PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT / ATTORNEY-CLIENT COMMUNICATION
APPENDIX VII-A
DERIVED DATA COMMERCIALIZATION PROTOCOL
Appendix to Article VII of the Master Data License Agreement
June 2023
C&H Matter No. 1025-00008
Prepared by: Wendy Schultz, Esq. Omar Siddiqui, Esq. Calderwood & Harkness LLP
Parties:
Quorum Insurance Holdings Corp., a Delaware corporation ("Licensee" or "Quorum")
and
Meridian Credit intelligence, Inc., a Georgia corporation ("Licensor" or "Meridian")
STATUS: DRAFT — Subject to further negotiation; for discussion purposes only. This document does not constitute an offer, acceptance, or binding agreement. All terms remain subject to the mutual agreement of the parties and execution of definitive documentation.
Right-click to update Table of Contents
PREAMBLE AND RECITALS
WHEREAS, Quorum Insurance Holdings Corp., a corporation organized and existing under the laws of the State of Delaware ("Quorum"), and Meridian Credit intelligence, Inc., a corporation organized and existing under the laws of the State of Georgia ("Meridian"), have entered into, or are in the process of negotiating and finalizing, that certain Master Data License Agreement (the "MDLA"), pursuant to which Meridian will license certain data products and data feeds to Quorum for use in Quorum's insurance underwriting, actuarial, risk modeling, and related business operations;
WHEREAS, this Appendix VII-A (the "Protocol") is intended to be attached to and incorporated into Article VII of the MDLA, and shall constitute an integral part thereof, governing the specific terms and conditions under which Quorum may commercialize Derived Data (as defined herein) created by Quorum using Licensed Data provided by Meridian;
WHEREAS, pursuant to the MDLA, Quorum will receive licensed data products from Meridian, including but not limited to: (a) the MCI Consumer Credit File™, comprising approximately three hundred ten million (310,000,000) consumer records; (b) the MCI CommercialView™, comprising approximately forty-two million (42,000,000) business records; and (c) the MCI AltSignal™, comprising approximately eighty-five million (85,000,000) consumer alternative-data records (collectively, the "Licensed Data," as more fully defined herein and in the MDLA);
WHEREAS, Quorum intends to create Derived Data by blending, transforming, and processing Licensed Data with Quorum's proprietary first-party data — including, without limitation, Quorum's claims data, insurance policy data, and telematics data — to produce risk scores, propensity models, enriched datasets, and other analytics outputs for use in Quorum's insurance business and, subject to the terms of this Protocol, for Commercialization to third parties;
WHEREAS, the parties wish to establish a clear, objective, and measurable protocol governing the conditions under which Quorum may Commercialize Derived Data to third parties outside Quorum's corporate group, including the criteria for determining when Derived Data has been sufficiently transformed so as to be eligible for such Commercialization;
WHEREAS, Article VII of the MDLA provides that Quorum shall own all intellectual property rights in Derived Data, and this Protocol is intended to give effect to that ownership by establishing the permissible scope and conditions of Commercialization;
WHEREAS, this Protocol is intended to replace and supersede any subjective "sole discretion" standards previously proposed by either party for determining permissible Commercialization of Derived Data, and to substitute in their place objective, measurable, and scientifically grounded criteria that protect both Meridian's legitimate interest in safeguarding its Licensed Data and Quorum's legitimate interest in realizing the full commercial value of its proprietary analytics and data products;
NOW, THEREFORE, in consideration of the mutual covenants and agreements set forth herein, and for other good and valuable consideration, the receipt and sufficiency of which are hereby acknowledged, the parties agree to the following Protocol:
SECTION 1: DEFINITIONS
For purposes of this Protocol, the following terms shall have the meanings set forth below. Capitalized terms used but not defined in this Protocol shall have the meanings ascribed to them in the MDLA.
1.1 "Derived Data"
"Derived Data" means any data, datasets, models, model weights, parameters, scores, indices, risk assessments, propensity models, enriched datasets, analytics outputs, or other information created, generated, compiled, or produced by or on behalf of Quorum (including by Quorum's employees, contractors, agents, or service providers acting at Quorum's direction) using Licensed Data (in whole or in part) as an input, whether alone or in combination with Quorum's own first-party data, third-party data, or any other data sources. Derived Data expressly includes, without limitation:
(a) the outputs of machine-learning models, deep-learning models, statistical models, ensemble models, and algorithms that were trained on or with Licensed Data, whether or not the Licensed Data continues to be used as an input at the point of inference;
(b) feature-engineered datasets created using Licensed Data as a source or input;
(c) enriched datasets produced by combining Licensed Data with Quorum's proprietary claims, policy, telematics, or other first-party data;
(d) risk scores, propensity scores, classification labels, segment assignments, and other predictive analytics outputs generated using Licensed Data as a training or operational input; and
(e) indices, benchmarks, and aggregate statistical measures derived in whole or in part from Licensed Data.
