Ankit Sisodia
Research Subjects
Ankit Sisodia has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
Alex Burnap
1 shared publication- Alex Burnap
MIT Sloan School of Management
1 shared publication - Hortense Fong
Columbia University, New York, New York 10027
1 shared publication Jingyi Zhang
1 shared publicationK. Sudhir
1 shared publicationMayur Choudhary
1 shared publication
Latest Publications
- Open Access
- 26 October 2026
- Published by Elsevier in SSRN Electronic Journal
What Does Sincerity Sound Like? Designing Brand Personality Using Audio
Brands increasingly use audio logos, short and distinctive sonic signatures, as central elements of their identity across advertising, digital products and services, and points of sale. Listeners infer a brand's personality from these few seconds of sound, so design choices shape how a brand is perceived. Yet empirical guidance about which sound features convey which personality is limited, and whether such guidance would improve on designers' intuition and experience is an open question. We compile a novel dataset of 696 audio logos and combine interpretable acoustic features, grounded in music theory and perception, with flexible machine-learning models to identify the sound properties associated with each of Aaker's (1997) five brand personality dimensions. Focusing on Sincerity, a dimension practitioners consider important but difficult to convey through sound, we translate these relationships into directional design insights (e.g., major mode is associated with greater Sincerity) and test them in a field experiment in which professional sound designers redesign unreleased audio logos to sound more sincere. Designers' unaided intuitions align with many of the identified relationships, but designers who receive the insights are more likely to adjust features in the recommended directions and make larger adjustments. Their redesigns are rated significantly more sincere than the originals and marginally more sincere than logos redesigned without the insights, with no detectable shifts in other personality dimensions. These findings show that data-driven insights can sharpen professional intuition in designing sound to convey brand personality.
- 26 October 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessWhat Does Sincerity Sound Like? Designing Brand Personality Using Audio
Brands increasingly use audio logos, short and distinctive sonic signatures, as central elements of their identity across advertising, digital products and services, and points of sale. Listeners infer a brand's personality from these few seconds of sound, so design choices shape how a brand is perceived. Yet empirical guidance about which sound features convey which personality is limited, and whether such guidance would improve on designers' intuition and experience is an open question. We compile a novel dataset of 696 audio logos and combine interpretable acoustic features, grounded in music theory and perception, with flexible machine-learning models to identify the sound properties associated with each of Aaker's (1997) five brand personality dimensions. Focusing on Sincerity, a dimension practitioners consider important but difficult to convey through sound, we translate these relationships into directional design insights (e.g., major mode is associated with greater Sincerity) and test them in a field experiment in which professional sound designers redesign unreleased audio logos to sound more sincere. Designers' unaided intuitions align with many of the identified relationships, but designers who receive the insights are more likely to adjust features in the recommended directions and make larger adjustments. Their redesigns are rated significantly more sincere than the originals and marginally more sincere than logos redesigned without the insights, with no detectable shifts in other personality dimensions. These findings show that data-driven insights can sharpen professional intuition in designing sound to convey brand personality.
- Open Access
- 3 August 2026
- Published by Elsevier in SSRN Electronic Journal
Consumer Value of Privacy: Evidence from an Online Retailer
We study how much consumers value personal information using data from an online retailer's tiered loyalty program. Consumers receive points for disclosing personal information. While points awarded are fixed, their effective value varies quasi-experimentally with distance to the tier-upgrade threshold. Exploiting this variation, we estimate how incentives affect disclosure, and recover bounds on the value of personal information. Disclosure increases with incentives, and valuations differ across data types. Consumers value mobile number ($0.99) and payment card details ($1.02) more than date of birth ($0.20). More engaged consumers value their information more, suggesting privacy valuations depend on consumer experience with firms.
- 3 August 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessConsumer Value of Privacy: Evidence from an Online Retailer
We study how much consumers value personal information using data from an online retailer's tiered loyalty program. Consumers receive points for disclosing personal information. While points awarded are fixed, their effective value varies quasi-experimentally with distance to the tier-upgrade threshold. Exploiting this variation, we estimate how incentives affect disclosure, and recover bounds on the value of personal information. Disclosure increases with incentives, and valuations differ across data types. Consumers value mobile number ($0.99) and payment card details ($1.02) more than date of birth ($0.20). More engaged consumers value their information more, suggesting privacy valuations depend on consumer experience with firms.
