AI Shopping Recommendations and the Impulse Buying Economy

A good shopping recommendation solves a problem. It finds the right replacement part, narrows a confusing selection, or identifies something a customer actually needs. But the same technology can make an unnecessary purchase feel unusually convincing. As artificial intelligence becomes part of product discovery, the difference between helping someone shop and encouraging them to keep shopping deserves closer attention.

A new study in Sustainable Futures examines that tension through a survey of 642 Vietnamese online shoppers.¹ Researchers found that perceived usefulness, trust in recommendations, and emotional excitement were associated with stronger impulse-buying urges. Those urges were associated with consumption that respondents considered unnecessary or difficult to sustain.

For investors, this raises a question beyond whether AI can increase sales: can commerce platforms convert better recommendations into durable customer value? The opportunity is substantial, but measuring success requires looking beyond the next checkout.

What the AI Shopping Study Actually Measured

The researchers collected questionnaires between May 15 and 20, 2026, using online distribution and in-person surveys in Can Tho City. After excluding 59 responses for an invariant answering pattern, they analyzed 642 questionnaires. The reported sample comprised shoppers aged 18 to 33, making it especially relevant to younger consumers rather than the entire shopping population.

The study used the Stimulus–Organism–Response framework. In plain language, it examined how shopping experiences relate to what consumers think and feel, and how those internal responses relate to buying urges and consumption patterns.

Its shopping cues included interaction with livestream presenters, interaction with other viewers, perceived realism, and vivid product presentation. That context matters. Although framed around AI-targeted commerce, the research combines algorithmic shopping with the social and emotional features of livestream retail. It does not isolate a particular AI model or experimentally compare personalized recommendations with an otherwise identical shopping experience.

The following results come entirely from the paper. The coefficients describe associations within its statistical model, not percentage increases in purchases.

Unsustainable consumption here refers to self-reported behaviours such as purchasing without a clear prior need and shopping at a difficult-to-maintain pace. The study did not calculate emissions, examine household debt records, or track actual purchases. Its cross-sectional design cannot establish that AI caused the reported behaviour.

Why Useful Recommendations Can Encourage Impulse Buying

The most interesting finding is that usefulness and trust accompany impulsive urges. Persuasion does not necessarily depend on a shopper feeling deceived. A recommendation can seem relevant, credible, and convenient while still encouraging a purchase that was never planned.

Consider someone browsing for a winter jacket. A system might correctly suggest compatible boots. That suggestion may be useful, but usefulness alone does not establish whether the customer needs new boots or can comfortably afford them. Product relevance and purchase necessity answer different questions.

This distinction becomes important when companies train systems around observable actions. Clicks, cart additions, and purchases offer clear feedback. Whether an item remains useful six months later is harder to observe. A model rewarded primarily for immediate transactions could become excellent at identifying persuasive suggestions without learning which purchases deliver lasting satisfaction.

The absence of a significant direct relationship between perceived manipulation and buying urges also requires care. It does not mean manipulation never matters. Within this sample and model, recognizing persuasive intent was not reliably associated with the urge to buy. Disclosure alone therefore should not be assumed to resolve every behavioural concern.

Higher AI Conversion Rates Need Careful Interpretation

Commerce companies are already reporting measurable results. In its analysis of Q1 2026 activity, Shopify (SHOP ) reported that AI-referred shoppers converted at nearly 50% higher rates than organic-search visitors when comparing sessions beginning on product pages. Orders attributed to AI search also carried 14% higher average values.

Those figures indicate commercial promise, but they do not establish that AI created unnecessary spending. People who use an assistant to compare products may arrive with stronger purchase intent. Higher conversion could reflect better matching, more informed decisions, or a different mix of shoppers.

The survey and Shopify’s referral analysis answer different questions. One examines relationships among reported perceptions and urges; the other describes shopping outcomes across acquisition channels. Together, they support closer investigation of AI commerce, while leaving the quality and necessity of purchases unresolved.

Customer Trust Could Become a Commerce Advantage

A useful direction for retailers is to measure what happens after conversion. An extra order creates limited value if it brings an expensive return or discourages the customer from coming back. These are business implications to investigate, rather than outcomes demonstrated by the Vietnamese survey.

Retailers evaluating recommendation systems could pair sales metrics with:

• Return and cancellation rates for recommended products.
• Repeat purchases and customer retention over time.
• Customer satisfaction after delivery and use.
• Contribution margin after fulfillment and return costs.

This broader view connects with Securities.io’s recent examination of how trust influences automation returns. The commercial setting differs, but both topics show why human responses deserve attention alongside technical capability.

Responsible design can also preserve useful personalization. Budget filters, clear explanations of recommendations, and options to save suggestions for later could help shoppers make deliberate decisions. The paper recommends transparent personalization and reflective purchasing interventions, although it does not test their effectiveness.

Shopify Offers Exposure to AI Commerce Infrastructure

For investors interested in this development, Shopify provides a relevant public company to examine. Its connection extends beyond adding a chatbot to a storefront: it is building recommendation technology and infrastructure that connects merchants to emerging AI shopping channels.

In February 2026, Shopify described its generative recommendation system, which learns from sequences of searches, views, cart additions, favourites, and purchases. Timing helps the system account for changing preferences and seasonal context rather than treating each interaction independently.

Shopify reported that online A/B testing of an August model version increased Shop orders by 0.94% and conversion rates by 0.71%, both relative improvements. These production results provide a more direct test of a recommendation change than the Vietnamese survey, although they do not establish long-term satisfaction or environmental benefits.

In March, Shopify announced wider access to Agentic Storefronts, connecting merchants with AI channels including ChatGPT, Microsoft Copilot, and Google’s AI shopping experiences. This creates an investment angle around the infrastructure supporting discovery and transactions, even as consumer interfaces evolve.

The company was not evaluated in the academic study. Its relevance comes from its technology and market position. Investors still need to assess valuation, competition, operating costs, and whether AI activity generates incremental profitable business rather than shifting purchases between channels.

Latest Shopify (SHOP) News and Developments

The Next Test for AI Shopping Is Lasting Value

AI recommendations can reduce search time, introduce shoppers to overlooked products, and help merchants reach suitable customers. The study highlights why those benefits can coexist with stronger impulsive urges, especially when trust and emotional engagement reinforce one another.

The constructive response is to improve what systems optimize and what companies measure. A recommendation that produces a satisfied returning customer may be more valuable than one that simply produces an immediate sale. For the next generation of commerce platforms, proving that distinction could become an important competitive advantage.

References:

¹ Cao, L. T., Tran, T. T. B., Bui, K. T., Luu, T. T., & Nguyen, L. T. (2026). Dark side of AI in online shopping: Does algorithmic targeting increase unsustainable consumption. Sustainable Futures, 12, Article 102123. https://doi.org/10.1016/j.sftr.2026.102123