Levi’s virtual assistant – Source: Levi Strauss & Co.
How open are consumers to AI-driven shopping?
According to a study from The Integer Group, 78 percent of U.S. consumers feel “curious” about using artificial intelligence (AI) technology to shop, but 66 percent also said they were “cautious” about using it, as well.
One reason for the caution is the perceived threat to privacy. Asked to rank their top three concerns (from a list of seven) when it comes to AI in the future, 71 percent pointed to the protection of their personal information as being among their top three. That includes 41 percent who ranked it as their top concern, easily ahead of all other issues.
Only about half (52 percent) were open to sharing their past shopping history to power AI. That drops to about 25 percent for those willing to share social media page and profile information and slightly lower for sharing personal information (age, household income, etc.).
Integer Group wrote in the report, “It seems shoppers are inflexible or don’t realize that data fuels AI.”
Indeed, the report found many consumers remain confused about what AI is. The majority didn’t know online suggestions, search engine results and customized online ads were influenced by AI. Integer noted that, despite the privacy concerns, customers have widely accepted Google’s use of passive data collection for online browsing.
Beyond cost and accessibility, other primary concerns about using AI for shopping included its usefulness and effectiveness.
The study, for instance, found that respondents using virtual assistants such as Alexa and Siri only used them for a handful of simple tasks, like playing music or answering cooking questions. Using AI for tasks such as picking out clothing, automatically ordering dinner or purchasing groceries were seen to be “more of a hassle than a benefit.”
For mainstream shoppers, the study points to an opportunity to improve adoption by using AI to assist with the tasks seen as least enjoyable, such as shopping for everyday household items.
“However,” wrote Integer in the study, “the actual shopping decisions respondents said AI could do for them were quite personal and distinct. For instance, a shopper might say she would like AI to choose her produce, but not her shampoo.”
Discussion Questions
DISCUSSION QUESTIONS: Should retailers and brands focus most on privacy concerns, usefulness or effectiveness in efforts to spur mainstream adoption of AI for shopping? What conclusions do you draw from the findings that consumers are only using virtual assistants for simple commands?
Poll
BrainTrust
Zel Bianco
President, founder and CEO Interactive Edge
Jason Goldberg
Chief Commerce Strategy Officer, Publicis
It is incumbent on retailers to make customers feel comfortable about privacy issues regardless of the reason. AI can help personalize a shopping experience, suggesting merchandise that should interest the consumer based on past shopping patterns. Done well, with a customer who already trusts the retailer, it will only help create a better CX that potentially leads to more business.
As for consumers using virtual assistants for only simple commands … it’s because they don’t yet know the power of AI and the virtual assistant. It’s only a matter of time before they do, and then it will be as common as an airline passenger who books their ticket and checks in for a flight online, never talking to a human until they finally get to the airport on the day of departure.
Surveys like this don’t amount to much of anything. People that are not fluent/knowledgeable in the technology being surveyed often give uninformed answers or answers that they believe the pollsters want to hear.
AI, like many technologies before it scares users that are unfamiliar with it at first and in time they accept, adapt to and even seek it out. The key (as with anything) is utilizing AI in a manner beneficial to users, is reasonably transparent, that doesn’t abuse their trust, and ideally gives them control as to how much or how personal data is utilized or shared.
I suspect that most consumers are concerned about privacy as it relates to AI. That is the number one issue. But I also suspect that Millennials would less concerned than older shoppers.
Whether or not we are aware of it or not, AI is already being seamlessly integrated into not only the voice assistants but also our every day search and online shopping experiences. However, as we all desire personalized experiences, AI is a key component of that equation and retailers have to be very transparent about their privacy policies to ease any concerns.
In short, data and insights are the critical components of an enhanced customer experience. Our expectations for when we shop or interact with a brand online or in stores, is that they “know us,” and there is no need to request any more details via surveys or questionnaires. Ultimately, the user adoption of AI-powered voice or online commerce experiences will increase once the trust factor has been established.
