courts do not have to consider every factor in every case. Swatch Watch, S.A. v. Taxor, Inc. , 785 F.2d 956, 958 (11th Cir. 1986). Courts may also consider other probative evidence that does not fit within the factors. Id.
E. New Factors
Velex proposes two new factors the district court should have considered: first, that Velex had a valid federal trademark registration that covered the infringing products, and second, that PlayNation failed to produce survey evidence suggesting consumer confusion.
To support adoption of its first proposed factor, Velex argues that it ought to have the right to rely on decisions of the government-here, the Patent and Trademark Office (PTO). Velex relies on the Supreme Court's decision in B&B Hardware, Inc. v. Hargis Indus., Inc. , --- U.S. ----, 135 S. Ct. 1293, 191 L.Ed.2d 222 (2015). In B&B Hardware , the Supreme Court held that the likelihood of confusion standard used by the PTO for determining whether to register a new mark and the standard used by courts in trademark infringement cases are "the same." 135 S. Ct. at 1306-07. But the Supreme Court made that conclusion in determining whether issue preclusion applied. Id. The Court held that issue preclusion applied to a district court deciding a Lanham Act trademark infringement claim when the Trademark Trial and Appeal Board (TTAB) had previously decided the parties' marks were not confusingly similar for purposes of registration. Id. at 1307. There is no such finding by the PTO or TTAB in this case because PlayNation did not object to Velex's registration or litigate the issue before the TTAB. Issue preclusion does not bind PlayNation, because it was not a party to any prior proceeding. Cf. Tampa Bay Water v. HDR Eng'g, Inc. , 731 F.3d 1171, 1179 (11th Cir. 2013), overruled on other grounds by CSX Transp., Inc. v. Gen. Mills, Inc. , 846 F.3d 1333, 1340 (11th Cir. 2017).
B&B Hardware does not suggest that, absent a prior administrative finding, a district court deciding a trademark infringement action is bound-or even ought to be persuaded-by the PTO's registration of the defendant's mark. Id. ("[D]istrict courts can cancel registrations during infringement litigation, just as they can adjudicate infringement in suits seeking judicial review of registration decisions."). To the contrary, the Lanham Act expressly gives district courts the power to cancel federal trademarks issued by the PTO when they violate the Lanham Act. See 15 U.S.C. § 1119. Absent issue preclusion, it was not clearly erroneous for the district court not to consider the PTO's decision to approve Velex's trademark as a factor.
It was similarly not clearly erroneous for the district court not to consider Velex's second proposed factor: lack of survey evidence. Lack of survey evidence does not weigh against the plaintiff when determining likelihood of confusion. See Frehling , 192 F.3d at 1341 n.5 ("This Circuit ... has moved away from relying on survey evidence" and "the failure to adduce such evidence is not damaging to [a plaintiff's] case."). We are not persuaded *1170by Velex's reliance on Hickson Corp. v. N. Crossarm Co. , 357 F.3d 1256 (11th Cir. 2004). Hickson was a false advertising case, not a trademark infringement case. Unlike in a trademark infringement action, a plaintiff attempting to establish a misleading advertisement under the Lanham Act "must present evidence of deception in the form of consumer surveys, market research, expert testimony, or other evidence." Hickson , 357 F.3d at 1261 (internal quotation marks omitted). There is no similar requirement for trademark infringement, and we decline to impose one here.
The district court did not clearly err in holding that Velex infringed on PlayNation's trademark.
III. Remedies
Velex argues that the district court erred in fashioning a remedy in this case by first, awarding PlayNation damages based on an accounting of Velex's profits, and second, cancelling Velex's trademark registration. We review a district court's decision to grant relief, including cancelling a party's trademark registration, for abuse of discretion. See Planetary Motion, Inc. v. Techsplosion, Inc. , 261 F.3d 1188, 1203 (11th Cir. 2001).
