{"id":{"repo_id":"penn","oai_identifier":"oai:repository.upenn.edu:20.500.14332/62758"},"canonical_url":"https://search.dev.ndltd.org/etd/penn/oai:repository.upenn.edu:20.500.14332/62758","repository":{"repo_id":"penn","name":"University of Pennsylvania","base_url":"https://repository.upenn.edu/server/oai/request"},"display":{"title":"Essays on the Decision Value of Data in Marketing Measurement and Targeting","abstract":"This dissertation studies how marketers can make better personalization and measurement decisions when the usefulness of data is limited by heterogeneity, identification, and privacy constraints. Chapter 2 develops a model of actionable heterogeneity and shows that heterogeneity alone is not enough for personalization to outperform the best uniform policy. The chapter characterizes how within-treatment heterogeneity, cross-treatment correlation, and variation in average responses determine the expected gain from personalization, and applies the framework to large-scale field experiments on flu-vaccination nudges. Chapter 3 examines model specification in modern marketing mix models and shows that nonlinear and time-varying effects are often not separately identifiable from standard aggregate spending data. The chapter demonstrates the problem theoretically and empirically, quantifies its implications for budget allocation, and proposes practical guardrails for reducing conflation risk. Chapter 4 studies personalization under third-party privacy protections that restrict analysts to a limited number of noisy aggregate queries rather than raw individual-level data. It develops a strategic querying method based on Bayesian optimization, integral posterior updating, and a targeting-aware acquisition function, and shows that the method can recover most of the targeting value achieved by non-privacy-preserving machine learning benchmarks. Together, the three essays characterize when data is helpful for personalization and marketing measurement, when it is not, and how data collection can be redesigned to improve its decision value.","abstract_html":"This dissertation studies how marketers can make better personalization and measurement decisions when the usefulness of data is limited by heterogeneity, identification, and privacy constraints. Chapter 2 develops a model of actionable heterogeneity and shows that heterogeneity alone is not enough for personalization to outperform the best uniform policy. The chapter characterizes how within-treatment heterogeneity, cross-treatment correlation, and variation in average responses determine the expected gain from personalization, and applies the framework to large-scale field experiments on flu-vaccination nudges. Chapter 3 examines model specification in modern marketing mix models and shows that nonlinear and time-varying effects are often not separately identifiable from standard aggregate spending data. The chapter demonstrates the problem theoretically and empirically, quantifies its implications for budget allocation, and proposes practical guardrails for reducing conflation risk. Chapter 4 studies personalization under third-party privacy protections that restrict analysts to a limited number of noisy aggregate queries rather than raw individual-level data. It develops a strategic querying method based on Bayesian optimization, integral posterior updating, and a targeting-aware acquisition function, and shows that the method can recover most of the targeting value achieved by non-privacy-preserving machine learning benchmarks. Together, the three essays characterize when data is helpful for personalization and marketing measurement, when it is not, and how data collection can be redesigned to improve its decision value.","abstract_has_math":false,"creators":["Shchetkina, Anna"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Berman, Ron"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-07-24T03:45:15Z","subjects":["Marketing"],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://repository.upenn.edu/handle/20.500.14332/62758","outbound_label":"Repository record","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Berman, Ron"]},{"key":"dc:creator","label":"Author","values":["Shchetkina, Anna"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-06-05T16:13:34Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-06-05T16:13:34Z"]},{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:type","label":"Dc Type","values":["Dissertation/Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Marketing"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://repository.upenn.edu/handle/20.500.14332/62758"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["2026"]},{"key":"dc:description.abstract","label":"Abstract","values":["This dissertation studies how marketers can make better personalization and measurement decisions when the usefulness of data is limited by heterogeneity, identification, and privacy constraints. Chapter 2 develops a model of actionable heterogeneity and shows that heterogeneity alone is not enough for personalization to outperform the best uniform policy. The chapter characterizes how within-treatment heterogeneity, cross-treatment correlation, and variation in average responses determine the expected gain from personalization, and applies the framework to large-scale field experiments on flu-vaccination nudges. Chapter 3 examines model specification in modern marketing mix models and shows that nonlinear and time-varying effects are often not separately identifiable from standard aggregate spending data. The chapter demonstrates the problem theoretically and empirically, quantifies its implications for budget allocation, and proposes practical guardrails for reducing conflation risk. Chapter 4 studies personalization under third-party privacy protections that restrict analysts to a limited number of noisy aggregate queries rather than raw individual-level data. It develops a strategic querying method based on Bayesian optimization, integral posterior updating, and a targeting-aware acquisition function, and shows that the method can recover most of the targeting value achieved by non-privacy-preserving machine learning benchmarks. Together, the three essays characterize when data is helpful for personalization and marketing measurement, when it is not, and how data collection can be redesigned to improve its decision value."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["PhD"]},{"key":"dc:title","label":"Title","values":["Essays on the Decision Value of Data in Marketing Measurement and Targeting"]}]}],"canonical_facts":{"dc:contributor.advisor":["Berman, Ron"],"dc:creator":["Shchetkina, Anna"],"dc:date.accessioned":["2026-06-05T16:13:34Z"],"dc:date.available":["2026-06-05T16:13:34Z"],"dc:date.issued":["2026"],"dc:description":["2026"],"dc:description.abstract":["This dissertation studies how marketers can make better personalization and measurement decisions when the usefulness of data is limited by heterogeneity, identification, and privacy constraints. Chapter 2 develops a model of actionable heterogeneity and shows that heterogeneity alone is not enough for personalization to outperform the best uniform policy. The chapter characterizes how within-treatment heterogeneity, cross-treatment correlation, and variation in average responses determine the expected gain from personalization, and applies the framework to large-scale field experiments on flu-vaccination nudges. Chapter 3 examines model specification in modern marketing mix models and shows that nonlinear and time-varying effects are often not separately identifiable from standard aggregate spending data. The chapter demonstrates the problem theoretically and empirically, quantifies its implications for budget allocation, and proposes practical guardrails for reducing conflation risk. Chapter 4 studies personalization under third-party privacy protections that restrict analysts to a limited number of noisy aggregate queries rather than raw individual-level data. It develops a strategic querying method based on Bayesian optimization, integral posterior updating, and a targeting-aware acquisition function, and shows that the method can recover most of the targeting value achieved by non-privacy-preserving machine learning benchmarks. Together, the three essays characterize when data is helpful for personalization and marketing measurement, when it is not, and how data collection can be redesigned to improve its decision value."],"dc:description.degree":["PhD"],"dc:identifier.uri":["https://repository.upenn.edu/handle/20.500.14332/62758"],"dc:language.iso":["en"],"dc:subject":["Marketing"],"dc:title":["Essays on the Decision Value of Data in Marketing Measurement and Targeting"],"dc:type":["Dissertation/Thesis"]},"updated_at":"2026-07-24T03:45:15Z"}