Full opinion text
OPINION OF THE COURT BECKER, Circuit Judge. This is a trade dress infringement action in which plaintiff Versa Products Company, Inc. (“Versa”) contends that defendant Bifold Company (Manufacturing) Ltd. (“Bifold”) infringed the trade dress of Versa’s B-316 directional control valve, a device commonly used in control panels of offshore oil-drilling rigs to facilitate emergency shutdowns, by marketing its Domino Junior valve, which Versa maintains copies the product configuration of the B-816. The action was brought under section 43(a) of the Lanham Act, 15 U.S.C. § 1125(a) (West Supp.1994), New Jersey’s Unfair Competition Law, 56 N.J.S.A. § 4-1 to -2 (1989), and New Jersey’s common law of unfair competition. Following a bench trial the district court found that the trade dress of Versa’s valves had met the nonfunctionality and distinctiveness requirements of our trade dress jurisprudence, and that there was a likelihood of confusion of the sources of Bifold’s Domino Junior and Versa’s B-316 valves. Accordingly, the court entered a permanent injunction against Bifold, precluding it from selling its Domino Junior valve (in its present form) anywhere in the United States. Bifold appeals all aspects of the district court’s rulings. We need not reach the nonfunctionality and distinctiveness questions because the appeal may be disposed of on the likelihood of confusion issue, in connection with which we are called upon to determine whether the jurisprudence that lowers the standard to a “possibility of confusion” where the alleged infringer is a “second comer” applies in the trade dress product configuration context. We also must explicate the elements of the confusion standard in this context. We conclude that the lowered standard (applied by the district court) does not apply and that some but not all of the “Scott factors,” see Scott Paper Co. v. Scott’s Liquid Gold, Inc., 589 F.2d 1225 (3d Cir.1978), are pertinent, because of policy considerations applicable in product configuration cases. Applying this approach we conclude that the district court’s finding of a likelihood of confusion is clearly erroneous. We will, therefore, reverse the order of the district court and vacate the permanent injunction. I. Facts AND Prooedural History Versa, a New Jersey corporation with subsidiaries abroad, designs and manufactures pneumatic and hydraulic directional control valves. Bifold, an English corporation, competes with Versa and markets a line of control valves and related products and services to the offshore oil industry. Versa alleges that Bifold has engaged in unfair competition in its marketing of the Domino Junior — a valve manufactured by Bifold and adapted to the harsh offshore oil and petrochemical environments — by copying the trade dress, i.e., the distinctive appearance, of the product configuration of Versa’s B-316 valve. The most significant feature of valves designed for offshore applications is their stainless steel composition, used to withstand the corrosive effects of salt air and sour gas fumes. In offshore drilling platforms these valves are typically aligned in small control panels containing up to fifty modular valve bodies with a standard configuration but which, by attaching one of various actuators and making minor adjustments, may be adapted to a variety of applications. The panel design engineers devise the functional specifications of panels around the capacities of particular valves, selecting valves based on their functionality, reliability, availability, and price. The valves themselves are not visible from the front of a control panel when installed; only knobs, buttons, and status indicator actuators protrude. A Versa’s B-316 Valve Versa began producing brass valves (its ‘V” series) in 1949. Versa dresses the series, consisting at present of an entire line of valves well known in the industry, with contoured lines and shaping that the district court found “form a distinctive product appearance that has been associated with Versa for decades.” In the late 1970’s Versa designed the B-316 line of stainless steel valves, the subject of this litigation. Versa initially fashioned them of stainless steel bar stock, and the valves were plain and unadorned. Because of the waste of valuable metal associated with the machining process and the substantial manual labor needed to drill each valve individually, Versa converted to a cast version of the valve as soon as sales levels justified the substantial economic investment in a casting mold. The two versions of the valve serve the same function and are interchangeable. Ver-sa deliberately set about to give the cast version of the B-316 the “Versa look,” that is, to have it resemble in appearance the V-series of Versa valves. Versa’s desire to have the valve be clearly associated with Versa in the market was a primary impetus for its election to manufacture the cast version of the B-316. The modular B-316 valve is comprised of a valve body and, optionally, one or more attached actuators used to manipulate the moving parts in the valve body. The valve is a three-way valve, meaning that it has three ports or openings (for the ingress or egress of gas or fluid), which are threaded to NPT, a national standard and industry requirement. The ports open into an inner chamber in the valve body, which houses a spool moved by an actuator (such as a button, knob, or electronically controlled solenoid actuator) to open or close the port, controlling fluid flow. The configuration and function of the actuators provided with a valve are driven by customer demand; however, the use of certain actuators (most of which Versa purchases from other vendors) is standard with the B-316. Versa has used the same knob actuator for 40 years, although many others are available. It also uses a particular status indicator, which indicates the valve’s position; a particular button actuator, which is shrouded to prevent accidental actuation; a self-produced manual latch, which locks the valve spool in the open or closed position; and a self-produced pilot actuator, which responds to pressure in an attached fluid line to control the spool’s position. Located at each end of a valve is a flange, both of which serve as faces to mount the actuators and to provide a flat surface for attachment to the control panels; holes are drilled into the flanges to allow the actuators to be securely mounted. A longitudinal top rib runs along the top of the valve body to allow customers to attach solenoid (computer controlled) actuators, to provide strength, and to serve as a easting gate (an opening in the casting mold through which the molten metal is poured). A smaller bottom rib was added to provide parallelism in the product’s appearance and to assist in the easting process. Finally, the valve has three mounting holes which are positioned to provide stable mounting to a panel or other flat surface. Each aspect of the valve serves a specific function essential to the valve’s operation, cost, performance, or ease of manufacture. The design of each actuator is functional. Functionality dictates the overall cast design, but does not dictate its external appearance. The B-316 valve’s mold imprints the manufacturer’s name (“VERSA”) and place of origin (“N.J. U.S.A.”) on the valve. Versa also stamps a date code and rivets a metal label displaying the Versa name, logo, and part number onto the valve. Versa currently