concurring.
Although I agree with the outcome and rationale and thus join the opinion, I write separately to express my concern with the absence of analysis and resulting potential for uncorrectable legal error in the otherwise well-written Board opinion on the underlying question in this case: why the dissimilarity in the commercial impressions conveyed by the competing marks outweighs the Board’s findings in favor of Roederer with respect to all other relevant DuPont factors, including the strength of Roederer’s “very strong” senior mark and the similarity of Delicato’s and Roederer’s goods, customers, and channels of trade, between which the Board could “draw no distinctions.” Not only do these factors *1376support the opposer but they diminish what he must prove in order to prevail.
I believe that, in general, the Board’s expertise warrants great respect for its legal conclusion on likelihood of confusion even though clear error deference is limited to its underlying fact-findings. However, in the rare instances such as this in which the Board fails to provide adequate explanation of the reasoning on which it relied to reach its result, especially in the face of sharply conflicting findings, I question whether such respect is justified. To date, our court has said little on the subject of critical omissions in such opinions. The United States Court of Appeals for the District of Columbia Circuit, however, has made frequent reference to similar problems of analytical opacity in opinions in the context of the judicial review it is mandated to conduct of decisions by certain regulatory agencies. Most recently, for example, in American Lung Association v. Environmental Protection Agency, 134 F.3d 388 (D.C.Cir.1998), the court was asked to review an agency decision declining to strengthen existing air quality standards for sulfur dioxide. The court remanded the decision to the agency “to permit the Administrator to explain her conclusions more fully.” Id. at 393. In reaching this outcome, the court first noted that “[jludicial deference to decisions of administrative agencies ... rests on the fundamental premise that agencies engage in reasoned decision-making.” Id. at 392 (emphasis added). It went on to explain, however, that “[w]ith its delicate balance of thorough record scrutiny and deference to agency expertise, judicial review can occztr only when agencies explain their decisions with precision, for ‘[i]t will not do for a court to be compelled to guess at the theory underlying the agency’s action____’” Id. (quoting SEC v. Chenery Corp., 332 U.S. 194, 196-96, 67 S.Ct. 1575, 91 L.Ed. 1995 (1947)) (emphasis added).
Like regulatory agencies or other executive tribunals in their subject areas, the Trademark Trial and Appeal Board has acquired a high level of expertise in evaluating the DuPont factors and counter-weighing these factors to reach its ultimate conclusion — the likelihood vel non of confusion between competing marks. Nonetheless, the Board too should explain with reasonable “precision” not only its factual findings but the “theory underlying” its final conclusion. We need to be told and not be “compelled to guess at the theory” the Board applied to compare its conflicting findings and decide likelihood of confusion. Without such explanation wé are hard-pressed to review independently, as we must, whether the Board’s ultimate conclusion is legally correct. In such circumstances, we cannot assure ourselves that the Board did indeed “engage in reasoned decision-making.”
To be sure, every opinion need not be perfectly clear or detailed. Indeed, there are many oppositions in which cursory treatment of the law and relevant facts is adequate. For example, in some cases the Board may only receive evidence relevant to a single DuPont factor. In such a circumstance, the finding under that one factor would clearly be dispositive, and little farther discussion would be necessary. On the other end of the spectrum, some cases may involve extensive evidentiary presentations on many or all of the DuPont factors, but the Board might find all of the factors in favor of one of the parties in an opposition proceeding. Such cases would similarly require little explanation as to how the Board reached its decision on likelihood of confusion. Like the District of Columbia Circuit, then, we should “uphold an' agency decision of less than ideal clarity if the agency’s path may reasonably be discerned.” Common Cause v. Federal Election Comm’n, 906 F.2d 705, 706 (D.C.Cir. 1990) (emphasis added) (internal quotations and citation omitted). I believe, however, we should guard against affirming a Board decision that provides little insight into its “underlying theory” or “path,” based merely on an assumption that, as an expert tribunal, it must have decided correctly.
