Shopify has introduced Agentic Storefronts, a capability that makes products from eligible stores discoverable to AI shopping assistants — including ChatGPT, Google AI Mode, and Microsoft Copilot. The programme is rolling out progressively; stores may be enrolled without explicit opt-in depending on configuration and region.

The problem is that being enrolled and being recommended are completely different things.

Over the past two weeks, I manually reviewed 12 Australian SMB online stores across five different niches — candles, homewares, jewellery, handmade bags, and specialty food — to see how visible they were to AI shopping agents. The results were striking.

Not one store scored above 50 out of 100.

What We Actually Checked

Every store was run through the AICE (AI Commerce Enablement) diagnostic — 54 signals across 6 pillars that determine whether an AI agent can find, understand, trust, and recommend your products.

The six pillars are:

The Results: 12 Australian Stores Audited

Here's what every store scored. All are real Australian businesses — founder-operated, quality products, genuine brands.

Candles · Sydney NSW

38

🔴 Red Zone

Homewares · Blue Mountains

31

🔴 Red Zone

Candles · Brisbane QLD

48

🟡 Amber Zone

Handmade Bags · Melbourne

36

🔴 Red Zone

Jewellery · QLD

29

🔴 Red Zone

Soap · QLD

44

🟡 Amber Zone

Soap · Albany WA

31

🔴 Red Zone

Chilli Oil · NSW

36

🔴 Red Zone

Condiments · Southern Highlands

29

🔴 Red Zone

The average AICE Score across all 12 stores was 34/100. That's deep red. Every single store had at least 3 critical AI visibility gaps. Most had 8 or more.

The 5 Most Common Gaps We Found

The same problems appeared in almost every store, regardless of platform, niche, or size. Here's what they are and why they matter.

Gap #1 — No Schema.org Product Markup Found in 11/12 stores
What it is Schema.org is a standard way of marking up product data in your HTML so machines can read it. Without it, AI agents see your product page as an anonymous block of text — they can't formally identify what you sell, its price, availability, or what it's for. Why it matters When AI systems evaluate products for recommendation, structured data is a significant input. Stores without Schema markup may be at a disadvantage compared to stores that provide machine-readable product details — even when enrolled in Agentic Storefronts. The fix For Shopify stores, install a Schema app (JSON-LD for SEO is popular) or add it manually via your theme. For WooCommerce, Yoast SEO handles this. For other platforms, it requires a developer.
Gap #2 — No llms.txt File Found in 12/12 stores
What it is llms.txt is an emerging convention — similar to robots.txt — that provides structured information about a website to AI systems. It is not an official web standard, and support varies across platforms and AI providers. Why it matters Without it, AI systems must infer your store's purpose and product set from HTML, schemas, and feeds alone. Some stores may have an auto-generated LLMS-compatible endpoint via their platform — but this is not universally available. Where it exists, it can provide clearer orientation; where it doesn't, the file can be created manually. The fix For Shopify stores, check whether an LLMS-compatible endpoint is available in your configuration. If not, create an llms.txt manually and serve it via an app, proxy, or middleware that exposes it at yourstore.com/llms.txt.
Gap #3 — Product Descriptions Written for Humans, Not AI Found in 10/12 stores
What it is Most product descriptions are written to evoke emotion — "a beautiful handcrafted candle that fills your home with warmth." That's great for humans. AI agents need something different: clear use-case framing, comparison attributes, and literal specifics. Why it matters When someone asks ChatGPT "best soy candle for a bedroom," the AI scans product descriptions for use-case language. "Ideal for bedrooms, living rooms, and gifting" beats "a beautiful candle" every time. The fix Add a structured attributes section to each product: "Best for: [use cases]. Scent notes: [specific]. Burn time: [hours]. Wax type: [material]." Keep the emotional copy too — just add the structured layer.
Gap #4 — Trust Signals in Plain Text Only Found in 11/12 stores
What it is Reviews, return policies, and business identity are on most sites — but in plain HTML text that AI agents can't formally parse or cite as authoritative. Why it matters AI systems use trust signals when evaluating which merchants to recommend. A store with AggregateRating schema and structured trust data may be favoured over one with equivalent information only in plain text. The fix Add AggregateRating schema to product pages, hasMerchantReturnPolicy schema to your returns page, and ensure your business contact details are in structured data, not just plain text.
Gap #5 — AI Crawlers Partially or Fully Blocked Found in 8/12 stores
What it is Many stores unknowingly block AI crawlers in their robots.txt, or use JavaScript-heavy themes that render as blank pages to crawlers. GPTBot, ClaudeBot, and PerplexityBot all have to be explicitly permitted. Why it matters If AI crawlers can't access your site, none of the other fixes matter. This is the most fundamental gap — and the most common. The fix Check your robots.txt file (yourstore.com/robots.txt). Ensure GPTBot, ClaudeBot, PerplexityBot, and OAI-SearchBot are not disallowed. If using a JS-heavy theme, ensure core product content renders in HTML without JavaScript.