dominates the United States market for stainless steel valves. Aside from Bifold, only Versa sells cast stainless steel valves; other competitors use a cylindrical bar stock. B. Bifold’s DOMINO JUNIOR Series Valve As part of Bifold’s continuing efforts to expand its product line, in 1985 it introduced the Domino series modular valves. Although originally machined from standard bar stock (like Versa’s B-316), the Domino valves became sufficiently successful to warrant the investment needed to design a cast version. An outside casting company designed the cast version of the Domino; Bifold had no involvement in the process. Like a B-316, the Domino valve has three cylindrical ports, a top rib for housing a solenoid feed, flanges, and a range of actuators. Because the Domino was too large for many of its customers’ needs, however, Bi-fold designed the “Domino Junior” modular valve in 1990, producing it at first from bar stock. In late 1991, eight years after the east version of the B-316 became available, Bifold introduced its cast version of the Domino Junior. Bifold at the time erroneously believed that Versa had a “monopoly” on the wellhead control panel valve market, and ere-, ated the cast version of the Domino Junior to “bury Versa.” Bifold was aware of the B-316’s appearT anee and design features because it had seen the product at various trade shows. The district court did not credit Bifold’s claims that it designed the Domino Junior as a scaled down version of the Domino and that it did not copy the B-316. It found instead that, before and during its design of the Domino Junior east mold, Bifold examined and largely copied Versa’s B-316 valve, a sample of which it had obtained through its agent in Denmark. For example, the court found that Bifold, which regularly uses metric sizes in its valves, took measurements from the B-316 and used a metric conversion of the B-316’s imperial standard size. Bifold also sent a cast B-316 valve to Manchester Tool Services, which it selected to be the manufacturer of the cast version of the Domino Junior, as a model for the cast version of the Domino Junior. The two valves (the B-316 and the Domino Junior) are not, however, interchangeable in the field, and “replacing a Versa B-316 cast valve with a Bifold Domino Junior cast valve in an existing control panel could be problematic.” The district court also found that Bifold lacked the expertise to design the cast version of the Domino Junior, had presented misleading testimony about who produced a prototype drawing of the Domino Junior, and had backdated documents to create the false appearance that it had designed the valve. The court found that BSA Precision Castings, Ltd. (“BSA”), which had designed the cast version of the Domino valve, was involved in the design of the cast version of the Domino Junior valve. During the time that BSA participated in the design of the cast version of the Domino Junior, it had in its possession drawings and castings of the B-316’s major components, as well as Versa’s actuators. BSA had obtained the drawings, actuators, and actual samples of valve components from Versa three years earlier when it had provided Versa with a quote for the casting of the B-316. The court found that BSA used this data to design a casting of the Domino Junior valve as a look-alike of the Versa B-316, despite the fact that Bifold had asked BSA to give “due consideration to the appearance of the larger Domino valves.” Although the Domino Junior and B-316 do, in fact, look quite similar, the district court described a number of differences between the two manufacturers’ valves. Because the Domino Junior’s ports extend outward from the valve body, whereas the B-316’s ports are flush with the valve body, the Domino Junior is slightly wider than the B-316. The Domino Junior ports’ threads have metric dimensions. The Domino Junior is made of more metal and weighs about thirty percent more than the B-316. The stroke — the distance an actuator must move the internal spool to switch the valve’s ports — also differs from that of the B-316. Because Bifold electroplates its valves, the Domino Junior body has a duller finish than the B-316. The Domino Junior also has thicker flanges and ribs than does the B-316. The valves’ mounting holes have different centers. Bi-fold purchases its status indicators from a different company than Versa uses, and these indicators have a different size, appearance, and configuration from Versa’s. Like Versa, Bifold produces its own manual latch, but makes it out of two cast pieces and gives it rounded edges, whereas Versa uses a single piece of rectangular bar stock material. The Domino Junior valves are “block-before-bleed” (meaning fluid flow is completely blocked during the moment required to change states when the valve is activated), whereas the B-816 includes that feature only as an option. The Domino Junior ports are universal (meaning that any port can be an inlet or an outlet port), though Bifold customizes each valve as either normally open or normally closed; in contrast, Versa offers universal application only as an option. Like Versa, Bifold casts its name into the Domino Junior’s valve body and bolts onto it a metal label prominently displaying the Bifold name. Despite all these differences, the court concluded that “[t]he Bifold Domino Junior valve and actuators are virtually identical in external design and visual appearance to the Versa B-316 valve.” It found an identity in general body configuration, body length, flanges, distance between ports, valve mounting holes, actuator mounting holes, ribs, spring cap (which returns the spool back to its original position), buttons, button caps, status indicators, knobs, manual latching pins, and pilot caps. The valve bodies, buttons, button caps, knobs, status indicators, and manual latching mechanism used by other competitors in the industry look quite different from both the Domino Junior’s and the B-316’s. The court discounted the dissimilarity of the solenoid actuators attached to the valve bodies by Versa and Bifold because only a small portion of Versa’s B-316 valves sold include solenoid actuators. C. Marketing and Sales of the Valves In order to determine whether Bifold had engaged in unfair competition with Versa, the district court considered whether consumers were likely to confuse the sources of the two companies’ valves in light of the ways in which the two valves are marketed and sold. The court found that valves of this sort are not sold off the shelf or selected on sight. Rather, both manufacturers sell their valves based on functional specifications detailed in schematic diagrams, manufacturers’ catalogs, or specification sheets and samples available at trade shows and sales presentations. The valves are selected by multi-digit part numbers identifying the particular variation desired. The purchasers and users of the valves are qualified, knowledgeable persons who comprehend the installation and use of the valves. They typically prepare specifications designating which manufacturer’s valve they prefer to use in their system before placing the order. Versa, the more established manufacturer, has sold over 100,000 B-316 valves, and is currently selling over 16,000 per year. This gives it a fifty to fifty-five percent market share of valves sold for use in emergency shutdown systems in the United States. Bi-fold has only recently begun marketing its