I don’t think it’s important that consumers understand AI, what it is, how it relates to data or any of the details. What AI promises is to create more satisfying shopping experiences by leveraging a variety of data sources and machine learning. Retailers and brands should focus more on use cases and business outcomes first, then seek the options to test and rollout the right supporting technology solution.
It’s important to keep in mind that consumers must realize more value from technology driven interactions than their concerns about data privacy or ease of use. The use of virtual assistants for more than simple commands will happen, but to scale these interactions they must allow consumers to easily confirm their inputs, make changes and understand how payments work. For this reason, I think virtual assistants that include visual interfaces are probably going to be necessary to see more complex use cases work well at scale.
I don’t know whether this will still be true in time, especially if AI is used in a more gradual way at first. I think that is the key here. Yes perhaps younger, more tech-savvy consumers will adopt it earlier. Many already have. To get consumers other than early-adopters, slow and gentle is the way to go.
I think this is really a consumer education play. Shoppers are already being “guided” by AI agents now they just don’t recognize it. So I think the privacy issue should be incorporated into a general educational effort stressing fundamentals — what AI is and isn’t, its advantages and, of course, caveats such as discussions of how to manage privacy issues.
Privacy concerns need a risk/reward context. Would most consumers stop using Amazon, for example, just because an AI engine is recommending what they should read, listen to, watch and buy? As for the second question, AI and voice-activated agents are still new. Check back in five or 10 years and we’ll wonder why we asked this question.
The train has left the station in terms of consumer acceptance of data use for promotional marketing. Brand and retailer caution is well within the realm of acceptable behaviour generally, but retailers have to be diligent about hacking and data theft, which consumers can be justified in being concerned over. My Christmas wish is that no retailer suffer data losses as this erodes overall consumer confidence in technology-enabled commerce.
If consumers do not see a real value in using AI and know where their data is going with an assurance of privacy, it will not take off. Whether or not people understood how to use email, texting or sending pictures, they figured out how to do it because they wanted to use it. The same will be true with AI. They will have to want to use it. Before using it, they are likely to question privacy.
Great AI technology that matters shouldn’t be noticeable to the consumer being served. And if it IS noticed, then the retailer has made a big mistake.
As a result, this survey isn’t of meaning. It always amazes me how often people think good market research among that incredibly subtle world of “consumers” should directly ask questions. That’s the clearest way to know the research isn’t accurate — it just asks questions like “how do you feel about using AI?”
What I’ve always found is that mediocre research (research with the trappings to appear excellent but which doesn’t seek answers in the right way) is the biggest danger to corporations. Bad research stands out. Mediocre research appears meaningful but isn’t.
AI is so embryonic and the concept so nebulous that it is hard for consumers to predict how they might use it in the future.
All I would say is that AI will complement existing processes and decision-making rather than replace it entirely.
These kinds of studies are silly. First of all, “stated” preference studies for shopping behaviors are almost always wrong, as most shopping decisions are made continuously. For example, Admaster does a poll of Chinese shoppers before Singles Day every year. Per the “stated” preference survey 84 percent of shoppers planned to shop on Singles Day in 2015, only 71 percent in 2016 and only 64 percent in 2017, so Singles Day should be shrinking every year. Yet Singles Day grossed $14.3 billion in 2015, $17.8 billion in 2016 and $25.4 billion in 2017.
Even worse, in this case the study asks consumers to evaluate a tactic they can’t even see/experience/consume. How does a consumer know if the great outfit in their latest Stitch Fix box is the result of a stylist that really knows her or an AI algorithm?
Lastly, the definition of AI is constantly evolving. Aren’t all the suggested products on the Amazon and Tmall PDPs actually generated by AI?
AI, and more specifically deep learning, is a game-changing capability for commerce. But it’s a back-of-the-house tactic, not a customer-facing experience.
Jason, you are so right about lack of rigor in “stated preference” research. Consumers are rarely in a position to answer accurately about their intent — especially regarding a consumption method that is new or untried.
Population experience with cognitive technologies is still very limited, and the privacy-personalization tradeoff remains unclear, even after 20 years of speculation and experimentation.
