AI has been a tool for marketers for years, but recently it became a tool for consumers as well, and that shift is changing how the customer journey works.
As founder of Command Growth, an agency that provides brands and merchants direct access to insights that move the needle on growth, I’m watching the changes closely. Here’s what my colleagues and I recently talked through at our webinar, Adapting Marketing Tactics for Agentic Commerce.
The top of funnel is shifting
A meaningful portion of product research and evaluation is now happening inside AI tools like ChatGPT, Gemini, and Claude before a customer ever visits a brand's site. Andriy Boychuk, founder and CEO of Flowium, noted that traffic arriving from AI platforms converts at a higher rate than traffic from traditional sources, not because volume is higher, but because the research phase is already complete.
When an agent recommends a specific product, the customer arrives with intent and a level of trust already established. Brands that are not showing up in those LLM conversations are being filtered out of the consideration set before the customer has a chance to find them.
What "AI-ready" actually means for a product page
Emily Shea, in Customer Education at Klaviyo, put it plainly: if an AI agent cannot confidently answer a customer's question from your product page, you risk losing the sale. Agents pull from what is on the page and do not infer or fill in missing information.
AI-ready pages go beyond marketing language. They need specific product specs in structured meta fields, FAQs built from real customer support tickets, and reviews with enough detail to answer the kinds of questions a consumer might ask an agent. Shea also pointed out that the things that make a page more useful to an AI are the same things that make it more useful to a human shopper, so better content improves conversion across all traffic sources, not just agentic ones.
Reviews are now product education, not just social proof
Marta Maciel, director of retention at G.O.A.T. Foods described a shift in how she thinks about reviews: they have moved from a UGC and retention signal into an acquisition signal. AI agents draw on review content when assembling recommendations, and a five-star rating with a short generic comment gives an agent very little to work with.
The questions she asks customers now are not about star ratings but about context: why did you buy this, who was it for, and what problem did it solve? A review from someone who bought a sugar-free product because they are diabetic can surface in a future AI query about gifts for someone with dietary restrictions. That specificity is what earns the recommendation. Both Maciel and Boychuk described updating their review prompts to ask for context rather than sentiment.
Agentic traffic behaves differently when it lands on your site
Boychuk described a consistent pattern across client accounts: visitors arriving from AI platforms land directly on a product page, skip browsing, and have shorter sessions. They came with a specific product in mind and are ready to complete the purchase rather than explore.
Boychuk's team now runs a separate welcome flow for anyone who opts in on a product detail page, one focused on trust and brand credibility rather than a promotional offer, because this customer may be hearing about the brand for the first time.
Content strategy: optimize to be both cited & ranked
AI agents are not looking for keyword density. They are looking for the source that most completely answers the question being asked. That makes FAQs built from real customer queries more valuable than they have ever been, and product descriptions with detailed specs more useful than high-level marketing copy.
Boychuk's team uses abandoned cart email sequences as a research tool: the last message asks what question the customer had about the product that went unanswered. Response rates are low, but the answers are detailed and feed directly into FAQ content. Maciel added that blogs have a different job now. They are not primarily traffic drivers but reference material that LLMs read and cite, which means depth and specificity matter more than calls to action.
Zero-party data: the strategy gap
Shea identified a common pattern: brands collect preferences and then do not change how they communicate, which breaks trust quickly. If someone says they only want to hear about new arrivals and receives a batch-and-blast campaign instead, they will not share preferences again.
A real zero-party data strategy starts before collection, with a clear plan for what changes once that data is in hand: which segments it creates, which flows it affects, and what specific messages become different as a result. Sign-up quizzes that branch into different welcome series and post-purchase surveys that route customers into different product tracks are examples of what this looks like when it is working.
Maciel also noted that asking a qualifying question at sign-up lowers submission rates, but the smaller, better-qualified list drives more long-term revenue than a larger one with no context attached.
Loyalty when AI is influencing the consideration set
The group agreed that AI agents influence discovery and comparison but that the final purchase decision is still made by the customer. The goal of a loyalty or membership program is not to automate the purchase but to build a relationship strong enough that your brand is worth choosing when an agent surfaces options.
Maciel described how G.O.A.T. Foods approaches this: they call it a membership rather than a loyalty program. Members are 30 times more likely to repeat purchase than customers in a conventional points program. The shift is away from transactional loyalty toward trust, convenience, and exclusive access. She also noted that agentic purchases tend to be more considered than impulse buys, more likely to be gifts or subscriptions, and more likely to support long-term retention.
Where to focus for the rest of 2026
Invest in copy. Maciel's top recommendation is to audit every product description on the site. If the copy does not cover the range of use cases a customer might ask an AI about, it needs to be rewritten. In her testing, longer and more complete pages have shown better conversion across all traffic sources, not only agentic ones.
Strengthen owned channels. As discovery moves into LLMs and more of the consideration phase happens before a customer reaches a brand's site, email and SMS become more important rather than less. They are the channels brands control directly, and they are where the customer relationship continues after an AI-driven first touch. Emily's view on email is that its function is changing more than its relevance. It’s increasingly an activation layer for all the customer data a brand has, including data about where and how customers first discovered them.
Related content