Domino Junior valve in the United States, and immediately stopped its efforts to open the United States market pending the outcome of this litigation. Versa and its B-316 valve have an excellent reputation for producing a high quality product. This quality level is very important in emergency shutdown offshore drilling, for the failure of a valve could cause loss of human life and property as well as severe environmental damage. Versa has therefore subjected its B-316 valve to rigorous quality control tests, and the valve has performed faithfully in the field. The district court found that, because of the availability and outstanding reputation of the B-316 cast valve for over ten years, the overall appearance and contours of the B-316 have come to distinguish the valve as Versa’s (in the industry and to Versa’s customers), with the valve body constituting the most defining aspect of the Versa look. The valve’s overall appearance assists Versa in marketing its product, and Versa features this appearance widely in trade journals, cat-alogues, brochures, bulletins, trade shows, and sales exhibitions. The court also found that Versa deliberately created the B-316’s distinctive appearance (contrasting with the other valves available in the market at the time) to identify the valve in the market as a Versa product. The district court found that Bifold hired Versa’s former regional marketing manager of six years, James Carr, III, to sell its new Domino Junior valve. Carr sells both Versa and Bifold valves, sometimes to the same customers. To solicit Domino Junior sales in early 1993, Carr approached Gordon Fra-leigh, an employee of the Fraleigh Company, which for years had sold Versa products to customers in the petroleum industry. Carr represented to Fraleigh, whom he had known for many years, that Bifold had developed an “exact copy” of the B-316 which would “fit in perfect” as a substitute for the B-316, and inquired whether Fraleigh would like to distribute the Domino Junior. Thereupon, Fra-leigh became confused as to the relationship between Versa and Bifold. Bifold has also contemplated contacting other Versa distributors to sell its valves. D. Likelihood, of Confusion Beyond the confusion on Fraleigh’s part, the district court found that some, but not all, consumers of the B-316 valves are sophisticated, and that the likelihood of confusion as to the source of the Domino Junior is enhanced with respect to the unsophisticated consumers. This finding is undercut by the district court’s finding, see supra p. 196, that the purchasers are knowledgeable and understand the valves. We note, however, that these findings might be resolved by noting that the latter finding assumes expansion of Versa’s market. Indeed, the opinion later suggests that any unsophisticated consumers exist only as “potential new customers” in “untapped” markets, that is, that all the current customers are sophisticated. See Mem. Op. at 90. Because there will in those potential expansion markets be no likelihood of confusion based on the district court’s theory (namely, that consumers familiar with Ver-sa’s trade dress will mistakenly believe Bifold is affiliated with Versa, see infra at 214-16), we will ignore the court’s “expansion” finding and consider only its finding that the purchasers are knowledgeable. In addition, one customer forwarded to Bifold a telefax initially addressed to Versa. And Frank Vetter, a Vice President and Chief Operating Officer of Versa, testified that he was advised of confusion of the products at trade shows. Because of the valves’ “virtual identity in appearance” and the fact that Bifold has not sold products in the United States previously, the district court concluded that a Versa customer in the United States might reasonably assume that Versa and Bifold are related companies or that the Domino Junior is otherwise related to Versa. However, since Bifold had sold only two valves in the United States, both to distributors, the court found there had been little opportunity for Versa to document instances of actual confusion. II. The DISTRICT Court’s Legal Conclusions The district court noted at the outset that Section 43(a) provides a cause of action for unprivileged imitation of trade dress—defined as the “overall design or appearance of a product or its packaging”—because it involves actual or potential deception. Trade dress, it held, consists not of individual features, but of the overall appearance of the product. We ton then, to the court’s more specific legal conclusions. Recognizing that unpatented functional features may be freely copied regardless of any likelihood of confusion—and Versa has not patented the design of its B-316 valve—the district court first concluded that the trade dress of the B-316 was nonfunctional. The court next concluded that the B-316’s trade dress was “inherently distinctive,” or, in the alternative, possessed acquired distinctiveness because it had acquired secondary meaning. The distinctiveness finding is problematic because the district court evaluated inherent distinctiveness using a legal standard that this court has since held to be improper. See Duraco Prods., Inc. v. Joy Plastic Enters., Ltd., 40 F.3d 1431, 1441-42 (3rd Cir.1994) (rejecting the trademark distinctiveness taxonomy as the measure of inherent distinctiveness for trade dress in product configurations). And although we can scarcely blame the district court, whose analysis of functionality largely tracked this court’s various legal formulations, its finding of nonfunctionality is also problematic in view of the conflicting formulations of functionality used, as outlined in the margin. Although we have misgivings about these two issues, it is the district court’s third conclusion that forms the focus of our opinion today. The district court held that to prevail on a trade dress infringement claim, a plaintiff must demonstrate a likelihood of confusion, but not actual confusion. It held that Versa could do so here if it could show that an appreciable number of buyers are likely to become confused as to the origin of the Domino Junior valve. Importantly, the court further held that the threshold for likelihood of confusion is lower when a newcomer (or “second comer”) violates a long-established trade dress. The district court then seemed to apply the ten factors for likelihood of confusion that this court enumerated in Scott Paper Co. v. Scott’s Liquid Gold, Inc., 589 F.2d 1225 (3d Cir.1978). See CL 57. Under Scott, the threshold issue is the question of similarity of product appearances, and the court found that the Domino Junior’s appearance was “virtually identical” to the B-816’s and hence that there was a likelihood of confusion. The district court reasoned that Bifold’s clear designation on the product that it was the manufacturer, while relevant to Bifold’s duty to take reasonable steps to prevent deception, was only one factor to be assessed in resolving the confusion issue. The district' court found that “[a]n intent to copy trade dress and/or finding of copying by a junior user is often alone dispositive of a finding of likelihood of confusion,” and that since Bifold had copied Versa’s design there was a likelihood of confusion. The court also found a likelihood of confusion because of the “competitive proximity” of the goods, which “strongly favors a finding of confusion,” since the court found that the Domino Junior can replace the