Winning in Amazon's AI Era

Curtis Ritchie: Hi everyone, thank you for joining the webinar today. Good afternoon, everyone, and welcome to our Toucan Talks webinar. We're super excited to have you all join us for today's session. If you don't know me, I'm Curtis, our marketing lead here at Toucan, and we have Toby White, Toucan's senior client manager, who'll be talking about how you can win in Amazon's AI era. The session will run for around 30 minutes and we'll have a live Q&A at the end. If you have a question, please type it into the Q&A section at any time during the webinar, and I'll collect these while Toby is presenting. I'll now hand over to Toby.

Toby White: Brilliant, thanks Curtis. Thanks everyone for joining today - it's the most sign-ups we've had on a webinar for a while, so it's brilliant to have you all here. AI is a topic on everyone's mind at the moment, whether that's third-party tools using AI or Amazon incorporating it into the front and back end of their own systems, and the landscape is changing fast. What I want to focus on today is how conversational, AI-powered shopping is rewriting product discovery on Amazon, and what brands need to do today to succeed in this new era.

Toby White: Today's agenda: the evolution of Amazon product discovery, how AI is changing the shopping journey - specifically the shift to Amazon's AI-powered recommendations and conversational commerce - what AI looks for when recommending products (six core signals), why traditional Amazon SEO is no longer enough on its own and the shift from a keyword-focused stance to intent, context and answering consumers' questions, and finally, practical actions brands should take today.

Toby White: Why does this matter? Amazon has moved AI from a side experiment to the centre of its shopping experience in barely 18 months. This isn't a future trend, it's live and at scale, and it's already shaping what consumers buy and how they buy it. Some say this shift is as big as the move from desktop to mobile, or from organic to paid advertising, and brands that adapt early tend to win disproportionately. The good news is that the fundamentals that win in AI are clear: accurate, well-structured product information - things you can control on your own listings. So the game is winnable.

Toby White: If you look at the evolution of Amazon's homepage over the years, change is constant - not just on the homepage, but across the front end, back end, policies and best practices. In 2026, that means the integration of AI. If you zoom into a 2026 screenshot you'll see Rufus, recently renamed Alexa for shopping, in the top-left corner: a conversational, generative-AI-powered shopping assistant built into Amazon's app and website to help users research products, read review summaries and compare items. Traditionally, winning on Amazon meant winning the search box - customers typed a few keywords and Amazon returned a ranked grid of results. Winning meant the right keywords in your title and bullets, a strong sales velocity, competitive pricing, good reviews, and ad spend to buy the top slots. That's discovery-was-a-funnel-with-many-options: customers scrolled pages of results, opened several tabs, and compared listings themselves.

Toby White: Traditionally, this is where agencies like Toucan helped clients: content and SEO audits scoring listing quality and keyword performance, copywriting balanced for conversion and search visibility, hero image optimisation and AB testing, PPC to increase visibility across the funnel, and price analysis and strategic pricing recommendations, including around tentpole events like Prime Day and Black Friday. All of that remains important, but we need to acknowledge the ongoing shift in consumer shopping behaviour - understanding the rise of conversational shopping and Rufus adoption.

Toby White: Shoppers are bringing their AI habits from the rest of their lives - most people on this call have probably used ChatGPT, Gemini, or Claude - into how they research purchases on Amazon. They increasingly expect to describe a need rather than guess the right keyword. Research is moving upstream: people spend longer exploring within AI tools and arrive at the point of purchase already informed, rather than scrolling and comparing tabs on Amazon itself. And trust is being delegated - a meaningful share of consumers are comfortable letting AI summarise reviews and do the comparison work for them.

Toby White: Some stats: in a survey of over 4,000 consumers, 39% said they'd already used AI to shop, a behaviour that barely existed a year or two ago, and 79% of those felt more confident in what they bought having used AI to find it. More than 300 million customers used Amazon's AI assistant during 2025, and analysts expect roughly a quarter of shoppers to use AI chat while shopping in 2026. About a third of US consumers say they'd let AI actually make the purchase for them, and Amazon is testing an auto-buy feature integrated into Rufus in the US now - when a product hits a target price you've discussed with the AI, it buys automatically. If it's being tested in the US, at some point it's likely to be tested in the UK too.

Toby White: Instead of typing keywords and scrolling, customers are increasingly asking natural-language questions - 'what to buy for a home spa day' - and expecting a direct answer. Rufus sits inside the app or website, answers questions, compares products, and recommends a shortlist, often around five products, rather than the 50 or so you'd see across search result pages. In fact the search results page can be skipped altogether - you can ask Rufus a question on the homepage and click straight through to a product detail page. This collapses the funnel: AI does the scrolling and comparing on the customer's behalf, and the shopper sees a curated handful of products with reasons attached to each. The battle shifts from appearing somewhere on a 50-result page to being one of five AI recommends. The shelf just got dramatically smaller.