Notwithstanding the foregoing, Derived Data shall not include Inference Outputs (as defined in Section 1.3 below). The distinction between Derived Data and Inference Outputs is addressed further in Section 1.3.
1.2 "Licensed Data"
"Licensed Data" has the meaning set forth in Article I of the MDLA and, for the avoidance of doubt, means the data delivered by Meridian to Quorum under the MDLA, comprising the MCI Consumer Credit File™, MCI CommercialView™, and MCI AltSignal™ data feeds, in their raw, unmodified form as delivered by Meridian to Quorum. Licensed Data does not include any data that has been transformed, processed, blended, or otherwise modified by Quorum after delivery.
1.3 "Inference Outputs"
"Inference Outputs" means live predictions, scores, classifications, rankings, recommendations, or other outputs generated by a Quorum model at the point of inference (i.e., in production or in a live operational environment) using new, non-Licensed-Data inputs, where the model may have been trained using Licensed Data but the inference input data does not itself include or incorporate any Licensed Data. For the avoidance of doubt, Inference Outputs are generated when a trained model is applied to new data that is independent of and does not contain Licensed Data, even though the model's parameters, weights, or learned representations may have been informed by Licensed Data during the training phase.
Inference Outputs are not Derived Data and are not subject to this Protocol. Quorum shall have unrestricted rights to use, distribute, license, sublicense, sell, and otherwise Commercialize Inference Outputs without any notice to, consent from, or obligation to Meridian. Nothing in this Protocol shall be construed to limit or condition Quorum's rights with respect to Inference Outputs.
1.4 "Sufficiently Transformed Derived Data" (or "STDD")
"Sufficiently Transformed Derived Data" or "STDD" means Derived Data that meets all of the Objective Transformation Criteria set forth in Section 3 of this Protocol, such that Licensed Data cannot be reverse-engineered, reconstructed, re-identified, or otherwise extracted from the Derived Data through commercially reasonable efforts. Derived Data that qualifies as STDD is eligible for Commercialization in accordance with this Protocol.
1.5 "QuorumScore™ Outputs"
"QuorumScore™ Outputs" means the outputs of Quorum's proprietary QuorumScore™ predictive underwriting model (United States Patent and Trademark Office application pending, Serial No. 97/482,116), including, without limitation, risk scores, propensity scores, segment classifications, and related analytics outputs generated by the QuorumScore™ system, whether or not Licensed Data was among the inputs used in training or operating the model. QuorumScore™ Outputs are subject to the QuorumScore™ Carve-Out set forth in Section 5 of this Protocol.
1.6 "Commercialization"
"Commercialization" means the sale, license, sublicense, distribution, provision of access to, or other transfer or making available of Derived Data (or any portion thereof, or access thereto) to any third party that is not a member of Quorum's corporate group (as defined in the MDLA), in exchange for monetary or other consideration, whether pursuant to a one-time transaction, a subscription arrangement, a platform access agreement, an API-based data service, or otherwise. For the avoidance of doubt, Quorum's internal use of Derived Data within its corporate group, including use by subsidiaries, affiliates, and joint ventures in which Quorum holds a controlling interest, shall not constitute Commercialization for purposes of this Protocol.
1.7 "Transformation Assessment"
"Transformation Assessment" means the process described in Section 4 of this Protocol for evaluating whether specific Derived Data qualifies as Sufficiently Transformed Derived Data by satisfying the Objective Transformation Criteria set forth in Section 3.
1.8 "Independent Data Expert"
"Independent Data Expert" means a mutually agreed-upon independent third-party data scientist, statistician, or other qualified technical expert with demonstrated expertise in data privacy, statistical disclosure limitation, machine-learning model evaluation, or a related discipline, engaged to conduct or adjudicate a Transformation Assessment in accordance with Section 4 of this Protocol. The Independent Data Expert shall have no material financial, employment, or consulting relationship with either party (other than the engagement under this Protocol) and shall execute appropriate confidentiality and conflict-of-interest undertakings prior to commencing any review.
1.9 "Notification Period"
"Notification Period" means the thirty (30) calendar day period following delivery of a Commercialization Notice by Quorum to Meridian, during which Meridian may review and, if applicable, dispute the Transformation Assessment in accordance with Section 6 of this Protocol.
1.10 Other Defined Terms
Capitalized terms used in this Protocol but not defined herein shall have the meanings given to such terms in the MDLA. Without limiting the generality of the foregoing, the following terms shall have the meanings set forth in the MDLA: "Intellectual Property," "Confidential Information," "Personal Information," "Quorum Corporate Group," "Data Processing Agreement," "Data Security Addendum," and "Term."