B-316 at the point of conception of the panels. Although the district court concluded that Versa was not entitled to damages because Bifold had only sold two of its valves in the United States, it granted Versa permanent injunctive relief on the ground that the company’s goodwill was threatened by Bifold’s attempt to reap the benefits of Versa’s reputation (by basing the appearance of the Domino Junior on a Versa product). The court determined that the injunction had to cover not only the appearance of the article actually the subject of the lawsuit, but also all “confusingly similar” appearances. It therefore crafted an injunction enjoining Bifold from manufacturing, selling, etc., any cast valve which “has an external design and visual appearance confusingly similar to the cast Versa B-316 valve, described herein and shown in Exhibit A.” Order and Injunction at 5. The court then held Versa to be entitled to attorneys’ fees. Recognizing that attorneys’ fees can be awarded to the “prevailing party” only in “exceptional cases,” the court found Bifold’s deliberate and willful infringement to be exceptional. This appeal followed. We have jurisdiction under 28 U.S.C. § 1291. III. Discussion To obtain trade dress protection for the B-316 under section 43(a) of the Lanham Act, Versa had to prove that (a) the design was non-functional, (b) the design was inherently distinctive or distinctive by virtue of having acquired secondary meaning, and (c) there was a likelihood of confusion. See Two Pesos, Inc. v. Taco Cabana, Inc., — U.S. -, -, 112 S.Ct. 2753, 2758, 120 L.Ed.2d 615 (1992). As discussed supra, the district court found that Versa had met each of these requirements, and it therefore permanently enjoined Bifold from copying the B-316’s trade dress and ordered Bifold to pay attorney’s fees. We limit our discussion to the final element Versa needed to establish to prevail on its trade dress infringement claim — the likelihood of consumer confusion as to the source of Bifold’s Domino Junior. Such consumer confusion is, of course, at the heart of trademark law. See, e.g., Freixenet, S.A. v. Admiral Wine & Liquor Co., 731 F.2d 148, 151 (3d Cir.1984). Likelihood of confusion is a factual matter, subject to review for clear error, see Ciba-Geigy Corp. v. Bolar Pharmaceutical Co., 747 F.2d 844, 851 (3d Cir.1984), which exists when, “giving all due deference to the opportunity of the trial judge to evaluate the credibility of witnesses and to weigh the evidence,” Litton Sys., Inc. v. Whirlpool Corp., 728 F.2d 1423, 1445 (Fed.Cir.1984) (emphasis omitted) (citing Inwood Lab., Inc. v. Ives Lab., Inc., 456 U.S. 844, 855, 102 S.Ct. 2182, 2188, 72 L.Ed.2d 606 (1982)), we are “left with a definite and firm conviction that a mistake has been committed,” Anderson v. Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985). A. “Likelihood” vs. “Possibility” of Confusion Generally, “the law does not require that a competitor insure against all possible confusion or the likelihood thereof.” Charles E. MgKenney & George F. Long, III, Federal Unfair Competition: Lanham Act § 43(a), § 3.08[1], at 3-71 (1989, Release #5, May 1994) [hereinafter McKenney & Long, Federal Unfair Competition]. Rather, a plaintiff may prevail in a trade dress infringement action only if it shows that an appreciable number of ordinarily prudent consumers of the type of product in question are likely to be confused as to the source of the goods. See, e.g., Nikon, Inc. v. Ikon Corp., 987 F.2d 91, 94 (2d Cir.1993); West Point Mfg. Co. v. Detroit Stamping Co., 222 F.2d 581, 589 & n. 2 (6th Cir.1955). “The mere possibility that a customer may be misled is not enough.” Surgical Supply Serv., Inc. v. Adler, 321 F.2d 536, 539 (3d Cir.1963). Although this usual formulation of trade dress infringement requires a showing of a likelihood or probability of confusion, this standard has been relaxed in some cases. Where an alleged infringer was new to an area and the plaintiff was well-established, this court has at times replaced the “likelihood of confusion” requirement with a lower “possibility of confusion” standard. These cases have all involved actions for trademark or tradename infringement, not trade dress, and certainly not trade dress alleged in a product configuration. See Merchant & Evans, Inc. v. Roosevelt Bldg. Prods. Co., 963 F.2d 628, 637-38 (3d Cir.1992) (considering “possibility of confusion” with respect to “a name or mark,” in particular, a “ ‘Z’ logo” alleged to be confusingly similar to a “ ‘Zip-Rib’ trademark”); Country Floors, Inc. v. Partnership Composed of Gepner & Ford, 930 F.2d 1056, 1065 (3d Cir.1991) (directing application of “possibility of confusion” standard to “Country Tiles” and “Country Floors” names or marks); Telechron, Inc. v. Telicon Corp., 198 F.2d 903, 908-09 (3d Cir.1952) (“a case of a first coined word and a second coined word resembling it”). We must therefore consider whether the “possibility of confusion” standard should govern product configuration trade dress cases. Since unfair competition law regarding product configurations will diverge substantially in its incidents from the law regarding product packaging, Duraco, 40 F.3d at 1439, we begin our consideration by examining the rationale underlying the “possibility of confusion” cases. Telechron, Inc. offered some explanation for the lowering of the requirements for showing trademark infringement in certain situations. In that case the plaintiff used the name “Telechron” starting in 1919 as a trademark for its electric clocks and other timing and switching devices. The defendant Telicon Corporation began marketing radio and television sets under the “Telicon” name for the first time in 1946. Agreeing with the district court, this court held that “ ‘Telicon’ is a colorable imitation of ‘Telechron’ within the conception of trade-mark infringement.” Telechron, Inc., 198 F.2d at 908. In an opinion by Judge Hastie, the court explained that the “‘degree of resemblance necessary to constitute an infringement is incapable of exact definition.’” Telechron, Inc., 198 F.2d at 908 (quoting McLean v. Fleming, 96 U.S. (6 Otto) 245, 251, 24 L.Ed. 828 (1877)). We emphasized the strong aural similarity of the marks and the evidence of actual confusion, concluding that the evidence was “adequate substantiation of tendency to confusion inherent in the obvious similarity of the words themselves.” Telechron, Inc., 198 F.2d at 908. Only then, expressly as an additional consideration, did we observe that this was “a type of case where a court properly requires the second comer to stay clearly away from the original mark,” and thus that “ ‘any possible doubt of the likelihood of damage should be resolved in favor of the [first user].’ ” Id. at 908-09 (quoting Lambert Pharmacal Co. v. Bolton Chem. Corp., 219 F. 325, 326 (S.D.N.Y.1915) (Learned Hand, J.)). We recognize that application of the “keep clear” policy embodied by the trademark “possibility of confusion” standard would not be entirely senseless in the context of alleged infringement of trade dress, even where the dress consists not in a product’s packaging but in a non functional product configuration. To the extent that product configurations are protectable, a Johnny-come-lately copier arguably creates a greater risk than one who more promptly markets a copy that consumers will be misled by a substantially identical configuration into thinking the newcomer’s product to be that of the established business, for there will have been more time for the public to come to associate that configuration with a single source. In and of itself, however, that is no reason to change the measure of confusion (from “probability” to “possibility”) required to make out a Lanham Act violation. Rather, it is at most a factor properly taken into account in assessing the likelihood of confusion. The trademark “possibility of confusion” standard must therefore be supported by other considerations. We believe that the primary reasons for lowering the measure of confusion when a newcomer copies an established trademark are the general lack of legitimate reasons for copying a competitor’s mark, see, e.g., American Chicle Co. v. Topps Chewing Gum, Inc., 208 F.2d 560, 562-63 (2d Cir.1953) (“ “Why [the defendant] should have chosen a mark that had long been employed by [the plaintiff] and had become known to the trade instead of adopting some other means of identifying its goods is hard to see unless there was a deliberate purpose to obtain some advantage from the trade which [the plaintiff] had built up.’ ”), and the high degree of reliance by consumers on trademarks as indicators of the source of products. Whether or not these considerations translate to the realm of product packaging, we think that with respect to product configurations the significance of each of the factors is greatly diminished. First, the mere copying of product configurations does not suggest that the copier was necessarily trying to capitalize on the good will of the source of the original product. See Duraco, 40 F.3d at 1453; see also infra at 205-08 (discussing implications of defendant’s intent to copy). A presumption to the contrary would be mandated, if ever, only in the narrow class of cases where both (1) a product configuration is desirable to consumers primarily because of the configuration’s inherent or acquired identification with the original source, and (2) the copier adopts affirmatively misleading labeling and/or marketing for the copied product, cf. Quaker Oats Co. v. General Mills, Inc., 134 F.2d 429, 432 (7th Cir.1943) (“The pirate flies the flag of the one he would loot. The free and honorable non-pirate flies the colors of his own distinctive ensign.”). Second, although a product’s trade dress in the form of its configuration could function as an indicator of the product’s source, product configurations in general are not reliable as source indicators, for functional configurations are not protected and thus may be freely copied, see Duraco, 40 F.3d at 1441, 1448-49, 1451, and inherently distinctive configurations will be rare, see id. at 1446. Since substantially identical products are often sold by different manufacturers under different names, consumers are accustomed to relying on product packaging and trademarks to identify product sources. Indeed, if any modification of the likelihood of confusion standard is justified in the product configuration context, the standard might well be heightened, perhaps to a “high probability of confusion.” Nevertheless, we see no need to adopt such a standard today, preferring for now merely to reject the “possibility of confusion” standard for product configuration infringement cases, and adhering to the conventional “likelihood of confusion” standard. B. The Scott Factors in the Product Configuration Context Having concluded that the appropriate standard in this product configuration trade dress infringement action is a likelihood of confusion, we must determine what that inquiry entails in this context. Although the law of trade dress in product configurations will differ in key respects from the law of trademarks or of trade dress in product packaging, settled law provides the starting point for our analysis. We stated in Ford Motor Co. v. Summit Motor Prods., Inc., 930 F.2d 277, 297 (3d Cir.), cert. denied, 502 U.S. 939, 112 S.Ct. 373, 116 L.Ed.2d 324 (1991), that the analysis of the likelihood of confusion requires a court to evaluate a number of factors: (1) the degree of similarity between the owner’s mark and the alleged infringing mark; (2) the strength of [the] owner’s mark; (3) the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase; (4) the length of time [the] defendant has used the mark without evidence of actual confusion arising; (5) the intent of the defendant in adopting the mark; (6) the evidence of actual confusion; (7) whether the goods, though not in competition, are marketed through the same channels of trade and advertised through the same media; (8) the extent to which the targets of the parties’ sale efforts are the same; (9) the relationship of the goods in the minds of the public because of the similarity of function; (10) other facts suggesting that the consuming public might expect the prior owner to manufacture a product in the defendant’s market. Id. at 293 (citing Scott Paper Co. v. Scott’s Liquid Gold, Inc., 589 F.2d 1225, 1229 (3d Cir.1978)); accord Charles Jacquin et Cie, Inc. v. Destileria Serralles, Inc., 921 F.2d 467, 474-75 (3d Cir.1990); cf. Restatement (ThiRd) of Unfair Competition § 21 (Tent. Draft No. 2, Mar. 23, 1990). This test was developed not for product configuration cases but for “cases of alleged trademark infringement and unfair competition by a producer of a non-competing product,” see Fisons Horticulture, Inc. v. Vigoro Indus., Inc., 30 F.3d 466, 473 (3d Cir.1994), and not all of the factors will be appropriate for or function the same way with respect to trade dress inhering in a product configuration, so we consider them in turn. 1. Similarity of Appearance (Scott Factor 1) In trademark infringement cases, the first and primary factor to be considered in the likelihood of confusion inquiry is “the degree of similarity between the owner’s mark and the alleged infringing mark.” See Ford Motor Co., 930 F.2d at 293. In trade dress infringement cases where product packaging is at issue, the corresponding factor is the similarity of the protectable trade dress. Similarity of appearance is properly considered paramount in trademark and product packaging trade dress infringement cases, for unless the allegedly infringing mark or dress is substantially similar to the protecta-ble mark or dress, it is highly unlikely that consumers will confuse the product sources represented by the different marks or trade dresses. For the same reason, substantial similarity of appearance is necessarily a prerequisite to a finding of likelihood of confusion in product configuration cases. Unlike in trademark or product packaging trade dress cases, however, a finding of substantial similarity of trade dress in a product configuration does not by itself strongly suggest a likelihood of confusion. Consumers have grown accustomed to relying on trademarks as trustworthy indicators of the source of the product: that is the point of a trademark. Perhaps to a somewhat lesser extent, consumers also rely on other aspects of product packaging to identify the manufacturer. Such behavior is rational, for in a trademark or product packaging ease, all the consumer usually has to go on to identify the source of the product is the trademark and packaging (and any marketing featuring that mark or packaging). In a product configuration trade dress infringement case, by contrast, consumers do not have to rely on a potentially distinctive configuration to identify the source of the