- Open Access
- 8 April 2026
- Published by Elsevier in SSRN Electronic Journal
Consumer Adaptation to a Currency Reform: Illicit Cash Use and Digital Payments
We examine consumer adaptation in payment use to a major currency reform, using India's 2016 demonetization as a natural experiment. Currency reforms that abruptly withdraw or redesign legal tender aim to curb unaccounted cash and promote digital payments, but they can also trigger hidden, illegal adaptation. India's reform, which invalidated 86% of circulating currency, created both a temporary shortage of legal tender and an urgent need for consumers to dispose of banned notes. Using detailed customer-level transaction data from a leading e-commerce platform and cross-regional variation in corruption, we apply a forensic economics approach to detect illicit use of demonetized cash in cash-on-delivery (COD) transactions, by colluding with corruptible delivery-agents. We do so by leveraging three sequential stages of customers' purchase journey: order placement, acceptance at delivery, and returns after acceptance. We find consistent evidence of illicit cash conversion, including an increase in the use of COD for high-price items, a higher acceptance rate of COD transactions in more corrupt regions, and a higher average return rate in COD transactions, during the 50-day conversion window. A latent-class analysis shows substantial heterogeneity: roughly half of customers adopted or increased digital payment use, while about a quarter circumvented the reform using illicit cash via COD. Thus, although the currency reform spurred digital adoption, its effectiveness in curbing unaccounted cash was weakened by illicit consumer adaptation.
- 8 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessConsumer Adaptation to a Currency Reform: Illicit Cash Use and Digital Payments
We examine consumer adaptation in payment use to a major currency reform, using India's 2016 demonetization as a natural experiment. Currency reforms that abruptly withdraw or redesign legal tender aim to curb unaccounted cash and promote digital payments, but they can also trigger hidden, illegal adaptation. India's reform, which invalidated 86% of circulating currency, created both a temporary shortage of legal tender and an urgent need for consumers to dispose of banned notes. Using detailed customer-level transaction data from a leading e-commerce platform and cross-regional variation in corruption, we apply a forensic economics approach to detect illicit use of demonetized cash in cash-on-delivery (COD) transactions, by colluding with corruptible delivery-agents. We do so by leveraging three sequential stages of customers' purchase journey: order placement, acceptance at delivery, and returns after acceptance. We find consistent evidence of illicit cash conversion, including an increase in the use of COD for high-price items, a higher acceptance rate of COD transactions in more corrupt regions, and a higher average return rate in COD transactions, during the 50-day conversion window. A latent-class analysis shows substantial heterogeneity: roughly half of customers adopted or increased digital payment use, while about a quarter circumvented the reform using illicit cash via COD. Thus, although the currency reform spurred digital adoption, its effectiveness in curbing unaccounted cash was weakened by illicit consumer adaptation.
- Open Access
- 28 January 2026
- Published by Elsevier in SSRN Electronic Journal
Customer Prototypicality and the Effectiveness of Segment-Level Personalization: Evidence from a Field Experiment
Customer segmentation is central to personalization, yet managerially important heterogeneity in responses remains within segments. With firms increasingly using customer embeddings as bases for segmentation, these representations offer a natural way to account for within-segment heterogeneity. Existing work has used embeddings to create micro-segments or to construct ad hoc similarity metrics, but such approaches lack a clear theoretical grounding. We propose a theory-driven, embedding-based measure of customer representativeness (prototypicality) and test whether it reliably accounts for heterogeneous responses to segment-level personalization. In collaboration with a retailer that uses embedding-based value segments, we conduct a large-scale randomized field experiment that orthogonally varies the message framing (symbolic "for-you" cues vs. generic wording) and recommendation method (algorithmic vs. popularity-based). Prototypicality systematically moderates treatment effects: among high-value customers, symbolic cues paired with algorithmic recommendations increase response for highly prototypical members but decrease engagement for peripheral members. A similar moderation pattern emerges under popularity-based recommendations, suggesting that the effects are unlikely to be driven by algorithm-specific idiosyncrasies. Finally, demographics and global similarity measures do not consistently reproduce these patterns, underscoring that within-segment representativeness is an important source of heterogeneity. Interpreted through a reference-dependent lens, our findings align with the idea that salient personalization creates expectations, yielding recognition gains when realized fit meets those expectations for prototypical customers, but misrecognition losses when it falls short for peripheral customers.
- 28 January 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessCustomer Prototypicality and the Effectiveness of Segment-Level Personalization: Evidence from a Field Experiment
Customer segmentation is central to personalization, yet managerially important heterogeneity in responses remains within segments. With firms increasingly using customer embeddings as bases for segmentation, these representations offer a natural way to account for within-segment heterogeneity. Existing work has used embeddings to create micro-segments or to construct ad hoc similarity metrics, but such approaches lack a clear theoretical grounding. We propose a theory-driven, embedding-based measure of customer representativeness (prototypicality) and test whether it reliably accounts for heterogeneous responses to segment-level personalization. In collaboration with a retailer that uses embedding-based value segments, we conduct a large-scale randomized field experiment that orthogonally varies the message framing (symbolic "for-you" cues vs. generic wording) and recommendation method (algorithmic vs. popularity-based). Prototypicality systematically moderates treatment effects: among high-value customers, symbolic cues paired with algorithmic recommendations increase response for highly prototypical members but decrease engagement for peripheral members. A similar moderation pattern emerges under popularity-based recommendations, suggesting that the effects are unlikely to be driven by algorithm-specific idiosyncrasies. Finally, demographics and global similarity measures do not consistently reproduce these patterns, underscoring that within-segment representativeness is an important source of heterogeneity. Interpreted through a reference-dependent lens, our findings align with the idea that salient personalization creates expectations, yielding recognition gains when realized fit meets those expectations for prototypical customers, but misrecognition losses when it falls short for peripheral customers.