The Interesting Exception: Brand Authority vs. Technical Readiness

One thing that surprised us: two stores were already showing up in ChatGPT results despite low AICE Scores. Business 1 (48/100) and Business 2(36/100) both appeared when we searched their categories in ChatGPT.

Why? Brand authority. One business has 1,100+ reviews and significant press coverage. Another business has a high-profile founder story and strong social media presence.

But here's the important nuance: they're visible despite their technical setup, not because of it. A competitor who implements the technical fixes properly could improve their position significantly without having anywhere near their brand recognition. Their AI visibility is fragile. Stores with stronger technical readiness may be better positioned in AI results.

Brand authority currently overrides technical readiness in AI results — but only temporarily. As AI commerce matures and more stores optimise their data, technical readiness will become the baseline, not the differentiator. The stores that move now will own those recommendation slots.

Your 7-Step AI Visibility Checklist

Based on our audits, here are the highest-impact fixes in priority order. A store that implements all seven would score 70+ on the AICE diagnostic.

AI Visibility Checklist — Start Here
01 Check your robots.txt — ensure GPTBot, ClaudeBot, PerplexityBot are not blocked Quick Win
02 Verify or create an llms.txt file — provides structured context for AI systems Quick Win
03 Add Schema.org Product markup to all product pages (name, price, availability, description) Medium
04 Add AggregateRating schema to product pages — make your reviews machine-readable Quick Win
05 Rewrite product descriptions to include "best for" use-case language and specific attributes Medium
06 Add a FAQ page with FAQPage schema — answer the questions AI agents ask on behalf of buyers Medium
07 Verify your store is enrolled in platform AI channels (e.g. Shopify Agentic Storefronts) Quick Win

What This Means for Australian SMBs

The stores we audited aren't failing because of bad products. They're failing because the technical layer between their products and AI agents doesn't exist yet. The gap is almost entirely fixable — most of the highest-impact changes are Quick Wins that take under an hour each.

The window to move early is right now. AI-powered product discovery is growing rapidly on ecommerce platforms. The stores that establish strong technical visibility in 2026 will be better positioned as these channels mature.

Research methodology note: The 12-store audit data in this article comes from manual reviews conducted by AICEscore in June–July 2026. Stores were assessed using the AICE diagnostic (54 signals, 6 pillars) based on publicly accessible signals: robots.txt, schema markup, llms.txt presence, page content, and review structures. No automated tools or private data were used. Stores were selected from public Australian Shopify store directories and covered candles, homewares, jewellery, bags, soap, condiments, and specialty food niches. Scores reflect technical signals at a point in time and do not measure conversion, brand quality, or business health. We do not claim that any single fix causes improved AI recommendations.

For Australian SMBs specifically, this matters more than in other markets. We don't have the brand authority of US and UK competitors. We can't rely on recognition alone to get recommended. Technical readiness is how smaller Australian businesses level the playing field against bigger international brands in AI results.

Want to know your AICE Score?

We audit Australian SMB stores manually and send you a full report — score, pillar breakdown, and prioritised fix list. Free, no commitment.

jay@aicescore.com →

Usually responds within 24 hours. No technical setup required — we work from your public-facing pages only.