Although we have not addressed this problem of opaque or incomplete opinions in the context of the Trademark Trial and Appeal Board, we have previously considered the difficulty they pose for appellate review of decisions of the Board of Patent Appeals and Interferences. In that context, we have explicitly held that, although we review deci*1377sions and not opinions, to warrant affirmance of a decision on obviousness, the opinion “must contain sufficient findings and reasoning to pemiit meaningful appellate scrutiny.” Gechter v. Davidson, 116 F.3d 1454, 1458, 1460, 43 U.S.P.Q.2d 1030, 1033, 1035 (Fed.Cir.1997) (vacating decision and remanding for “specific findings of fact and conclusions of law adequate to form a basis” for appellate review) (emphasis added); cf. Oakley v. International Tropic-Cal, Inc., 923 F.2d 167, 168, 17 U.S.P.Q.2d 1401, 1403 (Fed. Cir.1991) (vacating a preliminary injunction issued by the district court because its factual findings on the four Hybñtech factors supporting the injunction were “so limited and conclusory that meaningful appellate re-vieiv is not possible” (emphasis added)). I believe that similarly here, once it set forth its findings of fact with respect to each of the five DuPont factors it considered, the Board was further obligated to provide some brief analysis regarding the relatively little weight it chose to afford the four pro-confusion factors in determining, ultimately, no likelihood of confusion.
Other circuit courts have held expert tribunals they review to a similar obligation. Recently, for example, the United States Court of Appeals for the Ninth Circuit remanded for further explanation a United States Tax Court decision limiting a tax deduction based on an executive’s reasonable compensation. See Leonard Pipeline Contractors, Ltd. v. Commissioner of Internal Revenue, 142 F.3d 1133 (9th Cir.1998). In a section aptly labeled “The Opaque Opinion of the Tax Court,” the Ninth Circuit provided the following justification for the remand:
Despite the difficulties of determining what is reasonable compensation, it is the obligation of the Tax Court to spell out its reasoning and to do more than enumerate the factors and leap to a figure intermediate between petitioner’s and the Commissioner’s. We can see the pieces of the puzzle. We can only guess how the trial judge put them together. It is not impossible for us to review such a decision, but we could do so only by supplying a synthesis that it is the Tax Court’s duty to provide. A reasoned decision as to what is reasonable in this context must bring together the disparate elements and give some account of hcnv the judge has reached his conclusion____
It is not our task to put the factors together in the first instance....
Id. at 1135-36 (emphasis added); cf. Public Media Ctr. v. FCC, 587 F.2d 1322, 1331 (D.C.Cir.1978) (“Although our judicial duties demand great deference to agency expertise, we cannot defer, indeed we cannot even engage in meaningful review, unless we are told which factual distinctions separate arguably similarly situated licensees, and why those distinctions are important.”). I see no reason why the Board- should not be similarly obligated to “supply a synthesis” or “give some account of how [it] has reached” its conclusion so that we may engage in “meaningful appellate review.” If the Board simply enumerates its findings on the various factors, pro and con, and then “leaps” to its ultimate conclusion on likelihood of confusion, we are greatly impeded in our review function.
Because the Board in this case failed to “supply a synthesis” of its conflicting findings that would enable us to “discern the path” to its ultimate conclusion, I initially wondered whether its conclusion of no likelihood of confusion in this case was correct. As our per curiam opinion notes, however, it is perfectly lawful for the Board to determine in an appropriate case that one DuPont factor outweighs all others and thus disposes of the question of whether competing marks are confusingly similar. It was not immediately clear to me from the Board’s opinion, however, that this is such a case. See Specialty Brands v. Coffee Bean Distributors, Inc., 748 F.2d 669, 671, 223 U.S.P.Q. 1281, 1282 (Fed. Cir.1984) (holding that, in some instances, the appearance, sound, and significance of the marks may be dispositive, “but the similarity between words in the respective marks is only part of the inquiry into likelihood of confusion”). Thus, the difficulty here arises, in my view, because the Board explicitly found that four of the DuPont factors weighed in Roederer’s favor, but concluded nonetheless that the dissimilarities of the marks in appearance, sound, significance, and commercial impression weighed dispositively *1378in favor of Delicato. I for one am left wondering how the Board justified its disregard or diminution in weight of the four DuPont factors that it found supported Roederer’s Opposition, for the Board has failed to “bring together the disparate elements” to explain the basis for its conclusion of no likelihood of confusion.