product; rather, they can generally look to the packaging, trademarks, and advertising used to market the product, which are typically much less ambiguous. Consumers therefore have less need, and so are much less likely, to rely on a product configuration as an indicator of the product’s source. Accordingly, they are less likely to be confused as to the sources of two products with substantially similar configurations. Thus, in trade dress infringement suits where the dress inheres in a product configuration, the primary factors to be considered in assessing likelihood of confusion are the product’s labeling, packaging, and advertisements. “The most common and effective means of apprising intending purchasers of the source of goods is a prominent disclosure on the container, package, wrapper, or label of the manufacturer’s or trader’s name ... [and when] that is done, there is no basis for a charge of unfair competition.” Venn v. Goedert, 319 F.2d 812, 816 (8th Cir.1963), quoted in Litton Sys., Inc., 728 F.2d at 1446. Indeed, except where consumers ordinarily exercise virtually no care in selecting a particular type of product (as may be the case with inexpensive disposable or consumable items, cf. Venn, supra (cookies)), clarity of labeling in packaging and advertising will suffice to preclude almost all possibility of consumer confusion as to source stemming from the product’s configuration. Cf. Bose Corp. v. Linear Design Labs, Inc., 467 F.2d 304, 309 (2d Cir.1972) (“The presence of [the source’s] name on the product [stereo speaker cabinets] goes far to eliminate confusion of origin.”) (emphasis supplied); id. at 310 (“[T]here is hardly likelihood of confusion or palming off when the name of the manufacturer is clearly displayed.”). 2. Strength of the Owner’s Mark (Scott Factor 2) In trademark cases, the strength of the owner’s mark directly affects the likelihood that consumers will be confused as to the sources of products bearing substantially similar marks. Strength includes both “[distinctiveness on the scale of trademarks” and “[c]ommercial strength, or marketplace recognition.” Fisons Horticulture, Inc., 30 F.3d at 479. A strong trademark is thus one that carries widespread, immediate recognition that one producer (even if unknown) is associated with the mark, and so with the product. If a second comer adopts a mark substantially identical to a strong mark, there is a correspondingly high likelihood that consumers will mistakenly associate the newcomer’s product with the owner of the strong mark. The same may be said of a “strong” trade dress consisting of a product’s packaging. But these observations do not translate literally into the product configuration context. As we have explained elsewhere, the trademark distinctiveness scale is ill-suited for application to trade dress inhering in a product configuration. See Duraco, 40 F.3d at 1440-12. Having rejected the distinctiveness scale in this context, we are left with commercial strength as the measure of trade dress strength in a product configuration. Yet strength of a product configuration must mean more than the ability of large numbers of consumers to identify the configuration as coming from a particular producer. This would sanction too much reliance by consumers on product designs that, lacking the protection of a patent, are in large measure copyable at will. Cf. Duraco, 40 F.3d at 1447-48 (criticizing the “capable of distinguishing” interpretation of distinctive trade dress). Rather, “strength” of product configuration as relevant to determining likelihood of confusion on the part of ordinarily careful consumers should be found only if consumers rely on the product’s configuration to identify the producer of the good. This may perhaps be the ease with products purchased largely because of their appearance, such as “Carebears,” cf. American Greetings Corp. v. Dan-Dee Imports, Inc., 807 F.2d 1136, 1142 (3rd Cir.1986). Such focus, however, is not generally found in and should not be encouraged in the industrial design context, where product appearance typically plays a lesser role in buyers’ selection processes. Hence, to differentiate between these types of product configuration eases, courts should require evidence of actual reliance by consumers on a particular product configuration as a source indicator before crediting that configuration’s “strength” toward likelihood of confusion. 3. Attention Expected of Consumers (Scott Factor 3) The third Scott factor is “the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase.” “The greater the care and attention, the less the likelihood of confusion.” Fisons Horticulture, Inc., 30 F.3d at 476 n. 12. We believe that this factor takes on enhanced importance when a claim is made for infringement of trade dress in a product configuration, both as a result of the intersection of the patent laws with the Lan-ham Act, and as a function of the difference between a trademark and a product configuration. The penumbra of the federal patent laws restricts the degree to which courts may grant legal recognition of consumer reliance on product configurations as source indicators, for their limited scope of protection impliedly imposes restraint on the workings of Section 43(a). Accordingly, we must bear in mind the Supreme Court’s counsel that “mere inability of the public to tell two identical articles apart is not enough to support an injunction against copying ... that which the federal patent laws permit to be copied.” Sears, Roebuck & Co. v. Stiffel Co., 376 U.S. 225, 232, 84 S.Ct. 784, 789, 11 L.Ed.2d 661 (1964). “[T]he federal policy, found in Art. I, § 8, cl. 8, of the Constitution and in the implementing federal statutes, of allowing free access to copy whatever the federal patent and copyright laws leave in the public domain,” Compco Corp. v. Day-Brite Lighting, Inc., 376 U.S. 234, 237, 84 S.Ct. 779, 782, 11 L.Ed.2d 669 (1964), is “an ever-present consideration,” McKenney & Long, Federal Unfair Competition § 5.03, at 5-25. Furthermore, one expects a consumer exercising ordinary care to ascertain the source of a product to rely much more on packaging, trademarks, and advertising, which if not deceptive tend to reveal the product’s source unambiguously, than on the product configuration, which usually does not contain an explicit statement of the producer’s identity. While it might be shown that consumers in fact rely on a particular product’s configuration to identify its source, such deviation from the normal pattern (i.e., from reliance on trademarks, packaging, and advertising) would be rare. Because clear labeling thus should generally be legally and factually sufficient to remedy confusion where unpatented product configurations are at issue, clarity of labeling (and marketing) must be taken into account in considering whether there is a likelihood that consumers exercising ordinary care will be confused as to the sources of substantially identical products. Much as courts are required to police the boundaries of similarity within which a jury may be permitted to find a likelihood of confusion under the Lanham Act, Country Floors, Inc., 930 F.2d at 1063, courts must also establish the perimeters of ordinary care that constrain likelihood of confusion. The following non-exhaustive considerations should guide a court’s determination of the standard of ordinary care for a particular product. Inexpensive goods require consumers to exercise less care in their selection than expensive ones. The more important the use of a product, the more care that must be exercised in its selection. In addition, “the degree of caution used ... depends on the relevant buying class. That is, some buyer classes, for example, professional buyers ... will be held to a higher standard of care than others. Where the buyer class consists of both professional buyers and consumers, -the standard of care to be exercised by the reasonably prudent purchaser will be equal to that of the least sophisticated consumer in the class.” Ford Motor Co., 930 F.2d at 293. 