12 Experts on the Future of AI Shopping
AI is the new front door to commerce. Shoppers now start their research inside AI assistants, arriving on product pages already half-decided.
What's less clear is what brands should do about it, so we asked 12 experts who work on this every day.
The Expert Panel
- Lily Ray | Founder of Algorythmic
NYC-based SEO leader named the #1 most influential SEO expert in 2022 by USA Today. - Ethan Smith | CEO, Graphite
SEO growth leader with 18+ years of experience and a go-to expert on AEO. - Malte Landwehr | CPO & CMO, Peec AI
20+ year SEO expert recognized for enterprise SEO and AEO - Patryk Wawok | Founder, Organic Hackers
Technical SEO expert for Fortune 500 brands, with $50M+ in client revenue driven. - Phillip Jackson | CEO, Future Commerce
Commerce media founder and influential voice on the future of retail. - Charlie Marchant | CEO, Exposure Ninja
CEO and creator behind one of SEO's largest YouTube followings. - Rodrigo Stockebrand | Group Director, AKQA
Former Head of SEO @NASA, Amazon, Pfizer, Univision. Author: Answer Engine Optimization - Peter Rota | Senior Technical SEO Manager
Top 1% LinkedIn creator in the US for SEO, ranked #28 globally in the category. - Brodie Clark | Independent SEO Consultant
Australia's best-known name in SEO, with expertise in product feeds and SERP features. - Andreas Reiffen | Co-Founder & CEO, Pentaleap
Retail media pioneer and longtime authority on product advertising. - Freddie Chatt | Ecommerce SEO Consultant
Practitioner voice on catalog optimization across multi-channel surfaces. - Callum Lockwood | Director of Organic Search at Re:signal
Ecommerce SEO strategist working in search since 2010
The New Discovery Journey
The consensus
The first conversation about your product now happens without you in the room. A shopper asks an AI, gets a recommendation, and lands on your site already pre-sold – carrying whatever the engine decided about you on the way in. That makes them more trusting and far less patient: the bar for accurate, consistent, up-to-date product info just moved, because the AI already vouched for you. Your job at this stage isn't to make the first impression. It's to live up to the one the machine made for you.
1. How does AI-assisted discovery change what shoppers expect from the brands they find?
Patryk Wawok: "When an AI recommends your brand, the recommendation arrives with set up authority — the shopper is already pre-sold before they land on your site. ... Your brand is now defined by what AI systems have concluded about you from across the entire web and less by what you say about yourself on your site.”
Phillip Jackson: "According to our Future Commerce research, 80% of consumers started their holiday shopping sessions with AI platforms, and 63% say they trust AI as much as a brand website. ... we've accelerated trust with agents faster than any other technological adoption in history."
Brodie Clark: "They expect to receive up to date information and consistency in attributes provided. Shoppers will also expect there to be hallucinations and non existent products, or at least not be overly surprised when that happens."
2. Which product categories are seeing AI influence purchases most?
Andreas Reiffen: "The more complex or important the purchase, the more useful AI becomes. Electronics, home improvement, health & beauty — these are the categories where shoppers genuinely need help. AI is that sherpa. It guides, narrows, and recommends. But don't mistake that for autonomous purchasing. Humans still check out."
Callum Lockwood: "I don't think the question is really 'which categories are influenced most by AI?' anymore. The more important question for ecommerce businesses is: 'Which parts of the customer journey are now taking place within AI search and recommendation systems, and how do we become visible there?'"
Brand authority in AI search
The consensus
Getting AI to trust you comes down to two things working together. There's the earned side – what the rest of the web says about you in reviews, coverage, and comparisons – and it counts for a lot, because AI looks for agreement across sources before it recommends anyone. And there's the owned side – your own pages – where the practical wins are delivering the answer quickly and upfront, refreshing pages so they stay current, and owning a clear niche instead of spreading thin. The encouraging part is that a focused brand can move fast on both.
3. What signals does AI actually use to decide which brands are trustworthy enough to recommend?
Ethan Smith: “AI uses many of the same signals that search does. However, AI summarizes for consensus. It combines the core model, which is trained on billions of documents as well as RAG (retrieval augmented generation), which is search. You can optimize for the citations that appear in RAG in ways similar to search. Specifically, targeting popular prompts, building landing pages, optimizing content, and building an offsite presence.”