- 16 December 2024
- Published by SAGE Publications in Journal of Marketing Research
- Vol. 62 (3) , 405-428
Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis
This article develops a method to automatically discover and quantify human-interpretable visual characteristics directly from product image data. The method is generative and can create new visual designs spanning the space of visual characteristics. It builds on disentanglement methods in deep learning using variational autoencoders, which aim to discover underlying statistically independent and interpretable visual characteristics of an object. The impossibility theorem in the deep learning literature indicates that supervision with ground truth characteristics would be required to obtain unique disentangled representations. However, these are typically unknown in real-world applications, and are in fact exactly the characteristics that need to be discovered. Extant machine learning methods are unsuitable since they require ground truth labels for each visual characteristic. In contrast, this method postulates the use of readily available product characteristics (such as brand and price) as proxy supervisory signals to enable disentanglement. This method discovers and quantifies human-interpretable and statistically independent characteristics without any specific domain knowledge on the product category. It is applied to a dataset of watches to automatically discover interpretable visual product characteristics, obtain consumer preferences over visual designs, and generate new ideal point designs targeted to specific consumer segments.
- 16 December 2024
- Published by SAGE Publications in Journal of Marketing Research
- Vol. 62 (3) , 405-428
Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis
This article develops a method to automatically discover and quantify human-interpretable visual characteristics directly from product image data. The method is generative and can create new visual designs spanning the space of visual characteristics. It builds on disentanglement methods in deep learning using variational autoencoders, which aim to discover underlying statistically independent and interpretable visual characteristics of an object. The impossibility theorem in the deep learning literature indicates that supervision with ground truth characteristics would be required to obtain unique disentangled representations. However, these are typically unknown in real-world applications, and are in fact exactly the characteristics that need to be discovered. Extant machine learning methods are unsuitable since they require ground truth labels for each visual characteristic. In contrast, this method postulates the use of readily available product characteristics (such as brand and price) as proxy supervisory signals to enable disentanglement. This method discovers and quantifies human-interpretable and statistically independent characteristics without any specific domain knowledge on the product category. It is applied to a dataset of watches to automatically discover interpretable visual product characteristics, obtain consumer preferences over visual designs, and generate new ideal point designs targeted to specific consumer segments.
- Open Access
- 30 June 2022
- Published by Elsevier in SSRN Electronic Journal
Automatically Discovering Visual Product Characteristics
Marketing models typically focus on how structured product characteristics impact consumer preferences. However, visual characteristics of products present in unstructured image data play an important role in impacting preferences for many categories. We seek to automatically discover and quantify visual characteristics (attributes) from image data using a disentanglement-based approach. While the deep learning literature has shown that supervision is required to obtain unique disentangled representations, ground truth visual characteristics are typically unknown. We develop a method that does not require such supervision, and instead uses readily available structured product characteristics as supervisory signals to enable disentanglement. Our method does not need prior knowledge of characteristics, yet we are able to discover semantically interpretable and statistically independent characteristics. Moreover, the method quantifies the levels of each discovered product characteristic, necessary for managerial tasks such as demand modeling and conjoint analysis. We apply this method to automatically discover visual product characteristics of watches, and discover 6 semantically interpretable visual characteristics providing a disentangled representation. Our results find the supervisory signal `brand' best promotes disentanglement relative to an unsupervised approach. We lastly demonstrate how consumers preferences may be assessed over these discovered visual characteristics using a choice-based conjoint analysis.
- 30 June 2022
- Published by Elsevier in SSRN Electronic Journal
Open AccessAutomatically Discovering Visual Product Characteristics
Marketing models typically focus on how structured product characteristics impact consumer preferences. However, visual characteristics of products present in unstructured image data play an important role in impacting preferences for many categories. We seek to automatically discover and quantify visual characteristics (attributes) from image data using a disentanglement-based approach. While the deep learning literature has shown that supervision is required to obtain unique disentangled representations, ground truth visual characteristics are typically unknown. We develop a method that does not require such supervision, and instead uses readily available structured product characteristics as supervisory signals to enable disentanglement. Our method does not need prior knowledge of characteristics, yet we are able to discover semantically interpretable and statistically independent characteristics. Moreover, the method quantifies the levels of each discovered product characteristic, necessary for managerial tasks such as demand modeling and conjoint analysis. We apply this method to automatically discover visual product characteristics of watches, and discover 6 semantically interpretable visual characteristics providing a disentangled representation. Our results find the supervisory signal `brand' best promotes disentanglement relative to an unsupervised approach. We lastly demonstrate how consumers preferences may be assessed over these discovered visual characteristics using a choice-based conjoint analysis.