4- Actual Confusion or Lack Thereof (Scott Factors 4 & 6) The fourth Scott factor is “the length of time defendant has used the mark without evidence of actual confusion arising.” While we hold that this factor applies to product configuration cases as well as to trademark and product packaging cases (for it is obviously relevant), we take this opportunity to underscore the role of the “lack of actual confusion” factor. If a defendant’s product has been sold for an appreciable period of time without evidence of actual confusion, one can infer that continued marketing will not lead to consumer confusion in the future. The longer the challenged product has been in use, the stronger this inference will be. “Evidence of actual confusion” (the sixth Scott factor bearing on likelihood of confusion) is similarly relevant: the more evidence of actual confusion that a plaintiff can muster, the stronger the likelihood of confusion in the future, but lack of evidence of actual confusion (at least where the time period that the two products have been in competition is short or “when the particular circumstances [do not] indicate such evidence should have been available,” AMF Inc. v. Sleekcraft Boats, 599 F.2d 341, 353 (9th Cir.1979)) does not raise an inference that there is no likelihood of confusion. As the case law makes clear, proof of actual confusion is not required for a successful trade dress infringement action under the Lanham Act. Ford Motor Co., 930 F.2d at 292 (quoting Opticians Ass’n v. Independent Opticians, 920 F.2d 187, 195 (3d Cir.1990)); accord 2 J. Thomas McCaRthy, TRADEMARKS AND UnfajR Competition § 23:2 (2d ed. 1984); id. § 23:20. We see no reason that these factors would not also apply to product configuration cases. However, we emphasize again, see supra at 204-05, that to make out unfair competition a plaintiff must show a likelihood that a consumer exercising ordinary care to discover the identity of the source would suffer confusion or be mistaken because of the appearance of the allegedly infringing product configuration. Thus, instances of actual confusion may not weigh in favor of a finding of likelihood of confusion unless the confused consumer was acting with the care expected of consumers purchasing the type of good at issue. See G.D. Searle & Co. v. Hudson Pharmaceutical Corp., 715 F.2d 837, 840 & n. 6 (1983) (ignoring testimony of witness who “was not acting as a reasonably prudent consumer of the type of goods in issue when purchasing the product”). 5. Defendant’s Intent (Scott Factor 5) The fifth Scott factor is “the intent of the defendant in adopting the mark.” Whatever merit this factor may have in the context of trademark and product packaging trade dress cases, we doubt that it is an appropriate consideration in a trade dress infringement case where the trade dress is alleged in the product configuration itself. In the likelihood of confusion inquiry in trademark cases and product packaging trade dress cases, we do not focus on a defendant’s bare intent to adopt a mark or product packaging substantially identical to a plaintiff’s mark or packaging, since there is little basis in fact or logic for supposing from a defendant’s intent to copy (without more) that the defendant’s actions will in fact result in confusion. Thus, what we have held is that a defendant’s intent to confuse or deceive consumers as to. the product’s source may be highly probative of likelihood of confusion. See American Home Prods. v. Barr Lab., Inc., 834 F.2d 368, 371 (3d Cir.1987) (product packaging case — color of pain relief medication); see also Fisons Horticulture, Inc., 30 F.3d at 479-80 (identifying “intent of promoting confusion and appropriating the prior user’s good will” as appropriate inquiry in forward confusion cases) (internal quotation marks omitted) (trademark case — marks “Fairway” and “Fairway Green”); Sands, Taylor & Wood Co. v. Quaker Oats Co., 978 F.2d 947, 960 (7th Cir.1992) (“[T]he defendant’s intent is relevant to the issue of likelihood of confusion only if he intended to palm off his products as those of another.” (internal quotation marks omitted) (trademark ease— words “Thirst Aid” used in advertising campaign)); First Brands Corp. v. Fred Meyer, Inc., 809 F.2d 1378, 1385 (9th Cir.1987) (“Intent of a defendant in adopting his trade dress is a critical factor, since if the trade dress were adopted with the intent of depriving benefit from the reputation of the plaintiff, that fact alone may be sufficient to justify the inference that there, is a confusing similarity.”) (emphasis supplied) (product packaging case — color and shape of antifreeze jug). Because American Home Products involved a claim that the color of a rival producer’s ibuprofen tablet infringed the trade dress of the plaintiffs Advil tablet, we believe that the case is closer to a product packaging ease than a product configuration case. Even were we to consider it a product configuration case, however, American Home Products is consistent with our present discussion of defendant’s intent. Judge Seitz’s opinion did not hold that independent significance must be accorded a defendant’s mere intent to copy; rather, it held that “intent to confuse might be highly probative of likelihood of confusion” and that “[a]t most, defendant’s intent is a factor tending to suggest likelihood of confusion.” American Home Prods., Inc., 834 F.2d at 371 (emphases supplied). In what follows, we simply clarify this circuit’s intent-to-confuse rule for product configuration cases, delineating the circumstances under which a defendant’s intent to confuse or deceive consumers may be considered a factor in the likelihood of confusion inquiry. We realize that some courts have adopted a broader rule holding that a defendant’s intent to copy strongly supports an inference of likelihood of confusion. See, e.g., Bauer Lamp Co. v. Shaffer, 941 F.2d 1165, 1172 (11th Cir.1991); Mobil Oil Corp. v. Pegasus Petroleum Corp., 818 F.2d 254, 258 (2d Cir.1987). Like the intent-to-confuse rule, this intept-to-eopy rule relies essentially on a (re-buttable) presumption of efficacy — although the intent-to-copy rule requires a double inference, see, e.g., Perfect Fit Indus., Inc. v. Acme Quilting Co., 618 F.2d 950, 954 (2d Cir.1980) (“If there was intentional copying the second comer will be presumed to have intended to create a confusing similarity of appearance and will be presumed to have succeeded.”) — since the defendant’s intent standing alone (without reference to the defendant’s competence and the nature of the defendant’s actions) reveals little about the probable outcome of the defendant’s conduct. The justification for these inferences in a trademark or product packaging case is that there is little or no competitive need to copy another’s distinctive symbol or presentation to sell one’s