Lily Ray: "Looking at the thinking process language models use to drill down into specific results, plus reviewing the fan-out queries themselves, are great ways to see how the models decide which brands are trustworthy. For example, ChatGPT appears to use site: searches heavily to filter its results down by specific high-authority domains, or it adds trusted entity names to the search to generate higher-quality results. Being a well-known and trusted website, entity, and/or brand in the space can help drive more citations and recommendations."
Rodrigo Stockebrand: "The strongest signals I've seen (outside of top traditional organic rankings) are:
- High Semantic Salience – against a target prompt and its fanout. I've also seen nearly a 3x out-performance of pages with higher-paced salient blocks outperform those with lower-placed salient blocks. In other words, it's not just about high relevance, but also about how quickly and upfront that relevance is delivered to the user.
- Minimal top-placed anchored links – this one has been surprising, but pages with minimal anchors (in body text) tend to significantly outperform those that use them, especially when they're used at the top of the page. My hunch is these internal link anchors are sending a negative signal to an LLM that the "answer" is actually referenced elsewhere.
- Freshness – I've seen a near 2x jump in brand mentions and citations when posts were updated with fresh and current information, especially on time-sensitive queries/prompts (ie. "2026") when more than 30%+ of the article's content was updated."
4. What role do reviews, editorial coverage, and digital PR play in earning consistent AI recommendations – and which matters most?
Ethan Smith: “Citations in LLMs are heavily fragmented. Across categories, no single domain accounts for more than 5% of all citations. So, if you focus offsite only on a particular domain, you’re missing most of the citations. Digital PR is a great way to gain broad citation coverage. Affiliates and influencer marketing is another way.
Reviews specifically matter for commerce or for b2b. Sites like G2 and Trust Pilot are often part of citations for companies.”
Peter Rota: “They play a huge role! LLMs look to understand your brand from both your website and external sources. They look to validate what you say across the web and for Conesus of how you're talked about across multiple platforms. I would say which matter the most are the ones that cite and mention your brand. But, all of them play to create the narrative around your brand.”
Rodrigo Stockebrand: “I've found that this significantly depends on the query/prompt itself. Some are highly driven by "earned" levers like news coverage or community mentions, while others are most driven by a site's own content and editorial strength (the "owned" levers). The key is to figure out which lever is most impactful for each query/prompt, which you can generally see using an AI visibility/tracking tool like Peec AI. From there, you'll be able to determine which specific third-party sources and content optimizations are most impactful simply by seeing who's already being mentioned and cited in AI results. Which you can also find using Peec AI.”
5. Can a smaller or newer brand build enough authority to compete with established players?
Charlie Marchant: "The key is niche topical authority. ... In some ways, smaller brands actually have an advantage here. Bigger, more established players often have broader positioning because they're trying to appeal to lots of different audiences across lots of different categories. That can dilute their relevance. So yes, smaller brands can compete: not by trying to be everything to everyone, but by becoming the obvious recommendation for a very specific customer, need or niche."
Callum Lockwood: "In a recent Women's Fashion visibility study I conducted, brands with a niche focus and strong category alignment often outperformed much larger retailers. AI systems appear to favour brands that are strongly associated with doing one thing particularly well, rather than trying to cover everything."
Malte Landwehr: "Since very few brands have a good AEO/GEO practice right now, small and nimble brands have an opportunity to have outsized influence on the grounding process. Simply by spending a few thousand euros on third-party listicles and by being cringe enough to publish 'x vs y' and 'best [x]' style content that LLMs love to gobble up."
Product data for AI
The consensus
AI sees your product as text – it can't take in a lifestyle photo, so everything that does the selling has to be written down. A readable catalog is a complete one. Not "blue cotton dress" but who it's for, what it's made of, how it compares – plus the structured feeds and schema that let a crawler parse it in the first place. And it isn't only reading your page.Your product feed, and the reviews and comparisons living elsewhere, shape the recommendation as much as your PDP does – though most brands never treat them as the same job. The brands that win also answer the worries – quality, trust, safety – not just the specs.
Underneath all of it: AI forms its own view of each product, and that view is the one doing the recommending.