A. Damages
The Lanham Act allows a successful plaintiff to recover the defendant's profits. 15 U.S.C. § 1117(a). "In assessing profits the plaintiff shall be required to prove defendant's sales only; defendant must prove all elements of cost or deduction claimed." Id. "The Act confers upon the district court a wide scope of discretion to determine the proper relief due an injured party." Burger King Corp. v. Mason , 855 F.2d 779, 781 (11th Cir. 1988) (per curiam). An accounting of a defendant's profits is appropriate where: (1) the defendant's conduct was willful and deliberate, (2) the defendant was unjustly enriched, or (3) it is necessary to deter future conduct. Howard Johnson Co., Inc. v. Khimani , 892 F.2d 1512, 1521 (11th Cir. 1990). Conduct is willful if the infringer was "knowingly and deliberately cashing in upon the good will of [the infringed]." Burger King , 855 F.2d at 781.
The district court concluded that Velex willfully infringed on PlayNation's trademark based solely on Velex's continued sale of its Gorilla Gym products after it was served with PlayNation's complaint. The district court reasoned that PlayNation's complaint put Velex's on notice that it was allegedly infringing on PlayNation's trademark. But when it received PlayNation's complaint, Velex had a valid registered trademark on Gorilla Gym. PlayNation did not obtain a preliminary injunction, and at the time, no court had adjudicated whether Velex was infringing on PlayNation's trademark.
Under the district court's construction of willfulness, a valid trademark holder risks losing all of its profits if it does not immediately cease selling its trademarked product upon receiving a complaint indicating it may be violating the law. The district court determined that at the time Velex obtained its trademark, Velex did not intend to infringe on PlayNation's trademark. And from the time Velex obtained its trademark to the time the district court made its final determination, there was no further adjudication that Velex was infringing. This continued sale under the color of a valid federal trademark cannot alone establish willfulness. See cf. Maltina Corp. v. Cawy Bottling Co., Inc. , 613 F.2d 582, 587 (5th Cir. 1980) (finding willfulness that justified an accounting of profits where the infringer used the infringing mark after the patent office refused to register it based on the confusing similarity of the plaintiff's registered mark).
*1171There may have been other evidence of willfulness, but the district court did not rely on it in its decision. The district court also relied solely on the willfulness theory to support its accounting of profits. The other two theories for supporting an accounting of profits, deterrence and unjust enrichment, do not depend "upon a higher showing of culpability on the part of defendant, who is purposely using the trademark." Burger King , 855 F.2d at 781. Remand is appropriate so that the district court may consider whether other evidence of willfulness exists and whether those alternative theories support an accounting of profits.
B. Cancelling Trademark Registration
District courts have discretion to order the cancelation of a federal trademark registration in whole or in part. 15 U.S.C. § 1119. Cancelling a trademark registration is appropriate where the plaintiff shows (1) that it had standing to petition for cancellation because it was likely to be damaged by the infringer's continued use of the infringing mark, and (2) that there were valid grounds for discontinuing registration. Coach House Rest., Inc. v. Coach & Six Rests., Inc. , 934 F.2d 1551, 1557 (11th Cir. 1991). PlayNation's federal registration of its mark satisfies the first element. See id. at 1557-58. PlayNation must show only the second element-a valid ground for cancelling registration.
If the "mark at issue has been on the federal Principal Register in registrant's name for less than five years, the mark may be cancelled if petitioner can prove that the registration should have been barred in the first instance under Lanham Act § 2." Id. at 1558. "In order to cancel a registration under Lanham Act Section 2(d) in this case, petitioner must prove ... that the registered mark is likely to cause confusion when used in connection with the services of registrant." Id. at 1559. Establishing trademark infringement thus satisfies the second cancellation requirement.
Having concluded that Velex infringed on PlayNation's trademark, it was not an abuse of discretion for the district court to cancel Velex's trademark registration.
IV. Conclusion
The district court did not clearly err in holding that Velex infringed on PlayNation's trademark. The district court similarly did not err in canceling Velex's trademark registration on that basis. But the district court did abuse its discretion in holding that PlayNation was entitled to an accounting of Velex's profits due to willful infringement based solely on Velex's continued lawful use of its mark after Velex was served with the complaint in this action. As a result, we vacate the district court's judgment on that issue, and remand for further proceedings consistent with this opinion.
AFFIRMED IN PART, VACATED AND REMANDED IN PART.