For example, the Board explicitly found that the “CRISTAL” mark “has come to serve as a very strong indication of origin” for Roederer’s champagne. See Champagne Louis Roederer, slip op. at 9 (emphasis added). If Roederer’s mark is so strong, however, why was the mere addition of the generic, topographical word “creek” in Delieato’s application mark sufficient to overcome any potential for confusion? The Board provides no explanation. It may be that the Board determined this DuPont factor favoring Roe-derer was not entitled to any weight at all in this case, but there is no explanation as to why this should be. Furthermore, the Board’s apparent failure to consider the strength of Roederer’s mark conflicts with our case law, for we have held that “[wjhen an opposer’s trademark is a strong, famous mark, it can never be of little consequence. The fame of a trademark may affect the likelihood purchasers will be confused inasmuch as less care may be taken in purchasing a product under a famous name.” Specialty Brands, 748 F.2d at 675, 223 U.S.P.Q. at 1284 (internal quotations omitted); see also J & J Snack Foods Corp. v. McDonald’s Corp., 932 F.2d 1460, 1464, 18 U.S.P.Q.2d 1889, 1892 (Fed.Cir.1991) (“[Pjreference is accorded the prior user of a mark or family of marks, as against a newcomer. The newcomer has the clear opportunity, if not the obligation, to avoid confusion with well-known marks of others.”). It is not clear to me, therefore, whether the Board’s failure expressly to attribute any significance to the strength of Roederer’s mark shows it applied an improper standard and, if so, whether that was merely harmless error here.
Similarly, because of the absence of explicit limitations in the application, together with the failure by both parties to present other evidence, the Board explicitly found no basis upon which to “distinguish” the goods “as to price, channels of trade, or classes of purchasers.” Champagne Louis Roederer, slip op. at 12. Nonetheless, there is no express indication that the Board even considered these three findings in reaching its ultimate conclusion that the marks are not confusingly similar or recognized that as a matter of law they change the analysis. Again, this failure may be especially significant in light of previous Board decisions holding that the degree of similarity between competing marks, where the goods on which the marks are used are directly competitive, need not be as great before registration should be denied. See, e.g., Eci Div. of E-Sys., Inc. v. Environmental Communications Inc., 207 U.S.P.Q. 443, 449, 1980 WL 30140 (TTAB 1980) (“It is necessary to note at the outset that, where as here, the goods of the parties are similar in kind and/or closely related and travel in the same trade channels where they can be encountered by the same purchasers, the degree of similarity of the marks under which these products are sold need not be as great as in the case of diverse or different goods.” (emphasis added)). If the Board did in fact fail to account for the similarity of the goods, trade channels, and customers in its analysis of the confusion issue, it may have consequently held Roederer to an inappropriately higher burden of proof than appropriate in such an opposition proceeding. Again, if it did err at law, how in the face of silence can we determine whether such error was harmless?
In short, there is no explanation of how one DuPont factor in Delieato’s favor can be considered adequate to support the legal conclusion of no likelihood of confusion where four other DuPont factors support Roederer’s assertion that the competing marks are so confusingly similar that Delieato’s application for registration should be denied. Furthermore, the Board’s failure to explain raises for me the question whether the Board failed to give the requisite legal significance to the factors it did consider. Thus, as this ease illustrates, it is not always enough that the Board make findings with respect to each DuPont factor that is of record; some discussion of the relative weight to be accorded at least to conflicting factors is necessary, or the findings alone may be insufficient for *1379meaningful appellate review. In my view, the Board here failed to “explain its decision with precision,” or even provide minimal landmarks such that its “path may reasonably be discerned” to those of us charged with reviewing its decision.