SECTION 2: OWNERSHIP OF DERIVED DATA
2.1 Quorum Ownership
Quorum shall own all right, title, and interest (including all intellectual property rights, trade secret rights, and other proprietary rights) in and to all Derived Data, including without limitation the right to use, modify, enhance, create derivative works of, Commercialize, and otherwise exploit Derived Data in any manner and for any purpose, subject to the conditions and procedures set forth in this Protocol. This ownership vests in Quorum automatically upon creation of the Derived Data, without any further action, assignment, or approval required.
No license, right, title, interest, or other claim in or to any Derived Data is granted, reserved, or implied in favor of Meridian under this Protocol, the MDLA, or otherwise. Meridian shall have no license-back, co-ownership, or other right or interest in any Derived Data, and this Protocol shall not be construed to create or imply any such license, right, or interest. The ownership provisions of this Section 2.1 are consistent with and intended to give effect to Article VII of the MDLA, which vests ownership of Derived Data exclusively in Quorum.
2.2 No Revenue Share
No revenue-sharing, royalty, fee-splitting, or other financial participation obligation shall apply to Quorum's internal use or external Commercialization of Derived Data or QuorumScore™ Outputs. Quorum shall retain one hundred percent (100%) of all revenues, fees, and other consideration received in connection with the Commercialization of Derived Data, without any obligation to account for, share, or remit any portion thereof to Meridian.
This Section 2.2 expressly supersedes and replaces any previously proposed revenue-share arrangement, including without limitation any proposal requiring Quorum to remit fifteen percent (15%) of net revenues from third-party Commercialization of Derived Data to Meridian. The parties acknowledge that the absence of a revenue-share obligation has been agreed upon as a term of this Protocol and is reflected in the overall economic terms of the MDLA.
2.3 Relationship to Licensed Data
Quorum's ownership of Derived Data does not diminish, impair, or otherwise affect Meridian's ownership of and rights in the Licensed Data in its raw, unmodified form as delivered to Quorum under the MDLA. Meridian retains all right, title, and interest in and to the Licensed Data.
Quorum shall not Commercialize Licensed Data in its raw, unmodified form. For the avoidance of doubt, this Protocol governs only the Commercialization of Derived Data (including STDD and QuorumScore™ Outputs); it does not authorize and shall not be construed to authorize the resale, redistribution, or sublicensing of Licensed Data in its raw, unmodified form to any third party.
SECTION 3: OBJECTIVE TRANSFORMATION CRITERIA
3.1 Purpose
The purpose of this Section 3 is to establish clear, measurable, and objective criteria (the "Objective Transformation Criteria") for determining whether Derived Data qualifies as Sufficiently Transformed Derived Data eligible for Commercialization. The Objective Transformation Criteria are intended to replace any subjective, discretionary, or "sole discretion" standard, and to provide a transparent, repeatable framework that both protects Meridian's legitimate interest in the integrity and exclusivity of its Licensed Data and respects Quorum's ownership of and commercial interest in Derived Data. Derived Data must satisfy each of the criteria set forth in Sections 3.2 through 3.5 (subject to the presumption in Section 3.6) to qualify as STDD.
3.2 Statistical Irreversibility Test
Derived Data shall be deemed to satisfy the statistical irreversibility criterion if, using commercially reasonable statistical, computational, or analytical methods, it is not possible to reconstruct, derive, or reverse-engineer the underlying Licensed Data records (or any material subset thereof, defined as more than one percent (1%) of the records in the relevant Licensed Data feed) from the Derived Data.
The following quantitative threshold shall apply: the reconstruction accuracy of individual Licensed Data records from the Derived Data must be no greater than what would be achievable from publicly available data sources alone (the "Baseline Reconstruction Rate"), plus a margin of no more than five (5) percentage points. For purposes of this test:
(a) "Reconstruction accuracy" means the percentage of individual Licensed Data fields (at the record level) that can be correctly inferred, predicted, or reconstructed from the Derived Data using state-of-the-art data reconstruction techniques.
(b) The "Baseline Reconstruction Rate" shall be determined empirically by attempting reconstruction using only publicly available data sources (e.g., public records, census data, publicly available consumer data aggregations) and measuring the resulting accuracy.
(c) "Commercially reasonable" methods include, but are not limited to, standard data-reconstruction attack methods such as linkage attacks, membership inference attacks, model inversion attacks, attribute inference attacks, and reconstruction attacks described in the peer-reviewed data privacy and machine-learning literature as of the date of the applicable Transformation Assessment.
The testing methodology shall be performed using a testing plan agreed upon by the parties or, in the event of disagreement, determined by the Independent Data Expert. The testing plan shall specify the reconstruction methods to be employed, the sample sizes, the Licensed Data feeds to be tested, and the metrics for measuring reconstruction accuracy.
3.3 Aggregation and Abstraction Standard
Derived Data must represent a level of aggregation, abstraction, or transformation that renders individual Licensed Data records unidentifiable within the Derived Data. Specifically:
(a) No individual consumer record from the MCI Consumer Credit File™ or MCI AltSignal™ shall be identifiable or isolable within the Derived Data.
