DCT
2:19-cv-00496
Applied Predictive Tech Inc v. Marketdial Inc
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: Applied Predictive Technologies, Inc. (Delaware)
- Defendant: MarketDial, Inc. (Delaware) and John M. Stoddard
- Plaintiff’s Counsel: Morris, Nichols, Arsht & Tunnell LLP
- Case Identification: 1:18-cv-00963, D. Del., 09/17/2018
- Venue Allegations: Venue is alleged to be proper in the District of Delaware because Defendant MarketDial, Inc. is a Delaware corporation and therefore resides in the district.
- Core Dispute: Plaintiff alleges that Defendant’s business analytics software infringes a patent related to optimizing parameters for business initiative testing models, and further alleges misappropriation of trade secrets by a former consultant.
- Technical Context: The technology operates in the "big data" business analytics sector, providing software tools for enterprises to test and predict the financial impact of business initiatives like promotions, store remodels, or pricing changes.
- Key Procedural History: The complaint alleges that Plaintiff provided Defendant with actual notice of the patent-in-suit via a letter dated December 18, 2017, which Defendant allegedly ignored. The patent’s prosecution history is invoked to argue that the claimed invention is a specific improvement over prior art systems, and it is noted that the patent has been cited as prior art in seven subsequent patent applications, including one assigned to Google Inc.
Case Timeline
| Date | Event |
|---|---|
| 2004-01-30 | U.S. Patent No. 8,571,916 Priority Date |
| 2013-10-29 | U.S. Patent No. 8,571,916 Issue Date |
| 2015-02-01 | Defendant MarketDial allegedly founded |
| 2017-12-18 | Plaintiff provides actual notice of the ’916 Patent to Defendant |
| 2018-09-17 | First Amended Complaint Filing Date |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 8,571,916 - “Methods, Systems, and Articles of Manufacture for Determining Optimal Parameter Settings for Business Initiative Testing Models”
- Issued: October 29, 2013.
The Invention Explained
- Problem Addressed: The patent describes a problem faced by retailers where it is difficult to accurately measure the impact of business initiatives due to "inconsistent data" or "noise" in their performance metrics. This makes it hard to identify which new initiatives will actually be profitable (’916 Patent, col. 2:1-10; Compl. ¶29).
- The Patented Solution: The invention is a computer-implemented method that aims to find the best way to analyze a business initiative before the analysis is actually performed. It does this by running a series of "virtual tests" on historical data, trying out different sets of analytical parameters (e.g., how to handle outlier data, which time periods to compare). The system identifies the parameter settings that produce the least "noise"—defined as the inconsistency between virtual test sites and control sites. These "optimal" parameter settings are then used to configure the analytical model that will be applied to a real-world business initiative, thereby producing more accurate results (’916 Patent, Abstract; ’916 Patent, col. 17:33-49).
- Technical Importance: The technology claims to provide a specific technical solution to a "big data" problem by automatically and empirically determining the optimal parameters for a testing model, an improvement over prior art systems that lacked this efficiency (Compl. ¶27, 32).
Key Claims at a Glance
- The complaint asserts independent claim 1 (Compl. ¶114).
- The essential elements of Claim 1 include:
- Identifying a business initiative testing model having a set of parameter settings.
- Selecting a first parameter setting set for a "virtual test."
- Performing the virtual test on a set of virtual test sites, where no actual initiative has been implemented.
- Determining actual performance data for the virtual test sites.
- Determining actual performance data for a set of control group sites.
- Determining a "noise value" reflecting the inconsistency between the virtual test sites' performance data and the control group sites' performance data.
- Determining a set of "optimal parameter settings" that "best minimize noise" based on the results of the virtual test.
- Configuring the business initiative testing model with the determined optimal parameter settings to test an actual business initiative.
- The complaint reserves the right to assert additional claims after discovery (Compl. ¶128).
III. The Accused Instrumentality
Product Identification
- The "MarketDial System" (also "Accused System"), a software platform for business analytics (Compl. ¶7).
Functionality and Market Context
- The Accused System is a software-as-a-service platform that allows customers to "construct and analyze a perfect A/B test" for business initiatives such as promotions or pricing changes (Compl. ¶84). A marketing graphic from the Defendant's website shows features including "AUTOMATED DATA UPLOAD" and "STORE ATTRIBUTE ANALYSIS" (Compl. ¶84, p.24). A screenshot of the system's "Create Test" interface shows a tool that guides users in selecting the quantity of stores and test length to achieve a desired "Estimated Confidence" level (Compl. ¶85, p.25).
- The complaint alleges that Defendant was founded specifically to compete with Plaintiff, has approached Plaintiff's clients, and has caused Plaintiff to lose business (Compl. ¶6). The Defendant's website is alleged to advertise that the system was "[b]uilt by ex-McKinsey and -BCG consultants" (Compl. ¶51).
