Resources on the Use of Conjoint Analysis in Patent Cases

Back in 2013 I published a short post about an article by S. Christian Platt & Bob Chen titled Recent Trends and Approaches in Calculating Patent Damages: Nash Bargaining Solution and Conjoint Surveys, Bloomberg BNA Pat., Trademark & Copyright L. Daily (Aug. 30, 2013), in which the authors discussed the use in IP litigation of conjoint analysis--a technique originally developed by marketing researchers to "measur[e] consumer preferences for specific product features" by "break[ing] a product down into bundles of attributes, and then test[ing] various combinations . . . to determine customer preferences."  In theory, conjoint surveys could help determine the relative importance to consumers of (say) a specific patented feature embodied in a complex product, and thus might be useful in calculating damages (based on the premise that a willing licensee would be willing to pay as much as, but not more than, the value the feature promises to deliver in comparison with other alternatives).  As Chen & Platt noted, as of 2013 there were only a small number of cases in which litigants had tried to employ conjoint analysis to this end, and the results were mixed.  (Judge Alsup rejected the use of conjoint analysis in the Oracle v. Google copyright case, for example.)  The authors concluding by stating that for conjoint analysis to be accepted in court, an expert "must be able to provide a principled basis" for why he or she focused on certain product features and not others. 

To my knowledge, conjoint surveys still are not exactly a staple of U.S. IP litigation, but a quick Westlaw search discloses that since 2013 there have been several more cases in which litigants have used or attempted to use conjoint analysis, though the results are still mixed.  The only time the Federal Circuit has addressed the matter in a patent case was in Apple Inc. v. Samsung Elecs. Co., 735 F.3d 1352, 1367-68 (Fed. Cir. 2013), in which the court vacated and remanded an order denying Apple a permanent injunction, on the ground that Judge Koh erred in considering the conjoint survey irrelevant to the question of whether there was a causal nexus between the infringement and Apple’s alleged harm.  Nonetheless, on remand Judge Koh concluded that the proffered conjoint analysis did not prove causal nexus, because (1) it "does not provide a way to directly compare consumers' willingness to pay for particular features to the overall value of the infringing devices"; (2) Dr. Hauser's results showed that "substantial portions" of the "price premiums" consumers were willing to pay were "attributable to features other than the patented features"; (3) "the survey appears to have failed to adequately account for noninfringing alternatives to the patented features"; and (4) the survey appeared to give "undue emphasis" to the patented features.   Apple Inc. v. Samsung Elecs. Co., Case No.: 11–CV–01846–LHK, 2014 WL 976898 (N.D. Cal. Mar. 6, 2014).   

Anyway, I thought I would mention here a few more resources on conjoint analysis and its cousin, discrete choice analysis, that readers might find useful.  I think it's likely that we'll see more attempts to use these techniques in IP litigation, both in complex products litigation and (as Pam Samuelson suggested at a recent conference) in U.S. design patent cases (where, post Samsung v. Apple, the court now has to figure out the profit attributable to the relevant design patent infringing component).

First, there's an article I mentioned in a post back in 2015, Greg M. Allenby, Jeff Brazell, John R. Howell & Peter E. Rossi, Valuation of Patented Product Features, 57 J. L. & Econ. 629 (2014).  Here is the abstract:

Ultimately, patents have value to the extent to which the product features enabled by the patents have economic value in the marketplace. Products that are enhanced by inclusion of patented features should generate incremental profits. Incremental profits can be assessed by considering demand for products with patented features and contrasting that demand with demand for the same product without the patented features. Profit calculations must be based on valid estimates of demand as well as assumptions about how competitive forces affect demand via computation of market equilibria. A conjoint survey can be used to estimate demand. Recently, conjoint methods have been applied in the patent setting, but the measures of value used are purely demand based and do not involve equilibrium profit calculations. We illustrate our method using the market for digital cameras and show that current methods can overstate the value of a patent.

