Free Ways to Learn AI SEO: A Self-Study Course Built From Primary Sources

You don't need to pay for AI SEO training. The platforms that build AI answers publish how their systems treat websites, and that documentation is free. It's also the only material that changes when the platform changes.

Most paid and free courses repackage older SEO advice. Some of it is now contradicted by the vendors themselves. Google's guide to optimizing for generative AI features, published in July 2026, says you don't need new machine-readable files, AI text files or special markup to appear in Google Search, including its AI features. A course that still teaches those as levers is teaching you something Google has publicly rejected.

So build your own syllabus from three kinds of source:

• Platform documentation. Google Search Central, OpenAI's crawler page, Anthropic's and Perplexity's crawler docs, Meta's web crawler page and the Bing Webmaster blog.
• Open research. Papers on arXiv that test GEO tactics, read critically.
• Your own site. The raw HTML, robots.txt and server responses you can check today.

If you want a structured version with lesson order, the free AI search course syllabus lays one out. This page explains why that order works and what to skip.

Every AI answer passes through up to four stages. Courses that blur them sell tactics for the wrong stage. Learn these before anything else.

1. Parametric. What the model knows with no web access, frozen at a training cutoff. Nobody can change this quickly. OpenAI's model docs list a February 16, 2026 cutoff for the GPT-5.6 family and August 31, 2025 for the Instant path.
2. Access. Whether a crawler may fetch and store your page. This is where publishers have real control.
3. Retrieval. Whether your page enters the model's context for a given question. The platform decides, and it rewrites the user's question into its own search queries first.
4. Generation. Which retrieved pages the model uses and cites. The model decides.

Why this matters for your study plan: almost all published GEO experiments sit at stage 4. A critical survey of 45 GEO studies (arXiv, July 2026) concluded that already-retrieved content can change how it gets cited, but no reviewed technique shows a stable, cross-platform effect on being discovered in the first place. That single finding tells you where to spend your reading time: stage 2, where the documentation is clear and the controls are yours.

Read these in the order of the table. Each one is free and published by the operator or by researchers with open methods.

The Google entries are the deepest set. If Google is your main channel, the guide to learning Google's AI search guidance walks through those pages one by one. The difference between Google's two crawler tokens trips up many teams, so read the comparison of Google-Extended vs Googlebot alongside Google's crawler page.

A good course is honest about this split. Most aren't.

In your control

• Rendering. Vercel's crawler study found that ChatGPT and Claude crawlers download JavaScript but don't run it, so content that appears only after JavaScript runs may be invisible to them. Googlebot does render.
• Per-bot robots.txt rules. OpenAI states each of its crawler settings is independent. Blocking GPTBot stops training use. OAI-SearchBot governs whether you appear in ChatGPT search answers.
• Network reachability. OpenAI's help centre says eligibility needs OAI-SearchBot allowed in robots.txt and a host or CDN that lets OpenAI's published IP ranges through. A firewall can block a crawler that robots.txt allows.
• Snippet controls. Google and Bing document noindex, nosnippet, max-snippet:0 and data-nosnippet as ways to keep content out of AI answers. Templates sometimes apply them by accident.
• Stated facts. A specific price, number or one-line definition on a fetchable page gives an engine a sentence it can quote.

Not in your control

• Whether an engine searches at all for a given question. ChatGPT search fires at the model's discretion, and Claude.ai search is a user toggle.
• The real queries the engine runs. OpenAI says ChatGPT rewrites the prompt into its own targeted queries.
• Your position inside an answer. Google's AI report in Search Console shows impressions only, with no position data, as of October 2026.
• What the model learned in training.

The practical lesson: study the controls, then audit them. Stop studying tactics that promise outcomes nobody can control.

This plan uses only free material and your own website. Each week ends with something you check, not just something you read. For a deeper version aimed at individuals, see how to learn AI SEO from primary sources. For rolling this out across a marketing team, see AI SEO training for teams.

Week 1: the stages and Google's rules

1. Read Google's AI features page and its July 2026 optimization guide end to end.
2. Write down the four stages from memory and one control for each, if one exists.
3. Check your key pages are indexed in Google Search Console and not marked noindex or nosnippet.

Week 2: crawlers, bot by bot

1. Read the OpenAI, Anthropic, Perplexity and Meta crawler pages.
2. Open your robots.txt and list every AI user agent it names. Sort them into training bots and search bots.
3. Confirm you haven't blocked a search bot while trying to block training. A rule set that separates the two looks like this:

User-agent: GPTBot
Disallow: /

User-agent: OAI-SearchBot
Allow: /

The robots.txt Checker for AI Crawlers resolves these rules per bot if you want a second read.

