©B4LLS via Canva.com | Credit: Amazon
Amazon Looks To Elevate Discovery With AI-Driven Chatbot
Amazon is launching its first AI-powered shopping assistant to empower mobile shoppers to search by query with answers delivered in a conversational context.
Called “Rufus” after the dog of two early employees, the chatbot is based on a large language model trained on the company’s product catalog, customer reviews, community questions, and “information from across the web.”
Under its traditional search bar, shoppers manually type keywords or product names, seeking either a specific item or looking for one based on its features.
Rufus lets customers ask specific questions but also those that are vaguely worded to better support discovery. Customers can learn what to look for in product categories, shop for special occasions, get help comparing product categories, get recommendations, and ask specific questions about products.
In a LinkedIn post, Amazon CEO Andy Jassy offered a number of examples of questions the AI bot can handle:
- “What is important to consider when buying headphones?”
- “What’s the difference between trail and running road shoes?”
- “What are the best Valentine’s Day gifts?”
- “Is this pickleball racquet good for beginners?”
“Rufus lets customers ask shopping journey questions like, ‘What is the best golf ball to use for better spin control?’ or ‘Which are the best cold weather rain jackets?’ and get thoughtful explanations for what matters and recommendations on products,” Jassy said last week on the retailer’s fourth-quarter analyst call.
He explained, “You can carry on a conversation with Rufus on other related or unrelated questions, and it retains context coherently. You can sift through our rich product pages by asking Rufus questions on any product features, and it’ll return answers quickly.”
To use Rufus, customers start typing or speaking their questions into the search bar in Amazon’s mobile app, and a Rufus chat dialog box appears at the bottom of their screen. Customers can expand the chat dialog box to see answers to their questions, tap on suggested questions, and ask follow-up questions in the chat dialog box.
Customers can dismiss Rufus to return to their traditional search results by swiping down to send the chat dialog box back to the bottom of their screen. Amazon CFO Brian Olsavsky said on the analyst call, “The way we’re rolling it out is that it is additive to the current shopping experience.” He also said that customers can still search in the search bar if they know what they want.
If Rufus gains traction, Amazon could play a bigger role in the research phase of the shopper journey as well as “take ad sales away from Google and social media sites, where companies try to influence what customers decide to buy,” according to the New York Times.
Rufus became available in beta to a small group of users in the U.S. on Feb. 1, with a broader rollout planned in the coming weeks.
A PowerReviews survey of about 8,000 U.S. consumers conducted in March 2023 found 50% starting their online shopping journey at Amazon, followed by Google at 31.5%. For Gen Z respondents, Google bested Amazon (38% versus 36%, respectively). Gen Z was also the most likely group to start their shopping journey on social media (5%).
Of all respondents, 56% consider search results (on Google and other search engines) when making purchase decisions, with 63% saying they use search engine results to learn about products they haven’t purchased before.
Last spring, Microsoft and Google released chatbots and AI tools for their search engines, highlighting shopping-related uses. At the start of this year, Walmart unveiled its own generative AI shopping assistant in partnership with Microsoft.
Discussion Questions
How much of a threat is Amazon’s Rufus to Google and social media sites that support discovery?
Does the shopping assistant appear to complement and augment the Amazon shopping experience?
Poll
Loading …BrainTrust
Ryan Grogman
Managing Partner, Retail Consulting Partners (RCP)
Scott Benedict
Founder & CEO, Benedict Enterprises LLC
With a burgeoning product catalog, search is one of the most challenging things for Amazon to get right. Rufus is an attempt to use AI to make the process of finding the right products easier. Rufus is essentially an intelligent inference between the consumer and Amazon’s store that will allow consumers to hone in on what they want in a more efficient way. Rufus also adds an element of customer service as it will answer questions for shoppers, helping them to make more informed decisions about what to buy. This is very early days for Amazon in AI, but this has the hallmarks of something that could solve a real problem and be very useful for shoppers. It will also be interesting to see if and how Amazon integrates this with Alexa devices.