Second, J. Gregory Sidak and Jeremy O. Skog recently published an article titled Using Conjoint Analysis to Apportion Patent Damages, 25 Fed. Cir. B.J. 581 (2016).  Here is the abstract:

Expert economic witnesses increasingly present survey evidence to support their calculations of reasonable-royalty damages in patent-infringement cases. In this article, we assess developments in the case law on the admissibility of expert testimony based on conjoint analysis, which is a particular type of survey analysis commonly used in market research to measure the tradeoffs that consumers make among salient features of a product. In a growing number of cases, experts have used conjoint analysis to estimate consumers’ average willingness to pay for a patented technology. That estimate informs the implementer’s maximum willingness to pay to license the patented technology, which is the upper bound on the bargaining range for a reasonable royalty in a hypothetical negotiation. We identify and analyze the factors that courts have considered when determining whether evidence from an expert’s conjoint survey is admissible. Further, one can use conjoint analysis to argue whether an infringing feature in a multicomponent product drives the demand for that product, which is relevant to the legal test in the United States for determining whether the patent holder may obtain an injunction and for identifying the appropriate royalty base for calculating damages. Decisions by the Federal Circuit and district courts in several disputes between Apple and Samsung indicate that the Federal Circuit has not yet provided comprehensive guidance on how to determine whether a patented feature drives demand for a downstream product for purposes of deciding whether a patent holder may obtain an injunction. Other Federal Circuit decisions indicate that, for purposes of identifying the appropriate royalty base, only evidence that a patented feature motivates consumers to purchase the product at issue will suffice to show that the patented feature drives demand for that downstream product.

Third, Professor Samuelson directed me to an article by Rohit Verma, Gerhard Plaschka, and Jordan J. Louviere titled Understanding Customer Choices: A Key to Successful Management of Hospitality Services, Cornell Hotel & Restaurant Admin. Q. 15 (2002), which (though not directly relevant to the topic of IP litigation) provides an accessible discussion of discrete choice analysis.  Here is the abstract:

We know that hospitality customers usually make purchases by simultaneously evaluating several criteria. A typical buying decision might take into account service quality, delivery speed, price, and any special buying incentives, for instance. It is imperative that businesses take into account customer preferences and choices when making decisions regarding product and service attributes. Managers need to understand how customers integrate, value, and trade off different product and service attributes. By the same token, information about customer demands and preferences must be incorporated into the design and day-to-day management of service-delivery processes.

In this paper we describe a particularly effective way to determine those customer preferences and to assess the tradeoffs that customers make in considering various product and service bundles. The methodology we describe is discrete-choice analysis (DCA). After explaining DCA, we provide guidelines for incorporating customer-preference information into the design and management of business processes. The DCA approach provides a robust and systematic way to identify the implied relative weights and attribute trade-offs revealed by decision makers' choices (whether customers or managers).

Fourth, Romain de Nijs has published an article titled Sondages et évaluation des préjudices économiques en matière de propriété intellectuelle ("Surveys and the evaluation of economic harm in regard to intellectual property") in the February 2017 issue of Propriété Industrielle (pp. 18-21). Here is the abstract (my translation from the French):

This article presents a panorama of the utilization of surveys for characterizing and quantifying economic harm in regard to intellectual property (trademark infringement, patent infringement, and online piracy of protected content).

Page 20 discusses the use of surveys in patent cases in particular, and cites some of the economic literature on conjoint analysis (as well as some articles on the use of conjoint analysis in antitrust law).  The article also alerted me to two more articles discussing the possible use of conjoint analysis in patent infringement cases:  a 2013 AIPLA White Paper by Joel Steckel, Rene Befurt & Rebecca Kirk Fair titled Is It Worth Anything? Using Surveys in Intellectual Property Cases, and a 2011 paper by Christopher K. Larus & Bryan J. Mechell titled Using Consumer Surveys to Prove Patent Infringement Damages at Trial, from the December 2011 issue of The Intellectual Property Strategist.