By Steve Merrill | April 27, 2026
A client came to us frustrated. Their product was in stock. It had been in stock for months. But it had disappeared from ChatGPT Shopping results and Perplexity recommendations almost entirely. We ran an audit and found the problem in under ten minutes.
A single inventory flag. Set wrong.
The product's Shopify inventory_policy was set to "deny," which tells Shopify to mark the item unavailable when stock hits zero. The stock had briefly hit zero during a fulfillment delay months earlier. Then inventory came back. But the available flag in their products.json never flipped back to true the way it should have. The product was sitting in a weird limbo state — in stock in their warehouse, technically listed in Shopify, but broadcasting "unavailable" to every AI crawler that hit their feed.
They lost roughly three months of AI visibility on one of their top-selling products. No alert. No warning. Just gone.
AI shopping assistants don't browse your product pages the way a human does. They read structured data feeds. ChatGPT Shopping, Perplexity Shopping, and Google's AI Overviews all pull from your Google Merchant Center feed and your Shopify products.json endpoint. When those sources say a product is unavailable, the AI skips it.
There's no nuance. No "well, let's check the store to be sure." The flag says unavailable, the product gets excluded.
This is different from a few years ago when ranking algorithms had more tolerance for messy data. AI recommendation engines are tighter. They're built to surface products customers can actually buy right now. Anything that signals otherwise gets filtered out at the data layer, before it ever reaches a recommendation prompt.
According to research from Rethink Retail, availability and price accuracy are the two most common disqualifying signals in AI product feeds. Availability errors alone account for roughly 40% of product suppression cases in merchant audits.
Three things trigger this, and they all happen quietly in the background.
1. Inventory policy set to "deny" with a brief zero-stock event. When your inventory_policy is "deny," Shopify marks variants unavailable the moment stock hits zero. If you restock quickly, the available field is supposed to flip back. In practice, this sync can lag or fail entirely, especially if you're using third-party inventory apps that write directly to Shopify's database without triggering a full product sync.
2. Third-party app conflicts. Inventory management apps, pre-order plugins, and backorder tools all write to product availability fields. I've seen this in dozens of audits (a clean Shopify backend, a connected inventory app, and the two writing conflicting availability values where the app wins and marks things unavailable that aren't). The app works fine for orders. The AI feed is broken.
3. Variant-level availability discrepancies. Your product might show as "available" at the top level in Shopify, but if specific variants (a size, a color, a bundle) are flagged unavailable, AI crawlers may exclude the whole product. This depends on how the AI platform handles partial availability. Some are generous. Many aren't.
Start with your products.json. Go to yourstorename.myshopify.com/products.json and look for the available field on each variant. What you want to find is this specific combination:
inventory_quantity greater than 0
available: false
That combination is the bug. Product exists. Stock exists. Flag says no.
You can also check this in Shopify admin by filtering products by availability and cross-referencing your inventory report. Any product showing "unavailable" that you believe is in stock needs a closer look at the variant level.
If you're using Google Merchant Center, pull a diagnostics report and filter for "availability mismatch" errors. Google flags these explicitly, and they correlate directly with AI Shopping suppression. According to Google's Merchant Center documentation, availability mismatches are among the top five reasons products are disapproved from Shopping campaigns and AI-powered surfaces.
Third check: look at your Merchant Center feed directly. Export it and filter for availability = out of stock. Compare that list against your actual Shopify inventory. Any discrepancy is a problem.
Four steps. Most stores can do this in under an hour.
Step 1: Change inventory_policy from "deny" to "continue" on products that you want to show as available even during brief zero-stock windows. This doesn't mean you'll oversell — your order management can still block purchases. But it keeps the available flag accurate.
Step 2: Force a product sync. In Shopify, make a minor edit to affected products (add a period to a description, then remove it) and save. This forces a fresh export to your Merchant Center feed.
Step 3: Resubmit your feed in Google Merchant Center. Don't wait for the automatic daily crawl. Go to Feeds, select your primary feed, and click "Fetch now." According to the Google Shopping Content API documentation, manual fetches can accelerate re-indexing by 24-48 hours.
Step 4: Audit your third-party apps. Check the settings in any inventory or pre-order app to confirm they're writing the correct availability values. When in doubt, temporarily disable and re-enable the integration and run a fresh sync.
The client who lost three months of visibility? After fixing their inventory_policy and resubmitting their feed, they were back in ChatGPT Shopping results within two weeks. Not immediately. But faster than starting from zero.
