Your website usually doesn’t appear in AI answers because of nine mistakes: not giving a direct answer up front, accidentally blocking the AI crawlers, hiding content behind JavaScript, not using structured data, publishing with no author or E-E-A-T signals, writing generic claims with no data, a structure with no headings or lists, not answering questions the way people actually ask them, and having a brand that’s inconsistent or invisible beyond your own site.
The way people search is changing. Many no longer type loose keywords into Google: they ask ChatGPT, Perplexity or Gemini a full question and keep the answer they’re given, without visiting any website. Gartner predicts that traditional search engine volume will drop 25% by 2026, precisely because of the push from AI chatbots and virtual agents.
The good news: showing up in those answers isn’t magic or a secret algorithm. Language models pick as their source the content they can read, understand, extract and verify easily. If your site fails on any of those four points, it’s left out. These are the nine mistakes I see again and again on freelancer and small-business websites, ordered from most to least common.
The 9 mistakes that keep your website out of AI answers
You don’t give a direct answer up front
If your page opens with a long introduction and the real answer only arrives in the fourth paragraph, AI won’t find it. Models preferentially extract the first sentences when those answer the question completely and on their own.
Your robots.txt blocks the AI crawlers (without meaning to)
Many sites block the bots that feed AI by default and never realize it. If those crawlers can’t get in, your content simply doesn’t exist for them.
robots.txt and make sure you’re not blocking GPTBot and OAI-SearchBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot and Google-Extended, unless you want to exclude them on purpose.
Your important content depends on JavaScript
If the key text only appears after JavaScript runs in the browser, many AI crawlers never see it. To them, that page is half empty.
You don’t use structured data (schema)
Schema is how you tell machines what each thing is: who you are, what service you offer, what question each block answers. Without it, AI has to guess and often gets it wrong or ignores you.
Organization, Article and FAQPage. Validate it with the Schema.org validator (as of June 2026, Google’s Rich Results Test no longer validates FAQs).
You publish with no author or E-E-A-T signals
AI prefers sources with a face and a name. Text with no author, no credentials and no “About” page behind it conveys little trust, and trust is exactly what a model weighs before citing you.
Person schema.
You fill the text with generic claims and no data or sources
“Many companies trust us” says nothing citable. Models pick content that provides concrete data, with figures and a source, because that’s what they can safely reuse in an answer.
The structure can’t be extracted: no headings, lists or tables
A wall of text with no hierarchy forces AI to work harder, and it almost always goes for a cleaner source. Clear headings, numbered lists and comparison tables are exactly what gets extracted best for an answer.
You don’t answer questions the way people ask them
On AI, people don’t search “web maintenance price”: they ask “how much does it cost to maintain a website per month?”. If your content doesn’t capture those natural, complete phrasings, it doesn’t match what’s being asked.
Your brand is inconsistent or invisible beyond your own site
Models cross-check information from several sources to decide whether you’re trustworthy. If your name, address or details change from one place to another, or if you don’t appear anywhere else online, they don’t find enough signals to cite you.
Summary: the mistake, the signal you send, and the fix
A quick view to audit your site at a glance.
| Mistake | What it tells AI | How to fix it |
|---|---|---|
| No direct answer | “I can’t find what this page answers” | 40-60 word quick-answer block up front |
| Robots.txt blocks AI bots | “I can’t get in” | Allow GPTBot, ClaudeBot, PerplexityBot, Google-Extended |
| Content behind JavaScript | “This page is almost empty” | Serve key content in the initial HTML |
| No structured data | “I don’t know what anything is” | Validated Organization, Article and FAQPage schema |
| No author or E-E-A-T | “I don’t know if I can trust this” | Real author, bio with credentials and Person schema |
| Generic claims | “There’s nothing I can cite” | Concrete data, with figure, year and source |
| Non-extractable structure | “You’re too hard to read” | Clear headings, lists and tables |
| Doesn’t capture real questions | “You don’t match what I’m asked” | FAQs with questions the way people say them |
| Inconsistent brand | “I can’t find signals to trust you” | Unified NAP, Business Profile and external mentions |
Checklist before publishing any page
Tick these seven boxes and your content plays with an advantage to be cited by AI.
- ✓There’s a 40-60 word direct-answer block up front.
- ✓The first sentence of each section answers its heading.
- ✓There are at least 3 FAQs with a direct answer.
- ✓Schema is implemented and validated with the Schema.org validator.
- ✓The content has an author with a bio and a link to their page.
- ✓There are links to the related service and to 2-3 pieces on the same topic.
- ✓Every concrete figure is backed by a source.
Want your website to show up when AI answers?
I review your site against SEO for AI (AEO) criteria and tell you, mistake by mistake, what’s keeping you out of ChatGPT, Perplexity and Gemini, and where to start.
Frequently asked questions
How do I get my website to appear in ChatGPT?
For your website to appear in ChatGPT, the model needs to be able to read, understand and verify your content: a direct answer at the start of each page, structured data (schema), an identifiable author with experience, concrete data with sources and a consistent brand across the internet. It can’t be bought or guaranteed; it’s earned with reliable, well-structured content.
Why does my competition show up in AI answers and I don’t?
Almost always because their content is easier to cite: it answers specific questions in the first sentences, has authority and external mentions, and uses structured data. It’s not about paying more, but about how the content is built and backed up. Reviewing the nine mistakes in this article usually reveals the cause.
Does SEO for AI replace Google SEO?
No, it complements it. Many of the signals that help you rank on Google (useful content, clear structure, authority, structured data) are the same ones that make you citable by AI. It’s worth working on both at once, not choosing one.
How long does it take to see a website appear in AI?
It depends on your starting authority, but the technical improvements (schema, direct answer, crawler access) can be applied in days. Getting models to cite you regularly usually takes between a few weeks and a few months, depending on your sector’s competition and your external mentions.
How do I know if my website already appears in AI answers?
The direct way is to ask yourself. Put the questions your clients would ask to ChatGPT, Perplexity and Gemini (“best web designer for freelancers in my city”, for example) and check whether your brand appears cited or linked. Repeat the test every month to track progress.
WordPress Expert, SEO & UX Optimization | I help freelancers and SMEs grow their business. | Web Design and Development Specialist for Startups, SMEs, and Personal Projects.
