dissenting.
After a twelve-day trial during which both sides presented evidence about the teachings of the prior art, the jury made detailed factual findings pertaining to the obviousness of the '420 and '664 patents. The jury found, among other findings, that elements of the asserted claims were not present in the prior art. Based on the jury’s findings, the district court determined that the Defendants had failed to prove by clear and convincing evidence that the asserted claims were obvious. In reversing the district court’s judgment, the majority finds that the prior art discloses a key claim limitation that the jury found was missing, and also concludes that the district court erred in failing to use “common sense” to bridge the differences between the prior art and the claims. In my view, the majority fails to accord sufficient deference to the jury’s findings of fact. Moreover, I find that the majority’s use of common sense to bridge the gap between the prior art and the claims is unsupported by sufficient evidence and reasoning. I respectfully dissent.
At the outset, it is worth noting that obviousness is a mixed question of law and fact. Kinetic Concepts, Inc. v. Smith & Nephew, Inc., 688 F.3d 1342, 1356 (Fed.Cir.2012). Although we must examine the legal conclusion of obviousness de novo, we should tread lightly when reviewing a legal conclusion — reached by a trial court — that rests upon a jury’s findings of fact. See Polaroid Corp. v. Eastman Kodak Co., 789 F.2d 1556, 1558 (Fed.Cir.1986) (“Though it is well settled that the ultimate conclusion on obviousness is ... a legal conclusion, that does not mean ... that we may proceed on a paper record as though no trial had taken place. This court reviews judgments. Because we do not retry the case, [the appellant] must to prevail convince us that the judgment cannot stand on the record created at trial....”); cf. Haebe v. DOJ, 288 F.3d 1288, 1299 (Fed.Cir.2002) (“[G]reat deference must be granted to the *997trier of fact who has had the opportunity to observe the demeanor of the witnesses, whereas the reviewing body looks only at cold records.”) (internal quotation marks omitted). Where a jury’s findings concerning the prior art are supported by substantial evidence, and where a trial court makes its obviousness determination based on those findings, I would exercise caution in wielding our own common sense as part of our review of the judgment.
The asserted claims in this case are based, in part, on filtering techniques used in two types of information systems found in the prior art: content-based systems and profile-based (or “collaborative”) systems. Content-based systems filtered search results for relevance to a user’s query, as reflected in “conventional search engines.” See Majority Op. at 987 (quoting '420 patent col.2 11.15-18). Collaborative systems, meanwhile, filtered information for relevance based on “the user’s long-term information desires or preferences,” and incorporated the information preferences of similar users. J.A. 3690.
As the majority explains, the prior art suggested that content-based and collaborative filtering could be combined. See Majority Op. at 986-87. The majority identifies prior art systems that passed content-based results (which were returned based on the user’s query) over to a distinct collaborative filter. The query in these systems was used only to obtain the initial results; it played no role in subsequent filtering on the collaborative side. The asserted claims, however, require using the query itself — not just the results returned by the query — on the collaborative side, thus combining content-based and collaborative filtering. See '420 patent, col.28 11.1-15; '664 patent, col.27 11.27— 37.
At trial, experts for both sides testified about whether a person of skill in the art would have found it obvious to supply the key claim limitation missing from the prior art — the use of the query as part of a combined content-based and collaborative filter. I/P Engine’s expert testified that a person of skill in the art in 1998 would not have “appreciated the advantages of tight integration” of search systems and profile systems, particularly with regard to the “relevance to the query.” J.A. 3739. In response, the Defendants’ expert testified that the prior art did “feature a tight integration between the search system and the content collaborative system ... [because ... it would have been obvious to one of ordinary skill in the art that if you are filtering search results, it’s obvious to keep around the query and use that also for filtering.” Id. at 3172-73.
