ScholarSignal
Menu
Sign inEvaluate an idea

Evidence · Digital & platforms

Marketing analytics and AI: what peer-reviewed research shows

Research on marketing analytics and AI is dominated by frameworks and reviews, which map where data and AI change marketing work but offer few effect estimates. The empirical studies are narrower: machine learning can extract customer needs from user-generated content, consumers react differently to decisions made by algorithms than by people, algorithmic dynamic pricing lowers trust at first, and platform promotion decisions strongly shape what gets consumed.

Definition

Marketing analytics is the use of data and models to understand customers and allocate marketing resources, from personalisation to attribution. AI in marketing refers to systems that automate tasks, analyse data to reach decisions, or read and respond to human emotion. Both operate increasingly inside digital platforms whose algorithms decide what consumers see, what they pay and whom they deal with.

What the research finds

Frameworks map a changing field

Reviews organise how digital technology reshapes consumers, firms, channels, competition and the wider environment, and list open questions on search, social media, personalisation, attribution and privacy. Analytics in data-rich settings raises trade-offs between personalisation and privacy and requires data fusion, attribution models and new organisational capabilities. In retail, the move to omni-channel blurs the line between channels and touchpoints.[1][2][3]

What AI can take over

A widely cited framework separates mechanical AI (automating repetitive work), thinking AI (processing data to reach decisions) and feeling AI (analysing interactions and emotions), and maps each onto market research, segmentation, targeting and positioning, and marketing actions. It is a conceptual guide, not a test of results.[4]

Machine learning as a market research tool

A machine-learning approach applied to user-generated content identified customer needs more efficiently than relying on traditional interviews alone, making online reviews and posts a usable input for product research.[5]

Consumers judge algorithms differently from people

Across ten studies, consumers reacted less positively to a favourable decision made by an algorithm than by a person, while the gap shrank for unfavourable decisions. Consumers are also reluctant to accept AI in medical care. A meta-analysis of 943 effect sizes from 327 studies finds customers sceptical of robots, chatbots and algorithms but, in the end, choosing and buying from them much as from human agents, with conditions that differ by type of agent.[6][7][8]

Algorithmic pricing costs trust at first

In five studies, algorithmic dynamic pricing lowered trust in the retailer using it and lengthened price search, but the loss of trust faded as consumers got used to it, and search shortened once such pricing became the market norm. Price-matching can help retailers rebuild trust.[9]

Platforms shape what gets consumed

Inclusion in Spotify's own playlists has large, significant effects on the success of songs and artists, and the platform's playlists can be tested for bias towards particular content. Adopting music streaming sharply raises how much, and how varied, music people play, with effects that fade but persist after six months, and increases discovery of new music.[10][11][12]

How strong is the evidence?

The framework papers are influential reviews in leading journals, but they are conceptual: they set out questions rather than measure effects, and the studies reviewed here contain little direct evidence on the business returns of analytics or AI investments. Reactions to algorithms come mostly from controlled experiments, often with hypothetical decisions. The platform-power evidence comes from one industry, music streaming. Whether AI-driven personalisation weakens or strengthens the stable regularities in how people buy brands is an open question. Measurement of advertising effects is covered on the advertising incrementality page.

References

  1. P.K. Kannan, H. Alice Li (2016). Digital marketing: A framework, review and research agenda. International Journal of Research in Marketing.
  2. Michel Wedel, P. K. Kannan (2016). Marketing Analytics for Data-Rich Environments. Journal of Marketing.
  3. Peter C. Verhoef, P.K. Kannan, J. Jeffrey Inman (2015). From Multi-Channel Retailing to Omni-Channel Retailing. Journal of Retailing.
  4. Ming‐Hui Huang, Roland T. Rust (2020). A strategic framework for artificial intelligence in marketing. Journal of the Academy of Marketing Science.
  5. Artem Timoshenko, John R. Hauser (2019). Identifying Customer Needs from User-Generated Content. Marketing Science.
  6. Gizem Yalcin, Sarah Lim, Stefano Puntoni, Stijn M. J. van Osselaer (2021). Thumbs Up or Down: Consumer Reactions to Decisions by Algorithms Versus Humans. Journal of Marketing Research.
  7. Chiara Longoni, Andrea Bonezzi, Carey K. Morewedge (2019). Resistance to Medical Artificial Intelligence. Journal of Consumer Research.
  8. Katja Gelbrich, Holger Roschk, Sandra Miederer, Alina Kerath (2025). Automated Versus Human Agents: A Meta-Analysis of Customer Responses to Robots, Chatbots, and Algorithms and Their Contingencies. Journal of Marketing.
  9. Arnd Vomberg, Christian Homburg, Panagiotis Sarantopoulos (2024). Algorithmic pricing: Effects on consumer trust and price search. International Journal of Research in Marketing.
  10. Luis Aguiar, Joel Waldfogel (2021). Platforms, Power, and Promotion: Evidence from Spotify Playlists*. Journal of Industrial Economics.
  11. Luis Aguiar, Joel Waldfogel, Sarah Waldfogel (2021). Playlisting Favorites: Measuring Platform Bias in the Music Industry. International Journal of Industrial Organization.
  12. Hannes Datta, George Knox, Bart J. Bronnenberg (2018). Changing Their Tune: How Consumers' Adoption of Online Streaming Affects Music Consumption and Discovery. Marketing Science.

Bibliographic details only. Follow the DOI to read each study at its publisher.

Related topics

A synthesis of 12 studies, written by ScholarSignal. Last reviewed 2026-09-26.

Building on this evidence?

ScholarSignal checks a new research idea against the closest existing work and an editor rubric, and tells you whether the contribution is new enough for the journals you target.

Evaluate my idea for free