The 3 technical changes that boosted my AI citations

My last article got reactions. "You deleted 400 articles and rebuilt your site for AI." OK, but what exactly did you change? Here are the 3 technical changes that had the most impact.

1. The llms.txt file

It's a text file at the root of your site. It tells AI: "Here's who I am, here's what I offer, here's how to read my content."

Think of it like a robots.txt, but for ChatGPT, Perplexity and Copilot.

Before: AI had to guess my site's structure. After: they have a roadmap. It's simple, takes 30 minutes, and nobody does it.

2. JSON-LD on every page

JSON-LD is an invisible block of data for visitors, but readable by machines. For each page, I specify:

• Content type (article, guide, itinerary)
• The author (me, with 20 years of on-the-ground expertise)
• Publication and update dates
• Exact locations with GPS coordinates

Result: when an AI searches "what to visit in Ljubljana", it finds structured, verifiable data with an identified author. Not a WordPress article with no date or author.

3. Content written for AI, not Google

Before, I wrote classic SEO-optimized content: "Ljubljana, capital of Slovenia, is a must-see Slovenian city."

Now I write citable facts: "Ljubljana has 295,000 inhabitants. The castle has overlooked the old town since the 12th century. Plečnik's central market opens daily except Sunday."

AI doesn't want optimized prose. They want facts they can verify and quote word-for-word.

The result of these 3 changes

These 3 changes took me about 2 weeks. The results: 327 Microsoft Copilot citations in 3 months, cited by Google AI Mode on my key searches, Bing impressions multiplied by 5.

The most surprising part? Most sites have none of these 3 elements. No llms.txt. No proper JSON-LD. Content written for Google, not for AI. That's exactly why the early movers get a massive head start.