(b) No individual business record from the MCI CommercialView™ shall be identifiable or isolable within the Derived Data.
(c) Outputs must represent aggregated, scored, modeled, indexed, or otherwise abstracted information that cannot be disaggregated to the level of individual Licensed Data records through commercially reasonable efforts.
Where Derived Data takes the form of statistical models, model parameters, indices, aggregate risk scores, segment-level classifications, or similar analytics outputs, the Aggregation and Abstraction Standard shall be presumed satisfied unless Meridian demonstrates, through the Transformation Assessment process in Section 4, that individual Licensed Data records are in fact identifiable or isolable within such outputs.
3.4 De-Identification Requirement
To the extent Derived Data contains or is derived from personal information of consumers (including information derived from the MCI Consumer Credit File™ or MCI AltSignal™), the Derived Data must satisfy de-identification standards consistent with the definition of "deidentified" information under the California Consumer Privacy Act, as amended by the California Privacy Rights Act (Cal. Civ. Code § 1798.140(m)), and applicable requirements under the Fair Credit Reporting Act ("FCRA"), 15 U.S.C. § 1681 et seq.
Without limiting the generality of the foregoing, Quorum shall:
(a) Implement technical safeguards reasonably designed to prevent re-identification of de-identified Derived Data, including appropriate data masking, hashing, tokenization, generalization, or suppression techniques as applicable to the specific Derived Data at issue;
(b) Implement business processes that specifically prohibit any attempt to re-identify de-identified Derived Data once such data has been Commercialized;
(c) Make no attempt to re-identify de-identified Derived Data and shall contractually require any third-party recipients of Commercialized Derived Data to refrain from re-identification; and
(d) Maintain publicly available information, to the extent required by applicable law, regarding its de-identification commitments and processes.
3.5 No Direct Incorporation
Derived Data eligible for Commercialization shall not contain any raw Licensed Data fields, tradeline-level records, individual consumer credit records, individual business credit records, or other Licensed Data records in their original or substantially similar form. For purposes of this Section 3.5:
(a) "Substantially similar form" means data that has been altered only by trivial modifications — including, without limitation, reformatting, field renaming, unit conversion, date-format changes, encoding changes, or other cosmetic or structural alterations — without substantive transformation of the underlying data content.
(b) The inclusion of a raw Licensed Data field (such as an individual consumer's credit score, tradeline balance, delinquency status, or payment history, as delivered by Meridian) within a Derived Data record, even if combined with other data fields, shall constitute direct incorporation for purposes of this Section 3.5.
For the avoidance of doubt, the use of Licensed Data as an input to a model, algorithm, or analytical process that produces a transformed output (e.g., a blended risk score, a propensity classification, or a segment assignment) does not constitute "direct incorporation" of Licensed Data, provided that the resulting output does not reproduce the raw Licensed Data field values in their original or substantially similar form.
3.6 Presumption for Model Outputs
Where Derived Data consists of the outputs of a machine-learning model, statistical model, actuarial model, or algorithm — including, without limitation, risk scores, propensity scores, classification labels, segment assignments, predicted loss ratios, and similar predictive analytics outputs — such outputs shall be presumptively deemed to satisfy the Objective Transformation Criteria set forth in Sections 3.2 through 3.5, provided that:
(a) The model was trained on Licensed Data combined with at least one other material data source (e.g., Quorum's proprietary first-party claims data, policy data, telematics data, or other third-party data), such that the Licensed Data is not the sole input to the model; and
(b) The outputs do not reproduce or allow reconstruction of individual Licensed Data records in their original or substantially similar form, as determined under the No Direct Incorporation standard in Section 3.5.
This presumption is rebuttable. Meridian may rebut the presumption only by demonstrating, through the Transformation Assessment process in Section 4 (including, if necessary, through review by an Independent Data Expert under Section 4.2), that specific outputs fail the Statistical Irreversibility Test set forth in Section 3.2. The burden of proof for rebutting the presumption shall rest with Meridian.
3.7 Updates to Criteria
The Objective Transformation Criteria set forth in this Section 3 may be updated or modified by mutual written agreement of the parties, such agreement not to be unreasonably withheld, conditioned, or delayed by either party. The parties acknowledge that developments in technology, data science methods, privacy-enhancing technologies, regulatory requirements, and applicable law may necessitate updates to the criteria from time to time.
Either party may propose updates to the Objective Transformation Criteria by delivering written notice to the other party specifying the proposed update, the rationale therefor, and any supporting data, research, or legal authority. The parties shall negotiate in good faith to reach agreement on any proposed update within sixty (60) calendar days of receipt of such notice.
If the parties are unable to reach agreement on a proposed update to the Objective Transformation Criteria, the disagreement shall be resolved through the dispute resolution mechanism set forth in Section 7 of this Protocol.