How much of a threat ?? That’s something best answered by users telling us how much they like it; but my guess is that – initially at least – the answer is “small” (actually if Amazon overhypes/underdelivers the answer may end up “none”) Obviously one advantage Amazon will enjoy is that search and purchase can essentially be combined in one step, tho how big of an issue that is I don’t know; and I think there will be concerns, among some , at least, about the results being biased. But, again, it’s hard to judge how much of an issue that is.
Amazon has already implemented generative AI to condense customer reviews, providing a time-saving and competitive advantage, particularly for products with tens of thousands of reviews. Navigating Amazon’s extensive product catalog can be daunting, and incorporating generative AI tools to streamline the consumer experience is a logical next step. The impact of this new tool on Google and other platforms remains to be seen. However, considering Amazon’s renowned reputation for competitive pricing, rapid shipping, and commitment to cutting-edge technology, they are well-positioned to stay at the forefront of innovation.
Given how much Amazon is currently relied upon for product search, making it better with its Rufus shopping assistant should only improve the experience. And if this happens, then it could eventually impact how frequently users visit other search sites. As we are already seeing with the integration of ChatGPT and Microsoft’s Bing search engine, it’s refreshing to get answers versus scrolling through countless pages of irrelevant or mistargeted, marketing-driven results. If Rufus really does deliver on the benefits Amazon is claiming, then this will be reflected in Rufus usage stats. And if it’s really effective, Amazon might even see an increase in their conversion rates as shoppers are better able to find exactly what they are looking for.
Anyone that discounts the impact that these types of advances will have on things like search, Amazon and otherwise, is not understanding the gravity of these tools. I’ve switched to using Perplexity for search and it provides insight rather than just the usual jumble of pages. With the amount of information to sort through on Amazon, this will likely make searching for products much easier and more efficient.
Thanks for highlighting Perplexity. I just tried it, and its approach to my question was more encompassing than my regular Google or Bing.
Glad you tried it and found it useful! Just FYI, there are also two Chrome plugins for Perplexity that allow you to replace Google as your search bar default if desired.
Thanks.
I think Amazon’s Rufus poses a significant threat to Google and social media platforms in terms of supporting discovery during the shopping journey. By offering a conversational AI shopping assistant directly within its mobile app, Amazon aims to capture users’ initial research phase, potentially diverting ad sales away from competitors.
Additionally, Rufus appears to complement and enhance the Amazon shopping experience by providing personalized recommendations and detailed product information, making it easier for customers to find what they’re looking for. Overall, Rufus represents a strategic move by Amazon to further solidify its dominance in the e-commerce landscape.
This sounds amazing, but I am immediately suspicious of the methodology. Can answers to question be bought…be put up for bid…??? Can brands and retailers buy their way to the front of the line? If I ask, “What is the best widget for…?” what are the variables that go into the answer? “Best” according to what metrics? There’s a big difference between serving up facts and product attributes and serving up opinions. Product reviews will naturally be included, but hopefully they will be given the proper context.
Exactly. It’s the economy of the ads…..
We are in early days here. Exhibit A is the “chatbot” experience, set up to answer simple questions like “are you open” or “how can I pay my bill”…never to answer something meaningful (which is why you need to chat with a real person anyway). It will be a long time before Rufus is really useful.
You read it here first!
In the initial launch, it’s a minimal threat. But it was inevitable that Amazon expand it’s current AI capabilities into the product search and research space, so it’s a necessary step for them. They are an innovative technology company and I have no doubt they’ll gain learnings quickly in order to evolve Rufus into a more efficient tool to drive higher conversion.
This fits in well with Amazon’s core value prop, enabling customers to find products across a vast catalog of available items. Anything that can curate their inquiries will be a win for the customer experience and for Amazon. Caveat- the results need to be accurate. That’s been a bit of an issue for Amazon in the past.
The threat to Google and other social media sites… I don’t think it will be a huge threat. Consumers who start their product searches on Amazon now have a new tool to get them to what they’re looking for faster. For consumers who look on other platforms, you would have to understand their motivation for doing so. Amazon’s results are probably not a big motivation to change behavior. Customers who specifically look elsewhere are also not likely to stop that behavior, their motivation is more likely to be “what other than Amazon is out there.”