6. What does a catalog that AI can actually read and recommend look like?
Freddie Chatt: "Most ecommerce catalogs are built for the shopper who already knows what they want. A title, a hero image, a few bullets, maybe some marketing copy. That works when the buyer has decided. It doesn't work when an AI is trying to figure out what to recommend. A catalog AI can actually read has the product attributes filled in properly. Material, dimensions, use case, who it's for, what problem it solves. Product schema with GTINs, ratings, availability, and proper category mapping. Reviews and Q&A on the page. Specs that go beyond what the brand cares about and cover what the buyer actually asks. The brands doing this well aren't doing it for AI. They were doing it for humans already. AI rewards them because the work was done."
Charlie Marchant: “It doesn’t just say “blue cotton dress” — it explains who it’s for, what problem it solves, what occasions it suits, what makes it different, what sizes and materials are available, how it compares to alternatives, and what customers say about it.
For ecommerce brands, this means thinking beyond traditional product copy. AI needs to be able to understand the decision-making context around the product. We want to help AI understand not just what the product is, but why it should be recommended, who it should be recommended to, and in what context."
Brodie Clark: "Using product feeds via GMC Next and also feeds directly to ChatGPT and Copilot (Shopify has an early version of this). Standard SEO principles for crawlability also apply."
7. Is there a meaningful difference between optimizing product pages for AI and for traditional search?
Freddie Chatt: "There's overlap, and there are real differences. The overlap: technical foundations, brand authority, structured data, content quality. All of it carries across. The difference is the unit of optimisation. Traditional search is about ranking for a query. AI search is about being part of the answer the model synthesises. That means brand mentions across the web, citations in third-party content, reviews on independent sites, and comparison pieces matter more than they used to. Ranking number one is less useful if the AI is pulling from a Reddit thread that doesn't mention you. The other big shift is the product feed. AI shopping experiences lean heavily on feeds from Google Merchant Center and similar sources. Most brands treat the feed as a paid media asset and ignore the SEO and AI implications. That's a mistake worth fixing."
Charlie Marchant: “AI will almost always look for an indication of pricing when it's trying to compare a product or service against competitors, because it knows that's the kind of information people want to see. If that information is missing, vague or hard to find, you're making it harder for AI to confidently recommend you."
Patryk Wawok: "There are fewer differences than some people would like you to think. ... The mechanism has shifted: AI tools retrieve documents and information and look for consensus across sources before generating an answer. What matters is being a content source a model wants to retrieve and cite, rather than chasing individual prompt rankings."
8. What are high-performing brands doing differently with their product content?
Freddie Chatt: “They build supporting content around the product, not just on it. Buying guides, "best X for Y" pages, comparisons between their own products. These give the AI somewhere to land and a reason to cite them. They take reviews and UGC seriously. Real customer language shows up directly in AI responses, and the brands with deep, authentic review content are getting referenced more often as a result.”
Brodie Clark: “They're doubling down on UGC. The more that you can differentiate your PDP and provide social proof goes a long way.”
Andreas Reiffen: "Turns out writing your PDP for a robot is good business. High-performing brands are investing in structured attributes, rich specs, and consistent taxonomy because the most important 'shopper' evaluating their content right now can't appreciate a lifestyle photo."
Patryk Wawok: "They don't focus only on their product content. They're identifying the specific anxieties people have about their brand and products (quality, trust, safety) and addressing those directly, rather than assuming a good product description is enough."
What's Next
The consensus
Most of what's described as "the future of AI shopping" is already turning up in small ways – agent-driven checkout, shopping built into the search box, buying without visiting a site. The experts disagree on where it leaves the balance of power between brands, marketplaces, and the platforms, and that argument is worth reading in full. But they agree on the part you can act on: the head start is real, and the brands paying attention to how AI represents them now will be the ones ready when the buying moves inside the engine.
9. As AI plays a bigger role in shopping, who gains the most power: brands, marketplaces, or AI platforms?
Phillip Jackson: "Brands still hold a tremendous amount of power in the short-to-midterm. ... 77% of shoppers still prefer to click through to the brand website and finish the purchase for themselves. ... those consumers arriving from AI platforms like ChatGPT convert at a higher rate, but those shoppers wind up purchasing with smaller, lower AOV baskets. Meaning, it's incumbent on the brand to adapt their pages ... to get shoppers to get their basket sizes up."