SECTION 4: TRANSFORMATION ASSESSMENT PROCESS
4.1 Self-Assessment by Quorum
Prior to any Commercialization of Derived Data (other than QuorumScore™ Outputs, which are subject to the carve-out in Section 5), Quorum shall conduct an internal Transformation Assessment applying each element of the Objective Transformation Criteria set forth in Section 3 to the specific Derived Data that Quorum intends to Commercialize.
The self-assessment shall be documented in a written Transformation Assessment Report (a "Report"), which shall be signed by a qualified member of Quorum's data science, data engineering, or analytics team. Qualified signatories include Rajesh Anand, Chief Data Officer; Priya Nair, Director, Data Engineering; or their respective designees, provided that any designee shall have sufficient technical qualifications and organizational authority to attest to the accuracy and completeness of the Report.
Each Transformation Assessment Report shall describe, at a minimum:
(i) the Derived Data at issue, including a general description of the dataset, output, score, model, or other deliverable to be Commercialized;
(ii) the Licensed Data inputs used in the creation of the Derived Data, specifying which Licensed Data feeds (MCI Consumer Credit File™, MCI CommercialView™, MCI AltSignal™, or a combination thereof) were utilized;
(iii) the transformation, processing, modeling, or analytical methods applied to the Licensed Data in the creation of the Derived Data, described at a level of detail sufficient to demonstrate compliance with the Objective Transformation Criteria without requiring disclosure of Quorum's proprietary algorithms, model architectures, or trade secrets;
(iv) the results of the Statistical Irreversibility Test conducted in accordance with Section 3.2, including the testing methodology employed, the Baseline Reconstruction Rate, the observed reconstruction accuracy, and the margin;
(v) confirmation that the Derived Data satisfies the Aggregation and Abstraction Standard set forth in Section 3.3, with a description of the aggregation or abstraction techniques applied;
(vi) confirmation that the Derived Data satisfies the De-Identification Requirement set forth in Section 3.4, with a description of the de-identification techniques applied and the technical safeguards implemented; and
(vii) confirmation that no raw Licensed Data fields are directly incorporated in the Derived Data in their original or substantially similar form, in accordance with Section 3.5.
Where the presumption for model outputs set forth in Section 3.6 applies, the Report may reference the presumption and provide the basis for its application (i.e., identifying the material non-Licensed-Data sources used in training and confirming non-reproduction of Licensed Data records), in lieu of conducting the full Statistical Irreversibility Test, unless the presumption is subsequently rebutted by Meridian.
4.2 Independent Data Expert Review (Optional / Dispute-Triggered)
Quorum may, at its sole election and expense, retain an Independent Data Expert to conduct or validate the Transformation Assessment prior to Commercialization. Quorum's voluntary election to obtain an Independent Data Expert review shall not create any obligation to do so for future Transformation Assessments.
If Meridian disputes the Transformation Assessment pursuant to Section 6.4 of this Protocol, an Independent Data Expert review shall be mandatory. In such event, the Independent Data Expert shall be selected as follows:
(a) The parties shall mutually agree on the selection of an Independent Data Expert within fifteen (15) business days of the date on which Quorum receives a valid Dispute Notice from Meridian.
(b) If the parties are unable to agree on the selection of an Independent Data Expert within such fifteen (15) business day period, each party shall, within five (5) business days following the expiration of the initial selection period, nominate one (1) candidate who satisfies the qualifications set forth in Section 1.8. The two (2) nominees shall then jointly select a third individual who satisfies such qualifications, and such third individual shall serve as the Independent Data Expert. If the two nominees are unable to agree on a third individual within ten (10) business days, either party may apply to the American Arbitration Association for appointment of a qualified expert.
(c) The Independent Data Expert shall, prior to commencing any review, execute (i) a confidentiality agreement with terms no less protective than those set forth in the MDLA, and (ii) a conflict-of-interest certification confirming the absence of any material financial, employment, or consulting relationship with either party.
Cost Allocation. If the Independent Data Expert determines that the Derived Data at issue satisfies the Objective Transformation Criteria, Meridian shall bear all reasonable fees and costs of the Independent Data Expert. If the Independent Data Expert determines that the Derived Data does not satisfy the Objective Transformation Criteria, Quorum shall bear all reasonable fees and costs of the Independent Data Expert.
4.3 Scope of Independent Data Expert Review
The Independent Data Expert shall evaluate whether the Derived Data at issue satisfies each element of the Objective Transformation Criteria set forth in Section 3 (including Sections 3.2, 3.3, 3.4, and 3.5), taking into account the presumption set forth in Section 3.6 where applicable. The Expert's evaluation shall be limited to the specific Derived Data identified in the Commercialization Notice and the specific criteria identified in Meridian's Dispute Notice; the Expert shall not undertake a broader review of Quorum's data practices, systems, or other Derived Data products.
