Estimated reading time: 8 min
If you’re reading this because you suspect your website platform might be holding you back from AI search, you’ve already spotted something most businesses haven’t. That instinct? Probably genius.
What changed: AI search engines like ChatGPT, Perplexity, and Gemini don’t rank websites — they recommend businesses based on structured knowledge and entity relationships.
Why it matters: Gartner forecast that traditional search volume would drop 25% by 2026 as AI chatbots absorb the difference. The businesses AI recommends today are building a compounding advantage that gets harder to catch every month.
What to know first: This isn’t about “bad SEO.” Wix is genuinely fine for Google. The problem is narrow and specific: Wix wasn’t built for the structured data architecture that AI uses to understand and recommend businesses.
No. Wix is not bad for SEO. That’s a 2015 talking point and it’s not true anymore. Wix has invested heavily in server-side rendering, automatic image optimisation, and built-in SEO tools. For traditional Google rankings — where humans type queries and scroll through blue links — Wix performs respectably. Credit where it’s due.
The issue is different, and it’s newer.
When someone asks ChatGPT “Who’s the best heritage architect in Melbourne?” or asks Perplexity “Which mortgage broker in Brisbane specialises in first-home buyers?”, those AI engines don’t scroll through Google results. They build a knowledge model from structured data, entity relationships, and clean content passages — and then they recommend whoever they understand best.
That process is called Entity Optimisation. And Wix, through no fault of its own philosophy, is structurally hostile to it. Not because it’s a bad product. Because it was designed for an era where “being found” meant showing up on page one, not being understood by a machine that synthesises answers from the entire web.
Here’s what that means in practice — and once you see it, you’ll understand exactly what to do about it.
Think of Entity Optimisation as the difference between AI knowing you exist and AI understanding why you’re worth recommending.
Traditional SEO tells Google: “This website is about architecture in Melbourne.” Entity Optimisation tells AI: “Jane Smith founded this firm, studied at RMIT, holds registration with the Victorian Architects Board, is a member of the AIA, and specialises in heritage conservation and sensory design for inner-city residential projects.”
See the difference? The first gets you listed. The second gets you recommended.
AI search engines build confidence through relationships — connections between people, organisations, credentials, and specialisations. The more relationships they can verify, the more likely they are to recommend you when someone asks a question in your area of expertise.
Those relationships are expressed through a technical language called JSON-LD Schema — structured code that explicitly maps how entities connect. And the platform your website runs on determines how much of that code you’re allowed to write.
This is where the platform question becomes a business question, not a tech question. If you can’t express your entity relationships in code, AI can’t learn them. And if AI can’t learn them, it recommends someone who can express theirs.
This is the core issue, and once you understand it, everything else clicks into place.
Wix has a built-in schema generator. It handles the basics well — Local Business, Product, Article. If all you need is to tell Google “a business exists at this address,” Wix covers it.
But AI recommendation requires nested schema — code where a Person is connected to an Organisation, which is connected to Credentials, which are connected to Specialisations, which are connected to a geographic area. Each connection builds AI’s confidence that you’re the right answer.
Here’s what each platform lets you build:
What Wix produces (flat schema):
{
"@context": "https://schema.org",
"@type": "LocalBusiness",
"name": "Smith & Co Architecture",
"address": { "@type": "PostalAddress", "addressLocality": "Melbourne" },
"telephone": "+61 3 9000 0000",
"url": "https://smithandco.com.au"
}
That tells AI one thing: a business exists in Melbourne. Nothing about who runs it, what they studied, who they trained under, what they specialise in, or why they’re different from the 400 other firms in the same postcode.
What WordPress lets you build (nested entity schema):
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Smith & Co Architecture",
"founder": {
"@type": "Person",
"name": "Jane Smith",
"alumniOf": {
"@type": "CollegeOrUniversity",
"name": "RMIT University"
},
"knowsAbout": ["Heritage Conservation", "Mid-Century Residential Design",
"Adaptive Reuse"],
"hasCredential": {
"@type": "EducationalOccupationalCredential",
"credentialCategory": "Registered Architect",
"recognizedBy": {
"@type": "Organization",
"name": "Architects Registration Board of Victoria"
}
},
"memberOf": {
"@type": "Organization",
"name": "Australian Institute of Architects"
}
},
"areaServed": { "@type": "City", "name": "Melbourne" },
"knowsAbout": ["Heritage Overlay Architecture",
"Sensory Design", "Luxury Minimalism"]
}
One schema makes you findable. The other makes you recommendable.
Wix caps structured data at 7,000 characters per markup and five markups per page. For basic schema types, that’s fine. For the deeply nested entity graphs that AI recommendation requires — connecting people to organisations to credentials to specialisations — you hit the ceiling quickly.