The addition of Rufus will make even more critical to find ways to mitigate fake reviews and other tactics for product manufacturers to manipulate the search function.
Am I the only person in the world that hates chatbots?
I don’t feel like they can do product discovery faster and better than a human, and I don’t want to be their learning vehicle.
This is a little bit like gilding the Lily. One also wonders if the chatbot recommendations will come before sponsored suggestions. I think not.
this is at best, a business model at war with itself.
If you read me regularly, you know I am an Amazon fan. When I am buying, Amazon is the first place I go.
However, if I want to research, i.e., What is the best? Most convenient? Most unique? I go to Google and get several review sites to check out. I don’t imagine the Amazon AI will provide that breadth, and I do imagine that just like people complain about Amazon’s favoritism, this will take complaints to a higher and louder level.
Once again, Amazon is pushing the technology envelope by providing personal shopping assistance on what used to be a static website. This is eCommerce personalization at a new level. I’m excited to try it out. Valentine’s Day is just around the corner and I’m looking for the perfect gift for my wife! 🙂
As for the question about Amazon being a threat to Google, there will always be competition. Amazon sees competition with Walmart, Target, and other big retailers trying to catch up to Amazon. For Google, there are plenty of other technologies and platforms that Google can and will consider a competitive threat, such as ChatGPT, Bing, etc.
As we witnessed at the latest NRF Big Show, GenAI capabilities are now embedded within the customer and store associate solutions to unlock value, driving efficiencies and outstanding customer experiences. We increasingly see intelligent GenAI chatbots and capabilities surface across search, discovery, and commerce sites.
Considering Amazon’s vast assortments and the challenging discovery experience, Rufus’s GenAI-backed capabilities can make this experience more seamless for customers. Until now, customers have had little to no interaction with Amazon customer service agents. The discovery experience has been limited to mobile app and desktop searches.
While there will be hiccups and challenges as this rolls out, with any GenAI capability and tools, Rufus will continuously improve and learn as it engages with customers at scale. It will be fascinating to see how this plays out as Rufus is potentially rolled out across the Amazon ecosystem.
Amazon could improve customer ability to find things very easily – restructure its search algorithm so what appears is based on what the searcher asks for instead of purely based on who paid Amazon the most (which is how it appears to work today).
I don’t have any faith that an AI assistant is going to add value and wish people would settle a bit until we discover where AI is valuable.
Most of all, return to the basics. Why is it that when I search a specific product on Amazon including brand name the search result never include that brand of product. They’ve got far bigger challenges than AI can fix – especially now that they’re hooked on ad revenue.
I view this as a significant advancement in the area of site search. Making a search query easier for a less tech-savvy shopper seems like a great solution that makes Amazon’s huge catalog easier to access and narrow down to a few relevant items for a shopper to consider.
Additionally, the logs of the query can be mined for insights that can help Amazon and its sellers refine product content from an SEO perspective and then leverage AI to refine catalog descriptions based on the data captured in the log.
I view this as a game changer that we’ll all want to watch in the months ahead.
It’s not so much the threat, rather that we are seeing both Amazon and Google “pick a lane” and start to experiment, using us, the consumer as the live learning lab for improved prompt engineering. Yes, the proverbial “product item master table” has always been a “garbage in/garbage out” equation – BUT (in response to the second question) those that will learn how to carefully craft and refine the prompts/inputs given to Rufus, will benefit from improved outputs/responses, as long as Amazon’s developers are doing their part in understanding yet pushing the capabilities of the underlying AI model, but also acknowledging the nuances (we the end user are creating) in the training data that is “raising Rufus”
No doubt this will augment Amazon’s customer experience. Amazon will learn over time what customers are thinking and wishing they knew prior to making buying decisions. Ultimately it will help Amazon prioritize what content is promoted and upping sales and profitability.
Watch for this tech to eventually be used for shopping at Whole Foods, and driving sales growth across web search platforms.