Overhauling the Text Structure (The ‘Blocks’ Method)
AI models chop your text into chunks for scanning. Therefore, write your pages in fixed blocks of a maximum of four to five sentences that independently answer a specific question.
Instead of a long-running story, place a clear, questioning heading above each text block. Think of the exact question someone types into ChatGPT or Perplexity.
Bad example (Floaty and vague):
“When you notice that the systems in your network are no longer performing optimally and data streams are stalling, it is time to look at a better setup. Our passion lies in streamlining these processes so that everything feels smooth again.”
Good example (Concrete and direct):
How do you resolve a slow database connection?
“A slow database connection in a local PHP environment is often caused by repeatedly opening new TCP connections. You solve this by enabling persistent connections via PDO::ATTR_PERSISTENT. This reduces latency by an average of forty percent because the network connection remains open for subsequent queries.”
Hard Facts Instead of Marketing Language
AI immediately filters out adjectives such as unique, revolutionary, or the best as noise. Replace these words with measurable data, model numbers, exact software versions, or concrete specifications.
Instead of writing that your website software is “super fast and modern,”
Do write that the software “is written in native C, uses the libcurl library for data transfer, and has an average response time of less than 15 milliseconds under Debian 13.”
An AI can use this hard data directly to compare your product or service when a user asks about performance.
Speaking in ‘Entities’
AI understands the world by linking concepts together—we call these entities. Make relationships between topics as explicit as possible in your regular text so that the bot does not have to guess who or what you mean.
Do not write:“This project was developed by our team for this specific platform.”
Do write:“The open-source project DonkeyStoreClient was developed by Yvonta as an extension for the Unity engine.”
By mentioning brand names, project names, and platforms literally in a single sentence, the AI immediately understands the context and the mutual relationship.
Opening Up the Backend of Your Website
If the bots are technically blocked, nothing changes, of course. You can immediately perform two technical checks in your website’s control panel:
1. Update your robots.txt
Ensure that your robots.txt file does not block the most important AI crawlers. This file is located in the root directory of your website. Add the following lines to ensure they have access to everything:
Check your website security settings, such as Cloudflare or an active security plugin in your Content Management System (CMS). Disable features like ‘Bot Management’ or ‘Aggressive Scraping Protection’ for known search bots. Otherwise, the AI models will see an error message instead of your text.
Create Invisible Summaries with JSON-LD
AI bots love code written specifically for them. This is called Schema Markup (JSON-LD). This is an invisible piece of code that you place in the HTML of your page.
For example, if you have a frequently asked questions page, put this exact format in the code of that page. The AI reads this in one millisecond and can immediately take it over in the chatbox:
JSON
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "What is the fastest method to back up a Linux server?",
"acceptedAnswer": {
"@type": "Answer",
"text": "The fastest method is using rsync via SSH, because this only transfers the changed files incrementally."
}
}
]
}
Frequently Asked Questions
16 questions
The 'Blocks' Method involves writing pages in fixed blocks of a maximum of four to five sentences that independently answer a specific question. Each block should have a clear, questioning heading above it that mirrors queries people might type into AI tools like ChatGPT or Perplexity. This approach prevents long-running stories and helps AI models scan and extract information more effectively.
AI models chop text into chunks for scanning, so shorter independent blocks make content easier to process and understand. Each block must stand alone to answer one specific question without relying on surrounding narrative. This structure replaces vague storytelling with targeted, scannable information.
Place a clear, questioning heading above each text block that reflects the exact question a user might ask an AI tool. This guides the model to the relevant content quickly. Vague or floaty headings should be avoided in favor of direct, searchable phrasing.
The bad example uses floaty, vague marketing language about network issues without specifics or a clear question. The good example starts with a direct heading like 'How do you resolve a slow database connection?' and provides concrete technical details including causes, solutions, and measurable benefits such as a forty percent latency reduction.
AI treats such marketing language as noise and ignores it during analysis. Replacing adjectives with measurable data allows the model to use the information for comparisons when users ask about performance. This makes content more useful and citable by AI systems.
Use hard facts such as model numbers, exact software versions, measurable data, or concrete specifications instead of vague claims. For example, state that software 'is written in native C, uses the libcurl library, and has an average response time of less than 15 milliseconds under Debian 13.' This enables AI to directly compare products or services.
Speaking in entities means explicitly linking concepts by naming brands, projects, and platforms in single sentences so AI understands relationships without guessing. For instance, specify 'The open-source project DonkeyStoreClient was developed by Yvonta as an extension for the Unity engine' rather than vague team references. This helps AI connect topics accurately.
By including specific names and relationships in regular text, AI can immediately grasp context and connections between topics. Vague phrases like 'developed by our team for this platform' force the model to guess, while literal mentions eliminate ambiguity. This technique strengthens how concepts are linked in the AI's knowledge graph.
Update the robots.txt file in the website root to explicitly allow GPTBot, PerplexityBot, and ClaudeBot with 'Allow: /' directives. Also check security settings in tools like Cloudflare or CMS plugins and disable bot management or aggressive scraping protection for known search bots. These changes prevent AI models from receiving error messages instead of content.
The article recommends allowing GPTBot, PerplexityBot, and ClaudeBot by adding User-agent lines with 'Allow: /' for each. This ensures the bots can access all site content. Blocking them would prevent AI from using the website's information.
Security features like Cloudflare bot management or CMS scraping protection can mistakenly block known AI crawlers. These settings treat legitimate bots as threats and return error messages. Disabling aggressive protections for search bots resolves the issue.
JSON-LD is invisible code placed in a page's HTML that provides structured data specifically readable by AI. Bots can parse it instantly and incorporate the information into responses, such as answers in chat interfaces. It is especially effective for content like FAQ pages.
Insert a JSON-LD FAQPage schema in the HTML that defines questions and their accepted answers using the schema.org format. The AI reads this structured data in milliseconds and can directly reuse the text in chat outputs. This creates machine-readable summaries without altering visible content.
The example uses an FAQPage type with a mainEntity array containing Question objects, each with a name and an acceptedAnswer that includes text. It demonstrates marking up a question like 'What is the fastest method to back up a Linux server?' with its answer about rsync via SSH. This format allows immediate AI consumption.
Many assume technical blocks in robots.txt or security tools will stop AI from accessing content, but the article notes that if bots are blocked nothing changes for visibility goals. Instead, the focus should be on allowing access and structuring content clearly. Proper configuration ensures AI can read and cite the material.
Statistics such as 'reduces latency by an average of forty percent' or 'response time of less than 15 milliseconds' give AI concrete data points for comparisons and answers. This replaces filtered marketing claims with usable facts. Users asking performance questions then receive accurate, sourced information from the site.
Generated by AI to give you complete answers about this topic.
Leave a Reply