With respect to both patents, the jury found that the prior art “did not disclose a tightly integrated search system, and could not filter information relevant to the query.” J.A. 4170, 4172. The majority downplays the significance of the jury’s findings, explaining that the jury “never determined whether it would have been obvious to one skilled in the art to bridge any differences between the prior art and the claimed invention.” Majority Op. at 990-91. Without such a determination, the majority suggests, we must resort to “common sense” to address the question left unanswered by the jury-that is, whether it would have been obvious to one of skill in the art to use the search query as part of the filtering of collaborative data.1
*998We have explained that “the mere recitation of the words ‘common sense’ without any support adds nothing to the obviousness equation. Thus, we have required that obviousness findings grounded in ‘common sense’ must contain explicit and clear reasoning providing some rational underpinning why common sense compels a finding of obviousness.” Plantronics, Inc. v. Aliph, Inc., 724 F.3d 1343, 1354 (Fed.Cir.2013) (internal citations omitted). As the Supreme Court emphasized, “it can be important to identify a reason that would have prompted a person of ordinary skill in the relevant field to combine the elements in the way the claimed new invention does ... because inventions in most, if not all, instances rely upon building blocks long since uncovered, and claimed discoveries almost of necessity will be combinations of what, in some sense, is already known.” KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 418, 127 S.Ct. 1727, 167 L.Ed.2d 705 (2007). We may not find a patent invalid for obviousness on the basis of “mere conclusory statements.” In re Kahn, 441 F.3d 977, 988 (Fed.Cir.2006).
Here, in support of its finding of a “strong prima facie case of obviousness,” the majority concludes that “retaining the query for use in filtering combined content and collaborative data was entirely predictable and grounded in common sense.” Majority Op. at 989, 992 (internal quotations omitted).2 The use of the query is a matter of common sense, the majority explains, “[bjecause the query was readily available and closely correlated to the overall relevance of search results.” Id. at 989. In support of its suggestion that one of skill in the art would find it obvious to use the readily available query, the majority cites the testimony of the Defendants’ expert:
As [the Defendants’ expert] explained, the query would be just “sitting there” with the results of a search, and it would have been obvious to one skilled in the art to “keep around the query and use that also for filtering.”
Id. (citing J.A. 3173).
I find this testimony inadequate to support the majority’s conclusion. The expert’s “sitting there” explanation tells us nothing about whether one of skill in the art in 1998 would have been struck by common sense to modify collaborative filtering systems so as to incorporate search queries. All prior art references are “just sitting there” in the metaphorical sense. What is needed — and what is missing from the cited testimony — is some explanation of why one would use the query as the asserted claims do. See Innogenetics, *999N.V. v. Abbott Labs., 512 F.3d 1363, 1374 (Fed.Cir.2008) (“[S]ome kind of motivation must be shown from some source, so that the jury can understand why a person of ordinary skill would have thought of either combining two or more references or modifying one to achieve the patented method.” (internal citations omitted)). Such evidence or reasoning is lacking here. The testimony of the Defendants’ expert amounts to a “mere conelusory statement” that may not serve as a basis for finding the asserted claims obvious. See In re Kahn, 441 F.3d at 988; see also InTouch Techs., Inc. v. VGO Commc’ns, Inc., 751 F.3d 1327, 1352 (Fed.Cir.2014) (holding that expert’s testimony could not support a finding of obviousness where “testimony primarily consisted of conelusory references to [the expert’s] belief that one of ordinary skill in the art could combine these references, not that they would have been motivated to do so”).
As for the majority’s observation that the query is “closely correlated to the overall relevance of search results,” no one disputes that the prior art taught that a query was relevant to a user’s content-based search. What is disputed is whether the prior art taught the query’s “overall relevance” to collaborative filtering. See J.A. 3172-73, 3739. To bridge the gap, we must identify a non-circular reason that would have prompted a person of skill in the art to appreciate the relevance of the query to collaborative data. See KSR, 550 U.S. at 418, 127 S.Ct. 1727.
The gap in the prior art references here is unlike gaps we have encountered in other cases — cited by the majority — where we have found patents obvious for merely adding “the Internet” or “a web browser” to a well-known prior art reference. See Majority Op. at 989-90 (citing Soverain Software LLC v. Newegg Inc., 705 F.3d 1333, 1344, amended on reh’g (Fed.Cir.2013); Muniauction, Inc. v. Thomson Corp., 532 F.3d 1318 (Fed.Cir.2008)). In those cases, we observed that skilled artisans had already performed the same type of combination with similar elements. See Soverain, 705 F.3d at 1344 (finding that the use of use of hypertext to perform the known process of transmitting documents “was a routine incorporation of Internet technology into existing processes”); Muniauction, 532 F.3d at 1326-27 (concluding that “[t]he record in this case demonstrates that adapting existing electronic processes to incorporate modern internet and web browser technology was similarly commonplace at the time the '099 patent application was filed”).