The Independent Data Expert shall issue a written determination (the "Expert Determination") within thirty (30) calendar days of engagement, or such longer period as the parties may mutually agree in writing. The Expert Determination shall set forth the Expert's findings as to each element of the Objective Transformation Criteria, the methodology employed by the Expert, and the Expert's conclusion as to whether the Derived Data qualifies as Sufficiently Transformed Derived Data.
The Expert Determination shall be final and binding on the parties, absent manifest error (as defined in Section 7.3), and shall not be subject to appeal, review, or challenge except through the manifest error review process set forth in Section 7.3.
Each party shall cooperate reasonably with the Independent Data Expert, including by providing access to data, documentation, and personnel as reasonably requested by the Expert. Notwithstanding the foregoing, Quorum shall not be required to disclose proprietary algorithms, model architectures, source code, or trade secrets to the Expert except to the extent strictly necessary for the Expert to evaluate compliance with the Objective Transformation Criteria, and any such disclosure shall be subject to the confidentiality protections referenced in Section 4.2(c).
SECTION 5: QUORUMSCORE™ CARVE-OUT
5.1 Express Exemption
QuorumScore™ Outputs are expressly exempt from the Transformation Assessment process set forth in Section 4 and the Notification requirements set forth in Section 6 of this Protocol. Quorum may Commercialize QuorumScore™ Outputs to any third party, at any time, without restriction, without prior notice to Meridian, and without any requirement to demonstrate satisfaction of the Objective Transformation Criteria. For the avoidance of doubt, the Commercialization of QuorumScore™ Outputs shall not require the preparation of a Transformation Assessment Report, the delivery of a Commercialization Notice, or any approval, consent, or review by Meridian.
5.2 Rationale and Basis for Carve-Out
The parties acknowledge and agree that the QuorumScore™ carve-out is based on the following considerations:
(a) QuorumScore™ is Quorum's proprietary predictive underwriting model, which is the subject of a pending application before the United States Patent and Trademark Office (Serial No. 97/482,116). QuorumScore™ represents Quorum's significant investment in data science, actuarial modeling, and software engineering, and constitutes Quorum's proprietary intellectual property.
(b) QuorumScore™ blends multiple data sources — including, without limitation, Quorum's own first-party claims data, insurance policy data, and telematics data — together with Licensed Data and other third-party data sources, to produce risk scores, propensity scores, segment classifications, and related analytics outputs. The QuorumScore™ system applies proprietary algorithms, feature engineering, ensemble modeling techniques, and other transformative processes to its inputs.
(c) QuorumScore™ Outputs constitute Inference Outputs and/or model-generated scores that are inherently transformed and abstracted. By their nature, QuorumScore™ Outputs satisfy the Objective Transformation Criteria — they do not reproduce raw Licensed Data, they represent highly transformed and aggregated model outputs, they are statistically irreversible, and they do not contain individually identifiable Licensed Data records.
(d) Quorum's planned business-to-business ("B2B") licensing of QuorumScore™ Outputs, with projected revenues of approximately forty million to fifty million dollars ($40,000,000–$50,000,000) by 2027, is a core element of Quorum's business strategy. The Commercialization of QuorumScore™ Outputs must not be impeded, delayed, or conditioned on Meridian's review, approval, or consent.
5.3 Limitations of Carve-Out
The QuorumScore™ carve-out set forth in this Section 5 is limited to QuorumScore™ Outputs as defined in Section 1.5. The carve-out does not extend to:
(a) the distribution, sale, license, or other Commercialization of underlying training data used to train the QuorumScore™ system, including any raw or processed Licensed Data used as training inputs;
(b) the distribution of raw feature sets or feature vectors derived from Licensed Data and used as inputs to the QuorumScore™ system, to the extent such feature sets contain or allow reconstruction of Licensed Data records; or
(c) any dataset or data product that would allow the reconstruction of Licensed Data records, regardless of whether it is associated with the QuorumScore™ system.
If Quorum materially modifies the QuorumScore™ system such that its outputs begin to incorporate raw Licensed Data fields (rather than transformed model outputs), the carve-out set forth in this Section 5 shall not apply to such modified outputs, and the standard Transformation Assessment process set forth in Sections 4 and 6 shall apply to any Commercialization of such modified outputs. For purposes of this Section 5.3, a "material modification" means a change to the QuorumScore™ system that results in QuorumScore™ Outputs containing identifiable, raw Licensed Data fields in their original or substantially similar form.