Monthly audits. Not weekly, not daily — monthly is enough for most stores, as long as you're disciplined about it.
Set a recurring task: on the first of every month, export your products.json, check available flags against your actual inventory, and run a diagnostics report in Merchant Center. Takes about 15 minutes once you know what to look for.
If you have a large catalog (500+ SKUs), consider a simple Python script that pings your products.json and flags any variant where inventory_quantity > 0 and available = false. Not complicated. Totally automatable. We've set this up for a handful of clients and it's caught issues before they become invisible months of lost visibility.
The bigger lesson here: AI shopping platforms don't give you feedback when they drop your products. No email. No penalty notice. No traffic drop you can trace. They just quietly stop showing you. The only way to know is to check.
This is one audit I'd run on every store right now, before the back-to-school and BFCM windows open.
Yes. When a product is marked "sold out" in Shopify, AI shopping assistants like ChatGPT and Perplexity treat it as unavailable and remove it from recommendations. If the item is still in stock but incorrectly flagged, you lose that visibility with no warning.
Recovery time varies by platform. Google typically re-indexes within a few days once the feed is resubmitted. Perplexity and ChatGPT can take 1-4 weeks to reflect corrected product data depending on their crawl frequency for your domain.
The inventory_policy field in Shopify controls whether a product shows as available when stock hits zero. If set to "deny," the product goes unavailable at zero. AI crawlers read availability status from your products.json and Google Merchant feed, so a wrong policy flag directly suppresses AI recommendations.
Yes. Go to yourstorename.myshopify.com/products.json and search for available: false. Look for variants where inventory_quantity is greater than zero but available is false. That combination is the bug.
Monthly at minimum. Inventory syncs, theme updates, and third-party app installs can all quietly change availability flags. A 15-minute monthly audit of your products.json catches these issues before they cost you weeks of AI visibility.
Check Your Store's AI Readiness →
This is where I document what I'm actually building and observing — not predictions about what AI might do someday, but what's happening right now with Shopify stores, AI shopping assistants, and the shift in how people find products online.
I run AI visibility audits on Shopify stores. I see the data. Most stores are invisible to ChatGPT, Perplexity, and Google AI Overviews — not because their products are bad, but because the structural signals AI crawlers look for aren't there.
That gap is what this blog covers.
AI Commerce Readiness
How AI shopping assistants like ChatGPT Shopping, Perplexity, and Google AI Overviews decide which products to recommend — and what Shopify stores need to do to show up. This includes structured data, product feed optimization, and content structure.
Answer Engine Optimization (AEO)
AEO is the practice of structuring your content so AI systems can extract it, quote it, and cite it. Different from SEO. Different signals, different ranking factors, different content requirements. I break down what it actually looks like in practice.
Real Data from Real Audits
I've audited hundreds of Shopify stores for AI readiness. The patterns are consistent. I share anonymized findings, before-and-after examples, and what the numbers actually show — not what anyone's guessing.
Agentic Commerce
AI agents that browse, compare, and recommend products are already live in ChatGPT, Copilot, and Perplexity. I cover what's changing, what Shopify's platform is doing about it, and what merchants need to do now before the window closes.
What is AI commerce readiness for Shopify stores?
AI commerce readiness is a measure of how well your Shopify store is structured for discovery by AI shopping assistants like ChatGPT, Perplexity, and Google AI Overviews. It includes your structured data (JSON-LD schema), product feed quality, robots.txt permissions for AI crawlers, and content extractability. Most stores score an F when audited for these factors.
What is Answer Engine Optimization (AEO)?
AEO is the practice of structuring your content so AI systems can find it, understand it, and cite it when answering user questions. Unlike SEO, which targets a ranked position on a results page, AEO targets a citation inside an AI-generated answer. The signals are different: question-based headings, structured Q&A content, clear definition blocks, and authoritative external references.
How is AI product discovery different from Google Search?
Google Search returns a list of links ranked by relevance. AI shopping assistants like ChatGPT and Perplexity synthesize a recommended answer — selecting specific products or brands based on structured data, citation patterns, and content credibility signals. 67.8% of pages cited by AI don't rank in Google's top 10, according to Surfer SEO's research. Optimizing for one doesn't automatically optimize for the other.
How do I know if my Shopify store is visible to AI shopping assistants?
Run a free AI Commerce Audit Here. It scores your store across the key AI discoverability factors — structured data, product feed coverage, content extractability — and identifies what to fix first.