· By Xhurian Shaba, co-founder and CEO of ecentic
13 min read
AI shopping agents now check local stock, price and availability before recommending a product. Here is how store owners keep their data accurate and visible.
If you landed here looking for how to run AI models on your own hardware, this isn’t that guide. We cover the other meaning: how AI shopping agents such as ChatGPT and Gemini decide whether your product is available to a shopper, including shoppers near a physical store, and what you can do so your listings pass that check. Agents can only recommend what they can read, trust and confirm as buyable. Local availability is part of that confirmation.
In classic search, availability was a line in a snippet. In an AI shopping flow it is a gate. When a shopper asks for a product, the agent builds a shortlist and drops anything it cannot confirm as purchasable, deliverable or in stock.
Three questions sit behind that gate:
Google describes the scale of what sits behind this. According to TechCrunch’s report on Google’s AI shopping update, AI Mode is powered by a Shopping Graph of more than 50 billion product listings, 2 billion of which are refreshed every hour. The same report says AI Mode answers can show details like price, reviews and available inventory, and that shoppers in the U.S. can search “near me” and use a “Let Google Call” option for categories such as toys, health and beauty, and electronics.
The practical reading: availability is checked continuously, by machines, and a merchant who feeds accurate data continuously has the advantage.
Honest numbers first. A study of 973 e-commerce sites reported by Frankfurt School researchers found that ChatGPT plays a limited role in online shopping overall. It creates value mainly for complex products with high information needs, and less so for standard items.
Separate analysis from Visibility Labs on 94 ecommerce sites found ChatGPT-referred traffic converting at 1.81% against 1.39% for generic organic search, about 31% higher, while representing less than 0.2% of total ecommerce traffic as of mid-2025. Revenue per session trailed established search channels but beat paid social.
What that means for you:
Agents prefer data they can parse without guessing. Availability, price, variants, shipping and return terms should exist as explicit fields, not only as text inside a description. Google’s own guidance for its Universal Commerce Protocol (UCP) is a good example of where this is going: the Merchant Center UCP guide explains that you update your product feed to signal eligibility and supply compliance data, using attributes such as native_commerce for checkout eligibility, consumer_notice for regulatory warnings and merchant_item_id, so agents can decide whether a product is eligible and calculate accurate totals.
If you want a broader primer on the data side, this overview of AI examples in ecommerce and machine-readable product information is a useful companion read.
An item that shows “in stock” in your feed and “sold out” on the page is a trust problem. The fix is boring and effective: one source of truth for inventory, synced often, with the page, the feed and the structured data all reading from it.
A common and costly mistake: your CDN or bot-management rules block the crawlers AI systems use. StoreFleet’s review of Perplexity and Google AI Mode for Shopify flags exactly this, naming PerplexityBot and Google-Extended, and recommends explicitly allowing them in robots.txt so your catalog stays visible. Check your firewall logs before you touch a single product description.
Availability gets you into the shortlist. Content decides whether you win it. Specs, materials, sizing, compatibility and honest use-case language let an agent match your product to a specific request. Our guides on optimizing product listings for AI shopping agents and on product knowledge graphs for Shopify and WooCommerce go deeper.
Many merchants ask how to “rank first” in ChatGPT. The same StoreFleet analysis notes that placement is probabilistic rather than a fixed position, and that agents favor relevant, structured data over generic SEO tactics. Services that promise guaranteed placement tend to be ineffective next to good hygiene in your product feeds.
Think in selection rates instead. If an agent is asked the same shopping question 100 times, in how many answers does your product appear? That is the number to move, and it moves when your data gets clearer than your competitors’ data.
There is also no single agent to win. AlixPartners’ analysis argues for a portfolio approach: fashion and apparel brands tend to gain more from discovery in ChatGPT and Perplexity, while replenishment-focused or commodity businesses often need stronger presence in native AI shopping flows like Gemini or Amazon Rufus. Check each agent separately.
Work through this in order. The first items cost nothing.
Say you run a Shopify store selling espresso machines, with one physical showroom. A shopper asks an assistant for a compact machine under a set budget available this week. The agent needs a price, a stock status, a delivery estimate and ideally the spec detail that proves “compact”. If your page says “ships soon” while the feed says “in stock”, and your dimensions sit inside an image, the agent has two uncertain fields and one unreadable one. A competitor with clean fields wins the shortlist even if your machine is better.
The fix is not clever copy. It is consistent fields, readable specs and an allowed crawler.
We built ecentic because merchants kept asking the wrong question. They asked how to rank in AI, when the useful question is why an agent picked a competitor’s product over theirs for a specific shopper request.
Our view is that availability is one factor among several the agent weighs, and you can’t see how it weighs them by looking at your own store. You have to simulate it. That is why we test against the actual agents, ChatGPT, Gemini, Claude and Perplexity, rather than a generic checklist, and why we tie every suggested change to a measured win rate and, where possible, attributed sales instead of to a feeling that the copy is better.
We also think most merchants should resist the urge to rewrite everything. Fix the data that makes you ineligible first: stock mismatches, blocked crawlers, missing fields. Then improve content on the few products where shoppers truly compare. Then rescan, because agents change and your competitors do too.
— Xhurian
If you run a Shopify or WooCommerce store, ecentic connects to it, simulates how AI shopping agents evaluate your product pages, and explains in plain English why you win or lose against competitors. You get listing rewrite suggestions, can publish changes in one click, and rescan to track how often agents select your products. There is a free introductory scan, so you can see your starting point before paying anything.
Published plans, from our pricing page:
| Plan | Price |
|---|---|
| Compete | $14.90 per month |
| Scale | $44.90 per month |
| Dominate | $149 per month |
| Performance fee | 3% per AI-attributed order |
Scan, Pay-as-you-go and Enterprise plans are also listed there without a published price. See how AI product listing optimization works, or the Shopify integration.
That is the other common meaning of the phrase, and it is about running models on your own hardware for privacy and control. This article covers something different: how AI shopping agents judge whether a product is available to a shopper, including nearby.
Not as a ranking purchase. Placement is probabilistic and driven by structured, relevant product data, so the dependable route is accurate feeds, consistent pages and allowed crawlers. Google does offer feed-based programs, so check the official Merchant Center documentation for your setup.
Google has announced a feature in the U.S. that calls local stores to ask about product availability for certain categories, and AI Mode can show inventory in answers. Availability for your products will vary by category and by how well your data is maintained.
It is small today, under 0.2% of ecommerce traffic in one mid-2025 analysis, but it converted better than generic organic search in that same data. Start with free fixes like crawler access and stock consistency, and put your effort on products where shoppers compare.
Guides and research on winning AI shopping agents.