I am further troubled by some of the Board’s particular factual -findings with respect to the dissimilarity of the competing marks. For example, the Board, without any discussion at all, found that the marks were different in sound and appearance. See Champagne Louis Roederer, slip op. at 13 (“Moreover, there are differences between the marks in sound and appearance.”). It failed, however, to identify the differences or explain why they were important. Because Delicato’s application was not part of the record on appeal, we cannot evaluate the similarity of appearance of the marks for ourselves, and the Board presents no discussion of the visual differences it presumably found. Furthermore, in the context of word marks, I question how important the “sight” factor can be for the overall comparison of commercial impressions. Even more troubling, although the Board concluded that “there are differences between the marks in sound,” id. at 13, it also found, inconsistently, that Roederer’s “CRISTAL” mark would be interpreted by most consumers as a phonetic equivalent, i.e., a misspelling, of the word “crystal,” see id. at 12, and that “a substantial segment of the purchasing public for goods of the type involved here would pronounce CRISTAL and CRYSTAL in a similar manner,” id. at 11. This suggests to me at least that the most significant component of the two word marks in fact are quite similar in sound. These seemingly contradictory findings, however, are never explained, must less reconciled, in the Board’s opinion.1
Because, to all outward appearances, the Board here failed to give any,weight to the similarity of the goods, customers, and trade channels or the strength of Roederer’s senior marks, and instead relied solely on its highly subjective finding that the commercial impression of the marks were dissimilar to determine no likelihood of confusion, I was at first -unsure of the correct resolution of this case". Thus, I hesitated to affirm the Board’s decision to dismiss the Opposition because of the crucial omissions in the analytical framework in its-opinion.
Nonetheless, despite my concerns as outlined above, I join the opinion affirming the Board' because, as the opposer, Roederer bore the burden of proof before the Board. See Sanyo Watch Co. v. Sanyo Elec. Co., 691 F.2d 1019, 1022, 215 U.S.P.Q. 833, 834 (Fed. Cir.1982) (holding that the opposer in a proceeding to test likelihood of confusion “bears the burden of proof which encompasses not only the ultimate burden of persuasion, but also the obligation of going forward with sufficient proof of the material allegations of the Notice of Opposition”); see also Yamaha Int’l Corp. v. Hoshino Gakki Co., 840 F.2d 1572, 1579-80, 6 U.S.P.Q.2d 1001, 1007 (Fed. Cir.1988) (noting that because section 2(d) of the Lanham Act “provides that no trademark shall be refused unless it is shown that there is a likelihood of confusion with another mark, the requirement that the opposer both establish a prima facie case of likelihood of confusion and carry the ultimate burden of persuasion on that issue is proper”); Clinton Detergent Co. v. Proctor & Gamble Co., 49 *1380C.C.P.A. 1146, 302 F.2d 745, 747, 133 U.S.P.Q. 520, 522 (CCPA 1962) (“Opposer, however, has the burden of proof to establish that applicant does not have the right to register its mark.”). The evidence Roederer presented to meet its burden before the Board was sparse and its persuasiveness questionable.
Moreover, decisions below are presumed correct. Thus, Roederer had the burden of persuasion on appeal as well. Yet Roederer has failed to persuade me that it met its burden at the Board level, or that the Board committed any legal, fact-finding, or other error that would warrant reversal of its ultimate conclusion of no likelihood of confusion. Thus, although the Board’s opinion in this case presents something of a “black box,” impervious to searching review on appeal, I agree to affirm. I do hope, however, that future Board opinions will not suffer from critical omissions such as the ones I have tried to identify here, but will instead focus on reconciling conflicting findings while perhaps omitting needless discussion. The result should be opinions that are shorter and quicker to write, yet will better facilitate meaningful appellate review.
. Although some may believe I am simply criticizing the Board’s well-written thirteen-page opinion here as too short, I do not suggest that a longer opinion or one with a more extensive factual assessment was necessary to correct the omissions I see. To the contrary, I believe the Board’s actual findings were sufficiently detailed, especially in light of the paltry evidence provided by the parties. Rather, I suggest the Board’s opinion — and thus, our review function — would have benefited from at least some brief, analytical explanation for why several DuPont factors, properly of record and weighing in the opposer’s favor, should be disregarded in favor of one factor weighing in the opposite direction, It is precisely that intermediate analytical step, however, that a panel member on our court must evaluate most closely on appeal, for on the ultimate conclusion our review is non-deferential. Thus, a brief additional sentence or two would have greatly improved our ability to review, and thus our confidence in, the Board’s ultimate decision here. Indeed, its opinion could actually have been shorter and yet still included the crucial analytical discussion. Nor need a better focused opinion have required more time or effort to write. Thus, there should be no concern for increasing delays or enlarging backlogs at the Board.