But anxiety about personal privacy is real, and the risks are underscored many times a day for each of us — whether in news stories about data breaches or in the online banner ads that seem to stalk us across the internet.
I’d argue, however, that AI cannot and should not be invisible and behind the curtain. Rather, I’d advocate a principle of complete transparency, revealed not in the depths of “terms of use” statements, but reinforced in the user interface itself. Ideally, every application of AI must answer this question to qualify: “How does this innovation give customers more confidence and control?”
AI is a technology and, as is the case with all technologies, should be implemented as an enabling vehicle — not as an end in itself. While privacy should always be respected, the focus should be on providing the shopper with the most valued and frictionless shopping journey — period. If AI can facilitate that then all of the secondary issues, concerns and challenges will melt away.
Assuming the retailers and brands who can afford to invest in AI will be the same that want to go global then, with the recent change in EU law, they will have little choice. While the change in EU law may not be relevant to some, no developed nation seems to be easing off on the use of personal data so brands and retailers have to put in the effort to protect consumers.
Consumers like retailers and brands are new to AI. Educating in the art of the possible with AI is a key part of any development in this area. Most of the tasks listed that consumers do are the very tasks shown in ads and introductory videos. Providers need to take them on the journey of learning this new capability.
Retailers are in the business of helping their customers achieve validation through the foods they eat, the gadgets they use, the clothes they wear and the products they buy for their living areas. While AI is prominent in all business conversations, that is not a topic that consumers are interested in when they go to make a purchase. Consumers are interested in the final outcome around value, uniqueness and experience.
Any discussion around AI that a retailer has with the consumer detracts from the real conversation they need to have about the consumer: their aspirations, dreams and fears. Retailing is about emotional connections and not about a discussion over technologies and analytics.
As to privacy, the more that trust is built between the consumer and the brand, the less this is an issue. I am willing to share more with a brand that has earned my trust versus one that I’m not familiar with. In itself, privacy becomes a hurdle to entry in the new data-driven world of retailing. That doesn’t mean that the retailer-consumer conversation is about privacy but it’s about establishing trust through multiple interactions.
We have had privacy concerns every time some new technology came to be. Each time we voiced and wrote our concerns. And each time we have become comfortable using the new technology. AI will be the same. Maybe we will continue with our concerns for another year; but after that we will get more comfortable with less worries about privacy.
While the Integer survey noted protecting personal information was the #1 preference for AI among shoppers, making things easier and lowering cost were #2 and #3. AI that keeps the user in control is most likely to be successful. This means:
As much as users say what they think about AI today, expect attitudes to change as retailers inevitably adopt more of it and shoppers realize it’s opt-out rather than opt-in.
In our own study on AI, 41% of over 3000 consumers tested gave AI the top two boxes for “increased chance of purchase with use of AI” — so, they’re ready.
It’s clearly shopping of the future, so incredibly convenient. It reminds me a lot of “one click” and the first time I used that: wow! Once you just start telling a box to buy you something or to get you the best price or where something is, you just can’t stop using it. It also makes the old method of shopping (you know, ecom, ha) seem ancient and clumsy. Look it up myself? Are you kidding?
Funny related side bar: heard a comedian tell a story about driving and missing the exit he was supposed to get off on. His GPS politely/calmly told him, “you missed your exit, re-calculating.” To which the comedian thought, “when you ride with people, could you imagine them politely/calmly telling you that?” Good point. Same with AI buying; there are no mistakes, you’re the boss.
Let’s face it, most consumers are benefiting from AI today, but they may not call it that or understand that is what is powering an aspect of their shopping. As long as AI offers value and builds trust with consumers, most will appreciate better selection, easier shopping and helpful advice.
Surveys that ask shoppers if they’re willing to share personal data invariably show that few shoppers are willing. However, reality shows that if the right, compelling offer is presented to the shopper, they’ll give up most anything to get a good deal.