IV. Analysis of Infringement Allegations
’916 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| identifying, by a computer, a business initiative testing model having a set of parameter settings | The Accused System is a business initiative testing model with parameter settings, including those affecting outlier data and the amount of historical data used. | ¶118 | col. 12:1-3 |
| selecting a first parameter setting set for performing a virtual test... | The Accused System allegedly selects a parameter set, such as how to treat outlier data (e.g., disregard or weigh) and the amount of historical data to analyze. | ¶119 | col. 19:10-14 |
| performing, by a computer, the virtual test on a set of virtual test sites, ... wherein each virtual test is a simulated business initiative test performed on test sites where no actual initiative test has been implemented... | The Accused System is alleged to perform a virtual test using historical performance data from a pre-period on sites where no new initiative has been implemented. | ¶122 | col. 12:4-13 |
| determining, by a computer, actual performance data associated with the set of virtual test sites | The Accused System allegedly obtains historical sales data for the virtual test sites from customer-uploaded transaction data. | ¶123 | col. 18:55-62 |
| determining, by a computer, actual performance data associated with a set of control group sites... | The Accused System allegedly obtains historical sales data for control group sites from customer-uploaded data. | ¶124 | col. 21:3-6 |
| determining a noise value for the first parameter setting set, the noise value reflecting an inconsistency between performance data... | The Accused System is alleged to use a virtual test to determine a noise value by comparing performance data between virtual test sites and control group sites. | ¶125 | col. 17:38-41 |
| determining, by a computer, a set of optimal parameter settings for the business initiative testing model based on results from the virtual test whereby the optimal parameter settings best minimize noise from the results | The Accused System allegedly chooses parameter setting options to minimize noise by finding the best match between the performance data of virtual test and control sites. | ¶126 | col. 17:45-47 |
| configuring, by a computer, the business initiative testing model using the optimal parameter settings to test a business initiative... | The Accused System allegedly uses the determined optimal settings to configure its "EASY STEP-BY-STEP TEST BUILDER" for analyzing an actual business initiative. | ¶127 | col. 17:47-49 |
- Identified Points of Contention:
- Scope Questions: The complaint's infringement theory is based almost entirely on "information and belief." A central dispute may arise over whether Defendant's marketing materials, which promise an "accurate, reliable testing toolkit," actually describe the specific, patented two-step process of first running a "virtual test" to optimize internal parameters and then using those parameters for a live test. The question for the court will be whether discovery confirms this specific mode of operation.
- Technical Questions: The claim requires determining a "noise value" that reflects an "inconsistency" between virtual test and control sites. A screenshot of the Accused System shows an "Estimated Confidence: 90.2%" metric (Compl. ¶85, p.25). A technical question will be what evidence shows that this "confidence" score is the claimed "noise value," as opposed to a standard statistical confidence interval or other metric that does not arise from the specific comparison required by the claim.
V. Key Claim Terms for Construction
The Term: "virtual test"
Context and Importance: This term is central to the claimed invention and the infringement allegations. The patent defines a "virtual test" as a "mock initiative test" performed in a "simulation environment" using historical data from sites "where no actual initiative test is planned or has been implemented" (’916 Patent, col. 12:4-13). Whether the internal processing of the Accused System meets this definition will be a critical issue.
Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification describes the process in broad terms as creating a "simulation environment" which could be argued by a patentee to cover any pre-analysis simulation on historical data to calibrate a model (’916 Patent, col. 12:13-14).
- Evidence for a Narrower Interpretation: The definition explicitly distinguishes a "virtual test" from an "actual initiative test" and states it is performed where no actual test is planned or implemented (’916 Patent, col. 12:7-9; col. 15:15-18). An accused infringer may argue this requires a completely separate, standalone process distinct from the calculations performed when a user sets up a real-world test.
The Term: "noise value"
Context and Importance: The patent's inventive concept centers on finding parameter settings that "best minimize noise." The complaint alleges the Accused System performs this step (Compl. ¶125-126). Practitioners may focus on this term because the infringement case depends on mapping a feature of the Accused System to this specific limitation.
Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification defines "noise" as a "quantified measurement of inconsistent performance data" and gives an example of a 0.5% difference in sales lift between test and control groups (’916 Patent, col. 17:38-39; col. 22:42-47). This could support an interpretation covering various statistical measures of error or variance.
- Evidence for a Narrower Interpretation: The claim language is specific, requiring the "noise value" to reflect "an inconsistency between performance data associated, with the set of virtual test sites and performance data associated with the set of control group sites" (’916 Patent, Claim 1). An accused infringer may argue that a generic "confidence score" or standard error from a regression model does not meet this specific comparative requirement.
VI. Other Allegations
- Willful Infringement: The complaint alleges willful infringement based on Defendant's alleged pre-suit knowledge of the ’916 Patent. The basis for this knowledge is a letter Plaintiff sent to Defendant on December 18, 2017. The complaint alleges that Defendant's continued infringement after this date has been deliberate and objectively reckless (Compl. ¶130-131).
VII. Analyst’s Conclusion: Key Questions for the Case
- A central evidentiary question will be one of operational reality versus allegation: will discovery show that the Accused System actually performs the claimed two-stage process—first, a "virtual test" to iteratively optimize internal model parameters by minimizing "noise," and second, using those optimized parameters to configure a model for an actual customer A/B test—or does it use a more conventional, single-stage statistical modeling approach that is merely described in marketing materials as "accurate"?
- A key legal question will be one of claim scope: can the term "virtual test," which the patent specification describes as a "mock" test performed where no "actual" initiative is planned, be construed to cover the internal computations that allegedly occur within the Accused System when a user is actively building a real-world test?
- A dispositive technical question will be one of functional equivalence: does the "Estimated Confidence" metric displayed in the Accused System's interface perform the function of the claimed "noise value," or is it a fundamentally different statistical measure that does not arise from comparing virtual test sites to control sites as required by the patent?