How people find products has shifted, and retail leaders can sense it before they can measure it. Discovery increasingly happens in places — ChatGPT, Gemini, Claude — brands don’t control.
Understandably, this brings marketing teams to an intersection where the need to control the brand meets the need to earn a place in the answer, all in the shadow of one uncomfortable question: are you even showing up? If you’re not, a competitor is. It nets out to urgency, fueled by a fair amount of anxiety.
Don’t give in to analysis paralysis. Instead, take action to be found in AI search. You won’t have a perfect, fully formed strategy, but guess what? Nobody does — the playbook simply doesn’t exist yet. At any rate, you have to start somewhere. The brands I see pulling ahead have made their peace with taking action despite uncertainty.
The Window Is Measured in Months, Not Years
When mobile shopping arrived, retailers could reasonably say, “We’re not ready yet; let’s watch for a couple of years.” AI search gives you no such runway. Every month brings new changes: new commerce protocols, shopping features, updates, and more. Retailers are already building their own shopping agents within these platforms, and some are launching their own branded AI experiences. The distance between the brands that are experimenting and those who choose to wait and see is widening quickly, growing each quarter rather than each year.
Agentic commerce raises the stakes further. AI agents are now handling discovery and evaluation, and with standards like Google’s Universal Commerce Protocol (UCP), they’re handling the purchase itself (and beyond). Your customer may never see your product page. Instead, agents read the website’s backend to pull reviews and attributes, then decide whether to include your product in the answer. The problem is that most of that information wasn’t built to be machine-readable. If an agent can’t build a clear, consistent picture of your product based on the information it finds, you’re not in the conversation… but your competitor is.
Earn the Right to Be Recommended
Authenticity is taking on a new meaning. For years, it meant the voice matched the brand and the shop window looked the part. That still counts, but it now runs deeper, into how coherent and comprehensive your product data is. When an agent evaluates you, it’s checks the context: are the reviews strong? Is the product information rich enough to answer a real shopper’s question? Does your product feed data align with what your website says? I always tell customers that it doesn’t matter how many products they have if AI can’t get a complete picture of their brand. A catalog of thousands of SKUs means nothing if it’s invisible to agents. You have to earn the right to be trusted, recommended, and surfaced in the right places.
Building this authenticity and trust with the new AI audience is hard to orchestrate when the teams responsible for it are scattered across the org chart. Marketing and product sit in one place; e-commerce and IT elsewhere. The customer is the diamond in the middle, and each team is polishing a different facet of it from afar. When our customers ask me why some organizations move while others stall, the honest answer usually comes down to structure more than to technology. The structures, processes and workflows that got us here won’t get us where we need to be, and admitting that requires self-reflection at the leadership level.
The businesses adapting well organize around the customer rather than the org chart, and that model comes from the top down. It needs a C-level sponsor, someone willing to get a little ‘startuppy’ about risk in an organization that has spent years being trained to avoid it.
A Readiness Audit You Can Run This Quarter
None of this stays abstract for long, so here are the questions I’d want a retail team to answer right now.
Start with your product data. Are your core attributes, titles, descriptions, pricing, and availability accurate for every SKU, and are extended attributes like size, material, and use case present and consistent? Agents match products to intent using exactly these fields, and missing context can make you invisible to a query.
Then check compliance. The protocols that enable agents to transact, ACP and UCP, are quickly becoming table stakes and continue to evolve. Are your product feeds and catalogs meeting current protocol requirements? And do you have a process for staying current as those requirements evolve?
Look at freshness and scale. Product data changes constantly. How are you managing updates? If something sells out mid-morning and your feed catches up a day later, you erode trust with both the platform and the shopper. Enriching a hundred SKUs by hand is manageable; doing it across a hundred thousand isn’t, not without automation.
Finally, be honest about measurement. How are you evolving your measurement program to map back to your other actions? You may not be able to measure the same way you always have, at least not yet, and the movements will be small and incremental at first. That’s fine. What matters is that you can see where, when, and how you show up, whether it’s you, someone acting on your behalf, or someone else talking about you, and connect that activity back to traffic and revenue.
Don’t Wait for Perfect
You won’t get all of this right the first time, and perfection is the enemy of success. Learn what you can’t control, control what you can, and put the measurement in place, so you actually know whether you’re winning. The brands starting now with answer engine optimization (AEO) are learning while the stakes are still low. Would you rather learn now, or when the stakes are higher, and 25 others in your category are already winning? I know I’d rather learn now and be ahead of my competitors.
The author, Charli Rogers, is Chief Customer Officer at Botify.