Week 3: rendering and reachability

1. Read Vercel's AI crawler post.
2. Fetch a key page with JavaScript disabled. Check that your pricing, product names and headings are in the raw HTML.
3. Look at your server or CDN logs for AI crawlers and the status codes they got. A free AI bot log check shows how to read them. A 403 to a search bot means that bot can't reach you.

Week 4: research and measurement

1. Read the abstracts of C-SEO Bench and the 2026 GEO survey. Note that the tested pages were already in the model's context.
2. List the buyer questions you care about using a free prompt research method.
3. See what AI platforms actually say for those questions, and note any wrong facts about your brand.

Free material is uneven. These are the errors that cost the most time.

• Trusting old posts. Plenty of free guides still teach schema, E-E-A-T and domain authority as AI levers. If a source says so, it's out of date, not new.
• Learning schema as a citation tactic. Google says structured data isn't required for generative AI search. Ahrefs' matched-control study of 1,885 pages found AI Overview citations fell 4.6% after schema was added. FAQ rich results were deprecated in May 2026. Existing markup does no harm, but it isn't a lever for AI answers.
• Chasing an E-E-A-T score. Google says E-E-A-T isn't a specific ranking factor. There is no score to raise.
• Publishing llms.txt as a fix. Google Search ignores it, and no other engine documents reading it.
• Confusing GPTBot with search. Allowing GPTBot governs training only. ChatGPT search depends on OAI-SearchBot.
• Quoting the 40% GEO figure. That number comes from a lab engine with sources loaded into context, not from live search.
• Assuming rank equals citation. Ahrefs found 37.9% of URLs cited in AI Overviews also appeared in the first 10 results (March 2026). Ranking helps sometimes. It isn't the whole story.
• Ignoring third-party sites. When engines cite other sites instead of yours, being named on those sites is associated with appearing in the answer. Check which sources and which competitors AI recommends for your topic.

Platform behavior changes, so a free course built on primary sources needs upkeep. Courses frozen in a video go stale the same way.

• Date everything you note. Write "as of October 2026" next to each rule. Google's FAQ markup changes and Cloudflare's September 2026 change to default AI crawler settings show how fast a rule can move.
• Re-read the source, not the summary. When a blog claims a platform changed, open the vendor page and check.
• Re-run your checks. Repeat the week 2 and week 3 checks each quarter. A CDN setting can block a crawler without anyone on your team editing anything.
• Watch answers, not just pages. The free checker on this site shows whether AI platforms name you for up to three buyer questions. It's a snapshot, so repeat it after changes.

The goal of self-study is judgement. You should finish able to read a new AI SEO claim and ask which stage it targets, what evidence backs it, and whether you can check it yourself.

Do I need to know traditional SEO before learning AI SEO? It helps, because the foundations overlap. Google says pages need to be indexed and eligible for a snippet to appear in AI Overviews or AI Mode, with no extra requirements. That means crawling, indexing and robots.txt basics carry straight over. You can learn both together by starting with Google's crawling and indexing docs, then adding the crawler pages from OpenAI, Anthropic, Perplexity and Meta.

Are free AI SEO courses from tool vendors worth taking? Some are useful, but check every claim against first-party documentation. Vendors, including this site's publisher, have a commercial interest in the topic. A fast test: if a course says schema, llms.txt, chunking or an E-E-A-T score will get you cited, it contradicts Google's own published guidance. Keep the parts that match vendor docs and discard the rest.

Can I learn AI SEO without technical skills? Yes, for most of it. Reading robots.txt, spotting a noindex tag and comparing a page with JavaScript on and off need no coding. Log analysis and firewall settings are more technical, so you may need a developer for those. Knowing what to ask for is the main skill: which bots are allowed, whether key facts are in the raw HTML, and what status codes AI crawlers receive.

How long does it take to learn AI SEO from free sources? The plan on this page runs four weeks at a few hours a week, but pace is up to you. The reading list is short because primary sources are concise. The slower part is applying it to your own site and fixing what you find. Expect to keep learning after that, because platform documentation changes and the research is still early.

Is there a free way to practise AI SEO on a real site? Use your own site or a test site you control. Fetch a page without JavaScript, read your robots.txt per bot, check your server logs for AI crawlers, and look up what AI platforms say about your brand for real buyer questions. Each check maps to a stage of the AI answer, so you learn the mechanism while finding things to fix.