Now, Wix does offer Velo, their developer platform, which allows programmatic schema injection. If you have a developer fluent in Wix’s proprietary syntax, they can technically write custom schema through the Structured Data API. But you’re building on a proprietary platform with skills that don’t transfer, limited documentation, and a dependency on Wix continuing to support the feature. You still face the character limits and code bloat. It’s the difference between jerry-rigging a workaround and building on a platform where schema control is native.
Here’s where Wix deserves credit for improvement — and where that improvement still falls short for AI recommendation specifically.
Wix now uses server-side rendering (SSR) by default. Pages are pre-rendered and cached on their CDN, which means Googlebot receives complete HTML rather than raw JavaScript. For traditional search crawling, this is a genuine fix. Wix has earned its improved SEO reputation.
But here’s the catch: SSR was built to satisfy Google’s crawler. AI crawlers are a different animal entirely.
GPTBot — OpenAI’s crawler — does not execute JavaScript. Period. It fetches the raw HTML from the initial request and processes only what’s there. Vercel’s research analysing billions of crawler requests confirmed that none of the major AI crawlers currently render JavaScript. And GPTBot’s crawl volume surged over 300% in the past year, all while processing only static HTML.
ClaudeBot and PerplexityBot behave the same way. Research from GSQI tested JavaScript-heavy sites across all three platforms and found consistent failures — ChatGPT, Perplexity, and Claude simply couldn’t find content that required client-side rendering. The bots saw empty space where the answers should have been.
Even with SSR in place, Wix sites still rely on JavaScript for interactive elements — accordions, expanding FAQ sections, tabbed content, dynamic widgets. The SSR cache delivers the page shell, but these interactive content areas often still need client-side execution to reveal what’s inside them.
Picture this: you’ve written thoughtful, detailed answers to every question your clients ask, wrapped them in a beautifully designed expandable FAQ widget, and the AI bot arrives, sees “Loading…” and moves on. Not penalised — invisible. The bot doesn’t know your answers exist.
For Google, this is a minor inconvenience. Google will usually render it eventually. For AI recommendation, it’s a dealbreaker. If the AI can’t read your expertise in the first request, your expertise doesn’t exist in its model.
This is the one that surprises people most — and it’s the easiest to understand once you see it.
AI systems don’t read web pages top to bottom like you do. They break content into passages — typically 100 to 500 tokens — and analyse each passage for its specific meaning. Google’s own Vertex AI documentation describes layout-aware chunking that uses “ancestor headings” to provide context. When AI encounters a chunk of text, it looks up the document structure to find the heading that chunk belongs to. That heading tells the AI what the text is about.
Simple concept: heading provides context, paragraph provides the answer. AI reads both, connects them, extracts the insight.
Here’s where Wix’s visual design flexibility creates an unintended problem.
To position elements precisely on screen — that beautiful drag-and-drop freedom — Wix wraps content in layers of <div> tags, inline styling, and positioning code. What you see as a heading followed by a paragraph looks very different in the actual code:
Clean code (WordPress):
<h2>How Much Does a Heritage Renovation Cost in Melbourne?</h2>
<p>Heritage residential renovations typically range from $800K to $1.2M
depending on overlay requirements and structural complexity.</p>
The AI sees the question. It sees the answer. The connection is instant.
What Wix’s editor actually outputs:
<h2>How Much Does a Heritage Renovation Cost in Melbourne?</h2>
<div style="position:relative">
<div data-mesh-id="comp-xyz">
<div data-mesh-id="comp-xyz-inlineContent">
<div data-mesh-id="comp-xyz-gridContainer">
<div id="comp-abc" class="...">
<p>Heritage residential renovations typically range from
$800K to $1.2M...</p>
The heading and the answer are now separated by layers of structural code. The AI’s chunking system may not recognise that the cost figure belongs to the heritage renovation question. The passage becomes an orphan — a piece of information disconnected from its meaning.
Research on content chunking for AI search describes this precisely: when content structure separates a heading from its supporting text, AI systems struggle to extract clean, accurate responses. They may grab partial information or miss the connection entirely.
Your beautiful words are all still there. Humans see them perfectly. But the machines making recommendation decisions? They’re reading through a fog of code.
This one’s less technical and more strategic — but if you’re thinking long-term about your business, it matters.
On Wix, your content lives inside Wix’s ecosystem. You can’t easily export your database, your structured data, or the entity relationships you’ve built. A comprehensive 2026 review of the platform estimated that migrating even a modest 30-page Wix site takes 12–20 hours specifically because of Wix’s data portability constraints. You’re renting their infrastructure.