SECTION 6: NOTIFICATION (NOT CONSENT) MECHANISM
6.1 Commercialization Notice
At least thirty (30) calendar days prior to any Commercialization of Derived Data (other than QuorumScore™ Outputs, which are exempt pursuant to Section 5), Quorum shall deliver to Meridian a written notice (a "Commercialization Notice"). Each Commercialization Notice shall include:
(a) A description of the Derived Data to be Commercialized, including the type of data product (e.g., enriched dataset, risk score, propensity model, index, benchmark), the general content and structure of the data, and the Licensed Data feeds used in its creation, at a level of detail sufficient for Meridian to evaluate the Transformation Assessment, but not requiring disclosure of Quorum's proprietary algorithms, model architectures, hyperparameters, or trade secrets;
(b) The identity of the proposed third-party recipient(s) of the Derived Data or, if the specific recipient(s) have not yet been identified at the time of the Notice, a description of the category of intended recipients (e.g., property and casualty insurers, reinsurers, financial institutions, insurtech platforms);
(c) A copy of the Transformation Assessment Report prepared by Quorum in accordance with Section 4.1; and
(d) A statement, signed by the signatory of the Transformation Assessment Report, that Quorum has determined, in good faith, that the Derived Data satisfies the Objective Transformation Criteria set forth in Section 3 of this Protocol and qualifies as Sufficiently Transformed Derived Data.
6.2 No Consent Required
The delivery of a Commercialization Notice is for informational purposes only. Meridian's consent, approval, authorization, or agreement is not required for Quorum to proceed with the Commercialization of Sufficiently Transformed Derived Data. The Notification mechanism set forth in this Section 6 is intended to provide Meridian with transparency and the opportunity to review Quorum's Transformation Assessment, but does not constitute a consent or approval mechanism.
Meridian's failure to respond to a Commercialization Notice, or failure to deliver a valid Dispute Notice in accordance with Section 6.4, within the thirty (30) calendar day Notification Period shall be deemed acceptance of Quorum's Transformation Assessment. Upon expiration of the Notification Period without delivery of a valid Dispute Notice, Quorum may proceed with Commercialization of the Derived Data described in the Commercialization Notice without any further obligation to Meridian with respect to such Derived Data.
6.3 Meridian Review Rights During Notification Period
During the thirty (30) calendar day Notification Period, Meridian shall have the right to review the Commercialization Notice and the accompanying Transformation Assessment Report. Meridian may request reasonable additional information from Quorum to evaluate the Transformation Assessment, and Quorum shall respond to such requests in good faith and within a reasonable time period (not to exceed ten (10) business days from receipt of such request).
Notwithstanding the foregoing, Quorum shall not be required to disclose the following in response to Meridian's information requests:
(a) Quorum's proprietary algorithms, model architectures, source code, or model hyperparameters;
(b) Quorum's trade secrets or proprietary business information unrelated to the Transformation Assessment;
(c) the identity of specific third-party data sources (other than Licensed Data) used in the creation of the Derived Data, except in general terms (e.g., "first-party claims data," "telematics data," "third-party geospatial data"); or
(d) confidential commercial terms of proposed Commercialization arrangements with third parties.
If Meridian does not deliver a Dispute Notice (as defined in Section 6.4) within the Notification Period, Quorum may proceed with Commercialization without further obligation to Meridian with respect to the specific Derived Data described in the Commercialization Notice.
6.4 Dispute Notice
If Meridian disputes Quorum's Transformation Assessment, Meridian shall deliver a written dispute notice (a "Dispute Notice") to Quorum within the thirty (30) calendar day Notification Period. To be valid, a Dispute Notice must include all of the following:
(a) Identification of the specific element(s) of the Objective Transformation Criteria (as set forth in Sections 3.2, 3.3, 3.4, and/or 3.5) that Meridian alleges are not satisfied by the Derived Data described in the Commercialization Notice;
(b) A reasonably detailed explanation of the basis for Meridian's objection, including the factual, technical, or analytical grounds upon which Meridian relies; and
(c) Any supporting data, analysis, test results, or other evidence in Meridian's possession or control that supports Meridian's objection.
A Dispute Notice that fails to specify the particular Objective Transformation Criteria at issue, fails to provide a reasoned basis for the objection, or consists solely of conclusory assertions without supporting detail or analysis, shall be deemed deficient. A deficient Dispute Notice shall not toll, extend, or otherwise affect the Notification Period, and Quorum may proceed with Commercialization upon expiration of the Notification Period notwithstanding the delivery of a deficient Dispute Notice.
6.5 Effect of Dispute Notice
Upon receipt of a valid Dispute Notice (i.e., a Dispute Notice that satisfies the requirements of Section 6.4), Quorum shall refrain from Commercializing the specific Derived Data that is the subject of the Dispute Notice until the dispute is resolved through the Independent Data Expert process set forth in Section 4.2 or, if necessary, through the dispute resolution mechanism set forth in Section 7.
For the avoidance of doubt, a valid Dispute Notice does not affect Quorum's right to:
(a) Commercialize other Derived Data that is not the subject of the Dispute Notice, including Derived Data for which the Notification Period has expired without a valid Dispute Notice;
(b) Commercialize QuorumScore™ Outputs, which are exempt from this Section 6 pursuant to the carve-out in Section 5;
(c) Commercialize Inference Outputs, which are not subject to this Protocol; or
(d) Continue to use, hold, and process the disputed Derived Data internally within Quorum's corporate group.