As more and more retailers employ true AI in their shopping experience, more shoppers will get comfortable with it. AI is only as good as the info it’s fed, so the more shoppers who use it, the better it becomes, and the self-fulfilling prophecy of ubiquitous adoption will ensue.
Consumer adoption of new things is a value exchange. As soon as the value proposition is right for a critical number of people it starts driving a culture shift. Privacy is cost of entry these days for everything and needs to be in place. Retailers should focus on the key drivers of broader adoption: usefulness, effectiveness and making sure consumers understand how it all works.
Consumers don’t need to know whether AI is behind their shopping experience or not. What they need to see and feel is convenience, understanding of context, and assistance. These will be brand differentiaters for retailers.
For example, I was recently shopping for wheels for my car. After a few clicks an AI system with picture and attribute recognition should have been able to determine the style and size I was looking for, and the many, many styles and sizes I was not.
If a website had this technology, and quickly began filtering my search for me, presenting me with choices consistent with my needs and preferences, it would have saved me time and demonstrated intelligent, personalized understanding of my needs. This benefit would have been noticed and appreciated, increasing the odds of conversion, and fostering loyalty.
AI processes the same data currently collected by retailers like consumer demographics, purchase history, and various inferred user behavior data, etc., to offer up “AI” recommendations. Same as current and past shopping applications apply the same collected consumer data through older systems of collaborative filtering, segmentation, inferred behavior, etc. AI cannot be “spurred to be adopted” just because it is AI. AI is not for the consumer to decide. It is the retailer who must invest time and expertise in understanding how they design and build an AI system to maximize the value for each of their individual customers, based on the retailers’ brand, business model, and products they sell. Knowing this, the developers writing the code for an AI application MUST know exactly the outcomes the retailer is seeking. In coding an AI solution, by virtue of being coded by a human, subjectivity is introduced [for shopping]. Each retailer must spend the time and resources to ensure the code in their AI solution is relevant to their customer, brand and product lines.
The question of why customers/shoppers only using Virtual Assistants for simple commands? Pretty simple. Order more toilet tissue, Alexa. Versus. Alexa, order a pair of cropped pants with a slight bell at the hem, slim through the knee and high waisted, but not too high waisted, in a soft denim. Alexa, you know how I like my pants and jeans to fit, with extra ease on the hips. And “oh yeah Alexa” make sure the pants do not touch my ankle, I like them shorter, but not too short! Alexa, you are a doll!
The first part of the question seems out of place. Retailers should not be spending any time trying to spur mainstream adoption of AI for shopping. They should be focused 100% of their efforts on solving consumer needs and if AI can help (which I believe it does and will even more in the future), then they should test and find best uses of it. We are just at the start of the voice era. It took e-commerce a solid decade for any real adoption. Like e-commerce, it will start slowly with minimal use and adoption but as it gets better and when it solves a need more efficiently or with better results, we will use it more and more and it will, by the nature of the increased use and AI, get better over time.
Transparency is crucial to any AI effort by retailers or brands. The minute shoppers think you’re trying to pull a fast one with their personal data, trust starts to erode. Conversely, if you’re upfront with the customer and present offers that help them shop better – that could mean price, value, assortment, service and more, depending on the individual or segment – then you can quickly build trust. Bottom line is adhere to the Golden Rule and you likely won’t go wrong.
I’m not surprised by the lack of understanding of AI, or the caution people feel. I think it’s really easy to forget that clothes shopping is a very personal experience, often driven by quite strong emotions.
While customers are happy to shop online, and value the convenience, it still doesn’t compare with the all-round sensory experience you get when shopping in a physical store in the high street. Many people still hanker for the intimacy of the physical store, and in fact many are just browsing online before they go to the stores in person.
I think that’s part of the reason why people don’t fully engage with retail AI – they are happy to use it for practical, functional tasks but they don’t really trust it. They still equate AI with a faceless machine or megacorporation that doesn’t have their best interests at heart.
If retailers want to improve engagement with AI they need to ensure customers see it as more human than machine – and that goes for every industry hoping to disrupt B2B using AI.