On WordPress, everything you build belongs to you. The schema, the content database, the entity relationships — all portable. If you change agencies, change platforms, or sell the business, your knowledge base comes with you.
When you invest in Entity Optimisation — building a structured, machine-readable representation of everything your business knows and everything that makes it credible — that’s intellectual property. On Wix, if you stop paying, it disappears. On WordPress, it’s yours.
You’ve spent years building your reputation. The digital translation of that reputation should be an asset you own.
Good news: “migration” sounds scarier than it is. Here’s the typical process for a professional services firm with 20–40 pages.
Weeks 1–2: The Platform Shift. Core pages are rebuilt on WordPress using a clean, semantic theme designed for machine readability, not just visual design. Every old Wix URL gets a 301 redirect to its new home so you don’t lose any existing search equity. Your media library comes across and gets properly optimised.
Week 3: The Entity Foundation. This is what makes it different from a standard website rebuild. Organisation schema, Person schema, LocalBusiness schema, and the nested entity relationships connecting your people to credentials to specialisations — all injected from day one. FAQ content is restructured into clean, chunk-friendly HTML that AI crawlers can parse. Schema is validated, tested, and connected to your Google Search Console.
Week 4: Launch and Verification. The domain points to the new site. Sitemaps are submitted. AI crawler access is confirmed. And for the first time, the machines deciding who gets recommended can actually read the full story of your business.
Your existing content is preserved — it’s just moved into infrastructure that lets AI understand it. You don’t lose what you’ve built. You unlock it.
Three to four weeks. That’s the distance between being invisible to AI recommendation engines and being fully readable by them.
Related: https://probablygenius.com/the-ai-visibility-engine/
Partially. Wix’s SSR improvements are real and meaningful for traditional Google crawling. However, SSR delivers the page shell — interactive elements like accordions, tabs, and expandable widgets still require JavaScript execution to reveal their content. Since AI crawlers like GPTBot, ClaudeBot, and PerplexityBot don’t execute JavaScript, content behind these interactive elements remains invisible to them. The SSR fix was built for Googlebot, which is sophisticated enough to render JavaScript. AI recommendation bots are not.
You can add some custom JSON-LD — Wix supports it. But you’re limited to 7,000 characters per markup and five markups per page. For basic types like LocalBusiness or Article, that’s workable. For the deeply nested entity schema that AI recommendation requires — connecting people to organisations to credentials to specialisations across multiple layers — those limits become a genuine constraint. Wix also offers Velo, their developer platform, for programmatic schema injection. But that means building on a proprietary system with non-transferable skills, limited documentation, and a dependency on Wix maintaining the feature. It’s a workaround for a limitation, not a native capability.
The core challenge — control over schema depth, clean HTML output, and content portability — applies to most closed-platform builders. Squarespace and Shopify have similar constraints in different areas. The common thread is that platforms designed primarily for visual design tend to prioritise how things look in a browser over how they’re structured for machine consumption. WordPress with proper tooling gives you the most control over the structured data layer that AI recommendation depends on.
Ask ChatGPT, Perplexity, or Gemini a question that should lead to your business. Something like “Who’s the best [your specialisation] in [your city]?” If you’re not in the answer — or if the answer describes your competitors with more specificity than it describes you — your entity infrastructure is the likely reason. An AI visibility audit can show you exactly where the gaps are and what’s fixable.
https://support.wix.com/en/article/adding-structured-data-markup-to-your-sites-pages-2546962
https://support.wix.com/en/article/troubleshooting-your-structured-data-markup
https://vercel.com/blog/the-rise-of-the-ai-crawler
https://www.gsqi.com/marketing-blog/ai-search-javascript-rendering/
https://seo.ai/blog/does-chatgpt-and-ai-crawlers-read-javascript
https://prerender.io/blog/understanding-web-crawlers-traditional-ai/
https://cloud.google.com/document-ai/docs/layout-parse-chunk
https://searchengineland.com/guide/content-chunking-seo
https://ipullrank.com/misinformation-about-chunking
https://dev.wix.com/docs/velo/apis/wix-seo-frontend/structured-data
https://ismelguerrero.com/blog/wix-website-builder/
Be the answer, not just the link.
Quality Verified
90%
This content scored 90% in the Probably Genius Publication Readiness Assessment, meeting standards for direct answers, section depth, proof points, citation quality, and AI extractability.
Your AI visibility starts with a free audit. We’ll run your business through our 109-point AI Search Audit, show you exactly where the machines can and can’t find you, and give you a clear map of what to fix first. Book your free AI visibility audit →
By Jax Baker
February 17, 2026