In this case, the majority does not identify analogous “routine” combinations that would render obvious the patents’ incorporation of the search query into a collaborative filtering system. The asserted patents did not merely combine information technology with “the Internet.” Rather, the patents combined elements from two known information filtering systems. The patents took the query data (input for a content system) and mixed it with collaborative data (input for a profile-based system). What was claimed was the combination of elements of two evolving systems in the field of information science, not a combination of a known process and a web browser.
Nor is this case like Perfect Web Technologies, Inc. v. InfoUSA Inc., 587 F.3d 1324 (Fed.Cir.2009), where we affirmed a district court’s conclusion that a patent was obvious as a matter of common sense. In Perfect Web, the asserted patent claimed a four-step method for distributing bulk emails. It was undisputed that the first three steps of the method were disclosed in the prior art. Id. at 1330. The fourth step — which was not present in the prior art — recited repeating the first three steps over and over until all the emails were delivered. As we described the *1000method, it “simply recites repetition of a known procedure until success is achieved.” Id. Moreover, we found that the “relevant art required only a high school education and limited marketing and computer experience,” and that no expert opinion was required to appreciate the value of repeating the three steps. Id.
In affirming the district court’s conclusion that the patented method was obvious, we observed that “simple logic suggests that sending messages to new addresses is more likely to produce successful deliveries than resending messages to addresses that have already failed.” Id. at 1331. To put it another way, one of skill in the art would have been motivated to add the fourth step because it would increase success — more recipients would receive email messages. We identified a benefit that would have been readily apparent to one of skill in the art at the time of the invention.
Here, by contrast, the record does not suggest a benefit or rationale that would have caused a skilled artisan to use the query as part of collaborative filtering in 1998. Although we know that the query was “sitting there,” we do not know why one of skill in the art would have thought that mixing the query with the collaborative filter would produce, to use the language of Perfect Web, “successful” filtering. We need something beyond the invocation of the phrase “common sense” or “simple logic” to demonstrate the reason to combine the prior art references in this case. See KSR, 550 U.S. at 418, 127 S.Ct. 1727; Innogenetics, 512 F.3d at 1374.
For these reasons, I respectfully dissent from the majority’s holding that the asserted claims of the '420 and '664 patents are invalid for obviousness.
. The majority also concludes that this claim limitation is taught by the Culliss patent because that reference “squarely discloses using combined content and collaborative data in analyzing items for relevance to a query.” Majority Op. at 987-88. But the majority’s conclusion squarely conflicts with the jury’s express finding that Culliss ”lack[s] any con*998tent analysis and filtering for relevance to the queiy.” J.A. 4170. Based on the record, I would defer to the jury’s fact finding. In the face of conflicting testimony about what Cul-liss disclosed, the jury was free to credit the opinion of I/P Engine’s expert. See Power-One, Inc. v. Artesyn Technologies, Inc., 599 F.3d 1343, 1351 (Fed.Cir.2010) (explaining that the jury was free to either credit or disbelieve expert testimony about "the differences between the prior art and the invention claimed”).
. The majority also takes issue with the jury’s findings on secondary considerations, noting that two of the findings with respect to the '420 patent and the '664 patent "appear internally inconsistent.” Majority Op. at 990. The majority generally characterizes the findings on secondary considerations as a “mixed bag.” Id. Even assuming, however, that the jury’s secondary consideration findings are as muddled as the majority describes, they are relevant only insofar as the majority is correct that its invocation of "common sense” may support a prima facie case of obviousness that must be overcome. See Dow Chemical Co. v. Halliburton Oil Well Cementing Co., 324 U.S. 320, 330, 65 S.Ct. 647, 89 L.Ed. 973 (1945) ("[Secondary] considerations are relevant only in a close case where all other proof leaves the question of invention in doubt.”).