Toby White: AI isn't just on the homepage either - it shows up on the product page too, offering shoppers up to a year of price history, and answering live questions like 'is this treat good for dogs with sensitive stomachs' or 'is this made in the UK', pulled from your listing content and reviews. If the answer's there, you win in the moment; if it's not, you risk losing that shopper to a competitor. Generative and agentic AI is now woven through the whole journey - discovery, comparison, decision, and increasingly the purchase itself.

Toby White: If AI is choosing roughly five products to recommend, the obvious question is what it's looking at. There are six kinds of evidence that matter most. First, product content quality - the AI reads your title, bullets, description, A+ content and Q&As as source material to answer shopper questions; thin or vague content leaves it guessing, while clarity of use case (who a product is and isn't for) helps it match your product to the right query. Second, product attributes and metadata - structured backend data, often hidden from the product page itself, that's now a primary lever; missing or inconsistent attributes make your products invisible to filtered, intent-based queries. Third, customer reviews and sentiment - the AI mines reviews for themes and specific use cases and cross-references your claims against them, so if you claim leak-proof and reviews say otherwise, it'll surface that gap. Fourth, product availability and pricing - the assistant factors in price competitiveness, price history and stock, and out-of-stock or erratically priced products get recommended less. Fifth, brand authority and trust signals - consistent brand information and credible signals across the web help build AI trust, since it looks beyond Amazon too. Sixth, shopping history and personalisation - the assistant blends a shopper's past purchases and browsing into what it recommends, so repeat purchase and subscribe-and-save behaviour compound into future visibility.

Toby White: Traditional Amazon SEO isn't dead, and everything you've focused on previously still matters, but on its own it's now the price of entry, not the path to winning. Your strategy needs to shift from solely keyword optimisation to also factoring in intent optimisation - not just what words a customer types, but what they're actually trying to achieve, and whether your product clearly answers that. Keep your keywords, but layer intent on top: cover the job, situation and outcome your product delivers in plain language. The old question was 'am I ranking for these keywords'. The new question is: if a consumer described their need out loud, could an AI confidently recommend your product and back it up with the content and reviews behind it?

Toby White: So, what should brands do today? None of this is rocket science, it just requires doing the fundamentals deliberately with AI as your reader in mind. Six moves you can start this quarter: audit your PDP content against one test - could AI answer a shopper's real question using only this page - and prioritise your highest-revenue or highest-traffic ASINs first. Improve your product attributes and metadata - complete every relevant backend attribute and treat empty ones as lost recommendations; you can pull a category listing report in Seller Central or a bulk catalogue export in Vendor to see the gaps. Strengthen your review strategy through compliant means like the Vine programme and post-purchase follow-ups, since recency and relevance now feed visibility, not just conversion. Build more informative content - rich A+ modules, thorough Q&As and comparison-friendly detail, informing first and selling second. Anticipate the comparisons AI will make with other products in your market. And build consistency across channels - your Amazon listing, your website, other retailers and third-party sources should all agree, because inconsistent specs or claims create doubt, and consistency earns recommendations.

Toby White: To wrap up: discovery is changing, from typing keywords to asking questions, with AI curating a shortlist of around five products rather than pages of results. AI recommends on evidence - content quality, structured attributes, reviews and sentiment, availability and pricing. Keywords aren't enough on their own; optimise for intent, questions and context too. Accuracy beats hype, because AI cross-checks your claims against reviews and the wider web, so be honest, complete and consistent. And start now with the basics: audit your PDPs, complete your attributes, strengthen your reviews, add informative content, and stay consistent across channels. The brands that win in this new era won't be the loudest or the cleverest at gaming search, they'll be the clearest - the ones whose products are well described, well reviewed, and that Amazon has every reason to recommend. That's entirely within your control, and the best time to start is now.

Curtis Ritchie: Thanks so much for that, Toby. With all the AI features coming out nearly every month, it's really important to keep up to date. First question: how would a brand actually test whether AI is recommending them today?

Toby White: Good question. There's no report yet that attributes questions to product sales, though maybe Amazon will build one in future. For now, the main thing you can do is log in through a private browser so it doesn't use your historical data, think about the questions consumers in your category would actually ask - allergens, breed types, whatever's relevant - and see if your products come up.

Curtis Ritchie: Thank you. We've had two questions from Adam - one on tracking impact, which you've just answered - and another: if you focused on one action above all others, what would it be?

Toby White: I'd say product attributes. A lot of them sit in the back end, so they're not the glamorous content you see on the PDP, and they often get overlooked. If you download the spreadsheets relevant to your categories, there are lots of attributes the AI can draw on, and that's probably the most undervalued lever right now.

Curtis Ritchie: Thanks for that. One more question: what's one of the most common mistakes you're seeing on listings right now?

Toby White: Historically it was keyword stuffing, and people are getting more conscious of avoiding that. Now the mistake is not finding the right balance between relevant keywords and optimising for AI at the same time.

Curtis Ritchie: Thanks, Toby. We've run slightly over, but I want to say a massive thank you to everyone for joining our Toucan Talks webinar today. If you'd like to get in touch with Toby about today's session, or see how Toucan can help you win in this Amazon AI era, his email is on screen: toby.white@toucanecommerce.com. Thanks again, and we hope to see you at another Toucan Talks event soon.