Rodrigo Stockebrand: "AI platforms are slowly shaping to hold more of the power ... but are also becoming the single largest source of first-party data on the planet. Which can lead to a whole new level of product/service personalization. And, gatekeeping. Yet, on the other hand, we're also seeing a great resurgence of brand equity like never before, as brands ... become the de-facto choice for users ... Only time will tell."
Andreas Reiffen: "Users will rely on AI to research and shape their purchase decisions. But I see AI behaving more like a marketing channel than a checkout button, driving traffic to websites where users will browse and buy on their own terms."
Callum Lockwood: "Each player controls a different part of the experience. The AI platforms sit at the centre because they control the interface, recommendation layer, and information synthesis. ... Brands are arguably in the strongest position when it comes to controlling product representation because they are closest to the source of truth ... Over time, I think the balance of power will depend on who is best able to structure, enrich, and distribute product information across these ecosystems."
Malte Landwehr: The answer engines will be even more extreme gatekeepers than search engines. ... I am sure we will see factory-to-LLM cases where Chinese manufacturers sell directly via ACP, UCP, and similar protocols. The biggest losers in this will be undifferentiated merchants and marketplaces.
10. If you had to make one bet on where ecommerce sits in 2027, what is it?
Phillip Jackson: "The real bet is that those who have invested heavily in AEO/GEO are the ones reaping rewards. That work compounds over time, if you haven't started, get started. The "zero-click phenomenon" will only continue to divert traffic away from authoritative sources; and with new features like Google's Universal Cart, consumers can now shop a brand from Gmail and YouTube without even visiting the website."
Charlie Marchant: "My biggest bet would be that by 2027, AI will become the new interface for consumer shopping decisions. I believe the majority of ecommerce discovery and decision-making will happen inside AI searches and interfaces, not just on ecommerce websites or traditional search results pages. People won't always start by searching "best moisturiser for sensitive skin" and clicking through ten websites. They'll increasingly ask an AI assistant for a recommendation, upload a photo of their skin now, explain their preferences and skin type, share their current products, compare options, ask follow-up questions, and potentially buy through that same experience. That doesn't mean ecommerce websites disappear, but it does mean the website's role changes. It becomes less of a standalone destination and more of a source of truth that AI systems use to understand, compare and recommend products."
11. What's the conventional wisdom about AI shopping that you think is wrong?
Freddie Chatt: "The idea that AI shopping needs a new playbook. It mostly doesn't. Most brands worried about AI visibility haven't done the ecommerce SEO basics yet. PDPs are thin, feeds are broken, category pages are filtered grids with no content, and schema is half implemented."
Andreas Reiffen: "That AI will kill retail or take over purchase decisions. For the vast majority of shopping, that's not happening. AI will make shoppers smarter, not replace them."
Callum Lockwood: "I think there's a misconception around agentic commerce feeling distant or unrealistic. In reality, we're already seeing the early stages of this shift happening now. ... The introduction of in-SERP checkout experiences alone has the potential to significantly reshape ecommerce journeys."
The AI shopping playbook: four moves
The consensus
For all the questions they answered differently, the experts agree on a few things. They agree that discovery has moved inside AI search — ChatGPT, Perplexity, Gemini, Google's AI Mode — and that getting recommended there comes down to the fundamentals, not a brand-new discipline. They agree that GEO and AEO mostly reward the ecommerce basics done properly, for a reader that reads your product instead of seeing it. And they agree the work compounds, so the brands that start now build a lead.
1. Get discovered
Find the AI queries your category triggers, and check whether your products show up in them at all. You can't capture demand you can't see.
2. Earn authority
Get cited across the web — reviews, digital PR, comparisons — and win a tight niche before chasing the broad terms. Citations are fragmented, so broad coverage beats one big placement, and focus beats scale.
3. Fix your product data
Get cited across the web — reviews, digital PR, comparisons — and win a tight niche before chasing the broad terms. Citations are fragmented, so broad coverage beats one big placement, and focus beats scale.
4. Prepare for what's coming
Agentic checkout is already arriving in pieces. Don't panic, but don't wait — get a read on how AI represents your catalog today, while the head start is still there to take.
None of it is exotic. It's the basics most brands already know they should be doing – clean product data, earned authority, a focused niche – just left undone. What's changed is the payoff. Right now, with so few brands doing AEO and GEO properly, the effort buys outsized visibility, and it compounds over time. Brands that put in the work now are the ones AI will keep surfacing later. The window doesn't stay open forever.