product, and that anyone who does so is most likely trying to cash in on the competitor’s goodwill attached to the competitor’s mark or packaging in order to sell his or her own product. See, e.g., 2 MgCaRthy, TRADEMARKS AND UNFAIR COMPETITION § 23.33 (“[W]e can readily read into defendant’s choice of a confusingly similar mark the intent to get a free ride upon the reputation of a well known mark.”). This presumption largely duplicates the weight given to the substantial-identity-of-appearance factor (Scott factor 1) in the likelihood of confusion inquiry, and the extra weight assigned to the intent to deceive is somewhat punitive. Cf. 2 McCarthy, Trademarks and Unfair Competition § 23.32 (“Where there is hard evidence of defendant’s intention to get a free ride on plaintiffs reputation, the court is free to engage in the traditional rhetoric which accompanies punishing the evildoer[.]”). Although these two types of inference from defendant’s intent do not directly serve the purpose of preventing consumer confusion or misappropriation of a producer’s goodwill— either of which might arise from good faith or bad faith actions — the inferences may serve as a deterrent to infringement. But where product configurations are concerned, we believe there is little room for deterrence if appropriate labeling and marketing are undertaken. One primary purpose of the Lanham Act is to foster fair competition. See, e.g., Merchant & Evans, Inc., 963 F.2d at 640 n. 13 (citing Park ‘N Fly, Inc. v. Dollar Park and Fly, Inc., 469 U.S. 189, 193, 105 S.Ct. 658, 661, 83 L.Ed.2d 582 (1985), and Jay Dratler, Jr., Trademark Protection for Industrial Designs, 1988 U.Ill.L.Rev. 887, 926 n. 10). Indeed, we have said that prevention of unfair competition is the doctrinal basis for trade dress infringement suits under the Act. American Greetings Corp., Inc., 807 F.2d at 1140-41 & n. 2. Where product configurations are concerned, we must be especially wary of undermining competition. Competitors have broad rights to copy successful product designs when those designs are not protected by (utility or design) pat ents. It is not unfair competition for someone to trade off the good will of a product, see Kellogg Co. v. National Biscuit Co., 305 U.S. 111, 121, 59 S.Ct. 109, 115, 83 L.Ed. 73 (1938); it is only unfair to deceive consumers as to the origin of one’s goods and thereby trade off the good will of a prior producer. See also Duraco, 40 F.3d at 1445. Unless very narrowly tailored, deterrents to copying of product designs — as opposed to product packaging or trademarks — would inhibit even fair competition, thus distorting the Lanham Act’s purpose. We believe that the best way to further Congress’s intent is to limit carefully the scope of any possible deterrence of competition. Cf. Merchant & Evans, Inc., 963 F.2d at 640 (“[C]ourts should tailor trademark remedies to decrease the likelihood of confusion without unnecessarily inhibiting competition.”). Recognizing that trademark and trade dress cases on one hand and patent eases on the other do not involve identical considerations, we nevertheless turn for guidance in this task to patent cases concerning defendants’ intent, for, as we have noted, we must decide product configuration cases so as to harmonize them with the federal patent laws. See supra at 204-05. In patent infringement cases, a defendant’s bad intent is relevant in at least two contexts. First, under the doctrine of inequitable conduct, infringement claims may be rendered unenforceable if a plaintiff intended to deceive the Patent Office by failing to disclose material evidence. See, e.g., Braun, Inc. v. Dynamics Corp. of America, 975 F.2d 815, 822 (Fed.Cir.1992); Allen Organ Co. v. Kimball Int'l, Inc., 839 F.2d 1556, 1567-68 (Fed.Cir.1988). Second, a plaintiff may receive increased damages where a defendant willfully infringed its patent. See, e.g., Braun, Inc., 975 F.2d at 822; E.I. du Pont de Nemours & Co. v. Phillips Petroleum Co., 849 F.2d 1430, 1440 (Fed.Cir.1988). In either case, however, the plaintiff must prove by clear and convincing evidence that the defendant had the relevant bad intent. See, e.g., Braun, Inc., 975 F.2d at 822 (intent to deceive Patent Office and willful infringement); Allen Organ Co., 839 F.2d at 1567 (intent to deceive Patent Office); E.I. du Pont de Nemours & Co., 849 F.2d at 1440 (willful infringement). Although in the present context we are not dealing with increased damages or actions taken by the Patent Office, product configuration trade dress cases nonetheless implicate patent-like restrictions on competition. Like the doctrine of inequitable conduct, a heightened evidentiary standard would serve to ensure that deviations from the “the federal policy ... of allowing free access to copy whatever the federal patent and copyright laws leave in the public domain,” Compco Corp., 376 U.S. at 237, 84 S.Ct. at 782, are not casually countenanced. And much as the burden of proof for willful infringement assures that competitors are not penalized by increased damage awards without compelling evidence, we think it similarly important to competitors — as well as the public — that competition not be hobbled by monetary damages or injunctive prohibitions absent similarly compelling evidence. Accordingly, for all the foregoing reasons, we hold that in the product configuration context, a defendant’s intent weighs in favor of a finding of likelihood of confusion only if intent to confuse or deceive is demonstrated by clear and convincing evidence, and only where the product’s labeling and marketing are also affirmatively misleading. Of course, a plaintiff might succeed in proving likelihood of confusion without evidence of affirmative deception. We only hold that, to be considered as evidence of a likelihood of confusion in a product configuration ease, the defendant’s intent must meet the conditions we have set forth. 6. Marketing Considerations (Scott Factors 7-10) The remaining factors identified by Scott as bearing on the likelihood of confusion address various aspects of the marketing of the products. In the product configuration context, none of these four factors tends to establish a probability of confusion, rather than a mere possibility, and thus we conclude that they should be treated as necessary but insufficient conditions for showing a likelihood of confusion. The seventh Scott factor is “whether the goods, though not competing, are marketed through the same channels of trade and advertised through the same media.” We believe that this factor, which is explicitly formulated for application to non-competing products, serves primarily to establish the possibility of confusion and carries little weight toward establishing the probability of confusion; if not shown, it may exonerate a defendant, but if established, it merely allows the plaintiffs case to go forward. Moreover, it will rarely need to be considered in a product configuration trade dress infringement case, for the goods at issue will, almost by definition be in competition. “The extent to which the targets of the parties’ sale efforts are the same” is the eighth Scott factor. Like the marketing channel inquiry, this factor was developed largely for non-competing products, see, e.g., Interpace Corp. v. Lapp, Inc., 721 F.2d 460, 462-63 (3d Cir.1983); Scott Paper Co., 589 F.2d at 1229-30, and relates more to the possibility than the probability of confusion. Part