How Voice-assisted Shopping Drives Up Sales

June 21, 2021

Voice-assisted shopping is a relatively new way to purchase products, and so is the research into its effectiveness. A new MSI working paper finds that consumers who use voice AI tend to browse 13.6% more and spend 19.5% more, which corresponds to an extra $493 million in sales revenue every year. Those are eye-popping numbers that reveal the promise and potential for this emergent technology, especially for firms that are considering its adoption.

“To our best knowledge, our paper is the first to provide a comprehensive analysis on the impact of voice assistant adoption on consumers’ purchase and browsing behaviors,” wrote the co-authors of the study, which relied on data collected from nearly 40,000 random customers and 5.59 million transactions on Alibaba, the world’s biggest e-commerce company.

The study is titled, “The Effect of Voice AI on Consumer Purchase and Search Behavior.” The co-authors are Chenshuo Sun, doctoral candidate at New York University’s Stern School of Business; Anindya Ghose, Heinz Riehl chair professor of business at NYU Stern; Xiao Liu, assistant professor of marketing at NYU Stern; Zijun (June) Sun, assistant professor of marketing at Hong Kong University of Science and Technology; and Xueying Li and Feiyu Xiong of Alibaba Inc.

MSI asked Ghose and Sun some questions about the study and its implications. Their answers appear below: 

MSI: Why did you want to study this topic, and what does your paper add to the literature?

Sun and Ghose: The original idea was not about voice shopping. While we were examining another dataset on couponing, it was just happenstance that we found that Tmall Genie [smart speaker system] adoption on Alibaba positively affects consumption. We felt that this is a truly important question for both academic research and practices that would yield useful insights for business, so we decided to work on this topic.

Our work builds on and extends several streams of research. First, it is related to the growing body of literature on how voice assistants affect consumers and service providers. More broadly, it also extends the line of literature that examines the impact of emerging technologies on consumer behavior and channel attribution. The literature review section of our paper has elaborated on the academic contribution.

MSI: What are the key takeaways for marketers and managers?

Sun and Ghose: There are several key takeaways:

  • Adopting voice AI leads consumers to purchase more and browse more.
  • There is a substantial heterogeneity in the treatment effects.
  • The voice channel does not cannibalize other channels; rather, it boosts the sales through the PC channel.
  • The potential mechanism by which voice AI adoption boosts purchase can be due to reduced search cost.


Our research speaks to the e-commerce firms on how they can capitalize on the emerging voice AI technology. E-commerce firms may want to develop their own voice shopping assistant for stimulating higher volume of sales and search. Moreover, according to our findings on the heterogeneous treatment effect of the voice AI adoption, managers may want to promote the voice AI device to consumers who tend to spend, purchase or search more after using the smart device.

Our results also show that the voice shopping channel has a positive spillover effect on the PC and mobile channels in terms of the spending amount. One possible cause of this complementary relationship could be that consumers use the voice AI and other channels together to make a purchase. Managers may want to keep this positive spillover effect in mind when trying to engage consumers through platform design. For example, e-commerce sites can keep records of where consumers ended in the voice search channel, and then show them to consumers when they open up the site on mobile or PCs.

Finally, our findings indicate that the adoption of the voice AI can induce larger increases in sales and search for categories with high purchase frequency or low substitutability. There are two immediate implications to the e-commerce firm. First, marketers may want to promote the voice AI together with product categories that are frequently purchased or of low substitutability. Second, the firm can try to target the voice AI to consumers who buy a lot from product categories with high purchase frequency or low substitutability. By doing as suggested, managers can save on marketing spending and achieve a higher return of investment.

MSI: Your paper finds that the voice channel has a “positive spillover effect” on other purchasing channels. What does that mean?

Sun and Ghose: That means this newly introduced shopping channel does not cannibalize the existing, conventional purchase channels, such as the PC and the mobile channels. This is actually very good news for the digital platforms that have launched the voice channel, because now they know that introducing the voice channel will help make their cake, meaning the total revenue, bigger and bigger.

MSI: You found some interesting differences between men and women, and among various age groups. Those under 25 were most likely to adopt voice-assisted shopping, while men were more likely to purchase. What explains these differences?

Sun and Ghose: Thanks for catching the nuance. The fact that those under 25 were more likely to adopt voice shopping is just a summary statistic of the data, whereas younger consumers and males being more likely to purchase more as a result of adopting voice AI is a causal effect. In general, younger consumers are more likely to be early adopters of new breakthrough technologies, and it was nice to see this insight being corroborated from our research.

MSI: Your study points out that voice-assistant shopping activity is highest right after the consumer purchases the device. Does that finding indicate that it’s just a fad?

Sun and Ghose: What we find in this research, using a 31-week dataset that includes a 17-week post-treatment period, is that even though the positive effects attenuate over time, they still remain statistically significant after eight weeks. This indicates that consumers will continue using the voice assistant for shopping. In other words, it is not just a fad.

MSI: Voice-assisted shopping has great potential, but it can be costly for firms to invest in this technology. What kinds of companies would gain the most from developing their own voice AI?

Sun and Ghose: It could be costly, but during the past years we are seeing an increasing number of voice AI products. Almost every big tech company has invested in and developed its own device. Examples are ample, like Amazon Alexa and Echo, Google Home and Apple Homepod. These are very famous products in the U.S. In China, millions of consumers have adopted Tmall Genie, Baidu Xiaodu and Xiaomi Xiaoice.

Currently, online digital platforms like e-commerce businesses may benefit the most as consumers can use their voice assistants for shopping purposes, which creates values directly for the platforms. This is the so-called voice commerce. Looking ahead, we believe that the evolving technologies in the next few years, such as better Internet of Things and mood detection, will further improve the voice commerce experience.

What’s more, the COVID-19 pandemic has fueled an increased demand for contactless shopping coupled with greater use of voice AI. We believe that such a hands-free and interactive way of shopping will have great potential in the future.

MSI: This field of research is rapidly evolving and ripe with possibilities. What do you hope to study next?

Sun and Ghose: During the past decades, technological innovations and digitization have profoundly changed the digital economy. Emerging technologies, such as the chatbots that we are studying here, along with virtual reality, live streaming and disruptive digital infrastructures are empowering a paradigm shift of academic research as well. For this voice-shopping project, we plan to study how we can design a “smarter” voice AI chatbot that can make consumers’ journeys smoother by modeling the communication costs from both sides, the sender and the receiver.

Meanwhile, we happen to be working on a new project. We [Sun and Ghose] are interested in how 5G is going to transform the digital economy.  Stay tuned!

By using you agree to our use of cookies as identifiers and for other features of the site as described in our Privacy Policy.