Marketplace Briefing: How Amazon sellers are rethinking SEO for AI search
This is the latest installment of the Marketplace Briefing, a weekly Modern Retail+ column about the ever-changing e-commerce marketplace landscape. More from the series →
About a year ago, when I interviewed Rajiv Mehta, Amazon’s vice president of search and conversational shopping, I asked a question that was on many sellers’ minds: Would brands need to rethink SEO to show up in AI-powered shopping tools like Amazon’s AI-powered shopping assistant Rufus?
Mehta, who oversees products including Rufus, gave a straightforward response. “I don’t think brands need to do anything different, or that it changes things for brands,” he said. “What’s most important,” he added, “is ensuring that you’ve got the most up-to-date, factual information about your products available in the store. But there’s nothing specific that I would encourage folks to do specifically or something that’s different.”
Amazon’s public guidance hasn’t shifted much since then. Asked recently whether there’s updated advice for sellers trying to improve their product mentions in Rufus, an Amazon spokesperson said the company didn’t have anything to share.
But that hasn’t stopped sellers from trying anyway. Even without official rules, sellers are adjusting their playbooks.
As Rufus becomes a more visible part of how shoppers search on Amazon, brands are experimenting with different tactics to figure out how to more prominently surface their listings in the chatbot’s responses. As Modern Retail previously reported, some of these tactics include adding more conversational language to product listings. For some, AI-friendly product listings have contributed to higher traffic and sales.
Amazon listings used to primarily rely on “keyword stuffing,” with sellers cramming as many keywords as possible into the listing to show up in search results. Unlike traditional search, Rufus is designed to understand context and intent, not just exact keywords, meaning a shopper searching for “gentle shampoo for sensitive scalp” could be shown fragrance-free or sulfate-free products, even if “sensitive scalp” isn’t spelled out in the product title.
Josh Blyskal, who leads answer engine optimization strategy and research at the AI search optimization startup Profound, gave an example. Before Rufus, a product title may look like a jumble of keywords — something like “gift chocolate valentines day dark milk assorted heart box candy 12 pack best gift” all crammed into a single line. Now, the title might look like, “Valentine’s Day milk & dark chocolate, 12‑piece heart box,” and the description would include conversational queries shoppers actually use: “best candy for Valentine’s Day” and “great for kids.”
This isn’t the only way product listings on Amazon are changing as a result of AI search. For instance, plant-based protein startup IQBar is making a bigger push into fiber next year, and what the company thinks will perform well with Rufus is impacting key decisions in pricing and packaging.
Because Rufus can read text on images, IQBar is updating its packaging and product listing images to prominently display the word “fiber” on the front of the package, the company’s founder and CEO Will Nitze told Modern Retail.
Despite seeing slightly higher costs from tariffs this year, IQBar is also keeping its prices stable. That’s in part because Rufus users commonly search by key price thresholds, such as, “What can I buy for under $20?” As Nitze put it, “Some of our products are below $10, some are below $25, some are below $20. These key price thresholds are advantageous for people searching on Rufus. If we’re at $19.99, that is going to help us as it relates to Rufus.”
It’s not just about adding more conversational language, but also including more detailed product information. Because Rufus relies heavily on product recommendations, the system needs more context to decide what to surface.
“Because GenAI chats are 25 words on average, versus three to four keywords in search, the engine holds way more context — name, location, preferences, even ‘I’m an autumn’ and ‘my scent profile is lavender,’” said Scot Wingo, author of the Substack Retailgentic and founder of ReFiBuy, a company that helps brands and retailers optimize for agentic AI. “For every product, you have to add a ton more content, and then add context to that content.”
Ryan Walker, head of retail at PMG, said brands trying to surface in Rufus need to focus on giving Amazon as complete a picture of a product as possible. That includes details that sellers may have ignored in the past, such as product dimensions, product compatibility, use cases, helpful advice related to the product, comparable products, product support information and the brand story.
“You want the most complete catalog upload possible,” Walker said.
Higher traffic, higher sales
Some sellers who have revised their listings for Rufus say they are seeing positive results, including increased sales and traffic. For example, one client of Profound’s, a confectionery brand, has seen a 58% year-over-year increase in sales, according to a test group of optimized products versus a control group of non-optimized products. Pattern, an e-commerce accelerator, previously told Modern Retail that brands with AI-friendly product listings can see median revenue lifts of up to 20%.
Mars Wrigley, the candy brand, sells its products on Amazon and has been experimenting with AI optimization on the platform. Six of the company’s brands — Pedigree, Starburst, Five Gum, Extra, M&M’s and Extra Refreshers — showed an average 8% increase in search visibility, according to Tory Bradley, Mars Wrigley’s global e-commerce retail search director.
And for clients of Katya Constantine, the CEO of ad agency DigishopGirl Media, products that used to be “slow-moving,” or had low sell-through rates, have seen sales surge after incorporating more conversational language to listings. “They’re still not top sellers, but they’re double of what they were selling previously,” Constantine said. She’s also seen traffic rise by as much as 35% after optimizing listings.
Sellers and consultants who spoke to Modern Retail for this story said they haven’t received guidance from Amazon on how to optimize for Rufus. At the same time, Amazon has AI-powered tools to help sellers create listings, including generating product descriptions, titles and bullet points. Amazon reports that sellers using these AI tools see a 40% increase in overall listing quality.
During its latest quarterly earnings call, Amazon said 250 million customers have used Rufus this year, with monthly users up 140% year over year. Customers using Rufus are also 60% more likely to complete a purchase. Amazon told investors Rufus is on track to deliver more than $10 billion in incremental annualized sales.
Still, at the end of the day, many of the fundamentals of traditional SEO on Amazon’s site, such as numerous, positive customer reviews, still apply. As Profound’s Blyskal put it, “I don’t think there’s a hack where I could go onto Amazon tomorrow and start ranking higher than DeWalt for power tools.”
What I’m reading
- Amazon is teaming up with Slope, a JPMorgan Chase &Co.-backed startup based in San Francisco, to offer financing for small online businesses, per Bloomberg.
- Rohit Prasad, an Amazon executive overseeing the company’s artificial intelligence division is leaving Amazon at the end of the year; Amazon has named Peter DeSantis to lead a new organization working on AI projects, according to a memo posted on Amazon’s website.
- The parent company of Roomba, called iRobot, has filed for bankruptcy due to mounting debt and tariff pressure.