SECTION 7: DISPUTE RESOLUTION FOR TRANSFORMATION DISPUTES
7.1 Initial Escalation
Any dispute arising under this Protocol regarding whether specific Derived Data satisfies the Objective Transformation Criteria shall first be referred to the senior business representatives of each party for resolution through good-faith negotiation. The designated senior business representatives for purposes of this Section 7.1 are:
For Quorum: Rajesh Anand, Chief Data Officer (or his successor in that role);
For Meridian: Keith Marlow, Senior Vice President, Data Licensing & Partnerships (or his successor in that role).
The senior business representatives shall meet (in person or by videoconference) within ten (10) business days of referral and shall use good-faith efforts to resolve the dispute within fifteen (15) business days of the initial referral. Either party may refer a dispute to the senior business representatives by delivering written notice to the other party identifying the dispute and requesting escalation under this Section 7.1.
7.2 Independent Data Expert Determination
If the senior business representatives are unable to resolve the dispute within fifteen (15) business days of the initial referral (or such longer period as the parties may mutually agree), the dispute shall be submitted to an Independent Data Expert for determination in accordance with Section 4.2 and Section 4.3 of this Protocol.
The Independent Data Expert's determination shall be final and binding on the parties, subject only to a claim of manifest error as set forth in Section 7.3 below. Neither party may seek judicial or arbitral review of the Expert Determination except on grounds of manifest error.
7.3 Manifest Error Review
A party claiming manifest error in the Independent Data Expert's determination may seek review under the dispute resolution provisions of the MDLA (Article [__]).
[Note: The MDLA's general dispute resolution article remains unresolved. Quorum's preferred position is exclusive jurisdiction in the United States District Court for the Southern District of New York or, alternatively, the Supreme Court of the State of New York, Commercial Division, New York County. Meridian's preferred position is binding arbitration administered by the American Arbitration Association under the AAA Commercial Arbitration Rules, with a panel of three (3) arbitrators, seated in New York, New York. This Section 7.3 should be updated to reflect the agreed MDLA dispute resolution mechanism once finalized.]
The scope of manifest error review shall be strictly limited to the following grounds:
(a) Demonstrable mathematical or computational error in the Expert's analysis or calculations;
(b) The Expert's demonstrable failure to apply the Objective Transformation Criteria as written in Section 3 of this Protocol (e.g., the Expert applied a standard different from the criteria specified herein); or
(c) A demonstrable conflict of interest on the part of the Independent Data Expert that was not disclosed prior to the Expert's engagement and that materially affected the Expert's determination.
Manifest error review shall not extend to disagreements with the Expert's judgment, methodology (provided the methodology is consistent with the Objective Transformation Criteria), or weighing of evidence, nor to any other ground not specified in clauses (a) through (c) above.
7.4 Interim Commercialization Rights
If the Independent Data Expert determines that the Derived Data at issue satisfies the Objective Transformation Criteria, Quorum may immediately proceed with Commercialization of such Derived Data. Any pending manifest-error review initiated by Meridian following a favorable Expert Determination shall not operate as a stay, injunction, or restraint on Quorum's right to Commercialize the Derived Data during the pendency of such review.
If the Independent Data Expert determines that the Derived Data at issue does not satisfy the Objective Transformation Criteria, Quorum shall not Commercialize the specific Derived Data that was the subject of the Expert Determination unless and until:
(a) Quorum modifies, further transforms, or reprocesses the Derived Data to satisfy the Objective Transformation Criteria, and completes a new Transformation Assessment in accordance with Section 4.1, followed by delivery of a new Commercialization Notice in accordance with Section 6.1; or
(b) The Expert Determination is reversed on manifest error review under Section 7.3, with a final determination that the Derived Data does in fact satisfy the Objective Transformation Criteria.
SECTION 8: RECORDS AND AUDIT
8.1 Record-Keeping
Quorum shall maintain complete and accurate records of all Transformation Assessments, Commercialization Notices, Dispute Notices (if any), Independent Data Expert Determinations (if any), and related documentation for a period of five (5) years from the date of each Commercialization of Derived Data. Such records shall include, at a minimum:
(a) The Transformation Assessment Report prepared under Section 4.1;
(b) Any Independent Data Expert Determination issued under Section 4.3;
(c) Copies of all Commercialization Notices delivered to Meridian;
(d) Documentation of each Commercialization, including the date of Commercialization, the general category of the third-party recipient (e.g., insurer, reinsurer, financial institution, insurtech platform), and a general description of the Derived Data provided; and
(e) Correspondence between the parties relating to any Dispute Notice or dispute resolution under Sections 6 and 7.
Records shall be maintained in a secure, organized, and readily retrievable format. Quorum shall designate a records custodian responsible for compliance with this Section 8.1.