After ChatGPT launched, a lot of companies started asking the same question: can we scale SEO with AI and win? Wix is one of the most interesting real-world examples — and the answer is more complicated than it looks.
On the surface, it looks like it worked. Their organic traffic doubled — from around 5M to 10M monthly visits. But when you dig deeper, the strategy and the risks become much clearer.
What Wix Actually Did
Wix scaled a large set of top-of-funnel "how to make money" articles — pages like "How to make money online", "How to make money on Instagram", "How to make money as a teenager", and "How to make money with AI." These pages were largely created around early 2023, right after ChatGPT launched, when AI content production exploded.
The important detail isn't just what they wrote. It's how they scaled: similar query patterns, similar article formats, similar intent (broad, informational, low commercial), and likely AI-assisted production with light editing. This wasn't a one-off test — it was a repeatable system applied across dozens of pages.
The Reality: This Is High-Risk SEO
Here's the key insight: this strategy works differently depending on your authority. Wix has a domain rating around 95. That gives them significant margin for error. If a smaller site tried the same approach, you might see a short-term spike — then a sharp decline.
This is what people are calling "Mount AI" — scale AI content quickly, traffic rises, then rankings collapse. Because large-scale AI content introduces system-level signals, not just page-level issues: repetition across URLs, predictable internal linking, similar topical coverage with low differentiation, and weak engagement signals over time.
⚠️ Mount AI: Once enough of these signals accumulate, the entire folder — or even the domain — can get devalued. This isn't a page-level problem. It's a system-level one.
The Biggest Problem: Patterns
The core issue isn't "AI content." It's detectable patterns across a system of pages. When you scale dozens of pages, patterns start showing up that Google reads as signals:
Repetitive content structures
Same intro format, same H2 flow (definition → steps → tips → conclusion), same list formatting across pages.
Similar writing style
Generic phrasing, no strong opinions or differentiation, surface-level explanations that don't add anything new.
Identical page templates
Same layout, same UI blocks, same content depth regardless of topic complexity.
Query pattern footprints
"How to make money + X" across dozens of pages with no expansion into adjacent intent — comparisons, tools, case studies, commercial pages.
💡 The key shift: Most people think Google evaluates pages. In reality, Google evaluates patterns across pages. If you can scroll through 10 URLs and they all feel the same, that's a problem.
Another Hidden Pattern: Outbound Links
Many AI-generated sites link to the same authority sources repeatedly — McKinsey, Deloitte, BCG. The logic is simple: add authority by citing trusted sources. But at scale, this becomes a footprint. Same domains linked across dozens of pages, same type of studies referenced, same placement in content. It's not that linking out is bad — it's that doing it the same way across 50+ pages is a detectable pattern of manufactured credibility.
Why Wix Hasn't Been Crushed (Yet)
There are a few practical reasons Wix is still ranking. First, domain authority — they can absorb weak content signals because they have a strong backlink profile, brand searches, and high trust signals. A smaller site doesn't have that buffer. Second, timing — they launched in early 2023, when there was less competition from AI content and less aggressive pattern detection. Third, relative scale — roughly 50 pages in this cluster on a domain like Wix is small. On a smaller site, that could be 80% of your content. And finally, Wix doesn't need these pages to survive. Even if this content underperforms, the business is unaffected.
The Information Gain Problem
This is where most AI SEO strategies fall apart. AI struggles to create genuinely new information. Most AI content rewrites existing ideas, combines known concepts, and doesn't introduce anything new. That creates two problems: no reason to rank (if 20 pages already exist on a topic and yours says the same thing), and no reason to be cited by AI (LLMs prefer data points, unique insights, and original examples).
What information gain actually looks like: original data (even small datasets), aggregated stats from multiple sources, real examples from your product or users, step-by-step workflows that aren't obvious. Even something simple like "we analysed 50 SaaS landing pages" is enough to differentiate.
What Actually Works Instead
The takeaway isn't "don't use AI." It's: stop using AI like a content factory. Break patterns by changing structure between pages, mixing formats, and varying depth. Expand beyond one query type — comparison pages, tool pages, templates, case studies, commercial intent pages. Add functional value with calculators, quizzes, or templates. Use AI as drafting, not publishing — AI first draft, human restructuring and examples, then add data or UX elements.
Final Thought
AI content isn't the problem. Predictable systems are. If your content strategy looks like the same structure, same sources, same intent, scaled across dozens of pages — it will work until it doesn't. The sites that last mix formats, add real utility, and create small but meaningful differences across pages. Because at scale, search engines don't just evaluate content. They evaluate patterns.