AI Visibility for Online Course Creators

Course buyers use AI assistants at two points: shortlisting and checking. The shortlist prompt is generic. The checking prompt names you. Here are realistic examples (the course names are hypothetical):

• "What's the best online course to learn Python for data analysis?"
• "Best Figma course for complete beginners under $200"
• "Is the Copy Lab Bootcamp worth it, or is it overpriced?"
• "Does the Copy Lab Bootcamp have a refund policy?"
• "Compare Copy Lab Bootcamp vs Write to Sell for learning email copywriting"
• "What does the Copy Lab Bootcamp cover, and is there a certificate?"
• "Cohort-based product management course starting next month"

The first two decide whether you're in the answer at all. The rest decide whether the answer about you is right. A wrong refund policy or a price from last year's launch can lose a sale the shortlist already won. Test both kinds, and write them in your buyer's words, not your sales copy.

No single engine covers the journey, and each one behaves differently. As of October 2026:

• ChatGPT decides per prompt whether to search the web. When it doesn't, it answers from training data with a fixed cutoff, so a launch price or a new module may simply be missing.
• Google AI Overviews and AI Mode only use pages that are indexed and eligible for a snippet in Google Search. Google says there are no extra technical requirements beyond that.
• Perplexity and Claude run their own search and fetching, with their own crawlers.
• Microsoft Copilot grounds its answers in the Bing index, so Bing coverage matters separately from Google.

One engine can name you while another describes a course you retired. Look at each engine on its own, which is why a blended score can mislead (see whether one dashboard can track every AI engine). For how ChatGPT's search side picks pages, read how ChatGPT search finds sites.

Course facts change more often than most businesses' facts. Launch pricing, early-bird tiers, cohort dates and bonus modules all expire. Each change leaves an old version somewhere. Four mechanisms explain most errors:

1. Training cutoffs. A model answering without search only knows what it saw before its cutoff. Your current price may be newer than that.
2. JavaScript-only content. Vercel found that ChatGPT and Claude crawlers fetch JavaScript files but don't execute them, so they can't read client-side rendered content. A price in a checkout widget or a syllabus in a tabbed accordion may be invisible.
3. Contradictions across your own pages. The sales page says 30-day refund. The terms page says 14. The platform listing says 7. An engine has to pick one.
4. Third-party pages that never updated. Review posts and roundups quote the price they saw when they were written.

The fix is unglamorous. State each fact once, specifically, in plain HTML: the price, the refund window in days, the module list, the certificate yes or no. A specific sentence on a fetchable page gives an engine something accurate to quote. If AI still describes a retired course, see why AI keeps recommending something that no longer exists.

For "best course" prompts, engines often cite other sites instead of yours. The evidence shows a strong association between being named on those sites and appearing in the answer. It doesn't promise an effect, but it tells you where to look. Check:

• "Best courses for X" roundups that rank for your topic. Are you listed, and is the price current?
• Review sites and forum threads where learners discuss your course. See whether reviews affect AI recommendations.
• Course marketplace listings, if you sell there too. They often outrank your own site and carry their own price.
• Podcast and guest-post pages that describe your course from an old launch.
• Affiliate pages quoting a discount that ended.

Where a page is wrong, ask the owner to update it. Where you're missing, earn the mention the normal way: a product worth recommending, and outreach to the people who write those lists.

Many creators sell from a hosted course platform and keep a thin personal site. That makes the platform sales page the page that matters most, and the one you control least. Run these checks before touching content:

• robots.txt per bot. Rules apply per crawler. OpenAI states that sites opted out of OAI-SearchBot won't be shown in ChatGPT search answers. GPTBot only governs training use, and OpenAI says each setting is independent. On a hosted platform you may not control robots.txt, so read what it says.
• Firewall and CDN. A crawler that robots.txt allows can still get a 403 from the firewall. OpenAI's help centre says eligibility requires allowing OAI-SearchBot and allowing traffic from its published IP addresses. Since 2026-09-15, Cloudflare blocks Training and Agent crawlers by default on ad-displaying pages for new domains and free-tier customers, so a site can lose access without anyone editing it.
• Rendering. Load the sales page with JavaScript off. If the price, refund terms or curriculum vanish, most AI crawlers can't see them either.
• Snippet controls. Check for noindex, nosnippet, max-snippet:0 or data-nosnippet on the sales page. Any of these keeps content out of AI answers.

A robots.txt that keeps search access while declining training looks like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

The AI Crawler Checker requests a page as each AI crawler and shows which ones the firewall turns away. If the platform blocks crawlers and you can't change it, put the core facts on a page you do control.

Creators are tempted to publish hundreds of AI-written lessons to "feed" AI search. Google's spam policy defines scaled content abuse as many pages generated "for the primary purpose of manipulating search rankings and not helping users", and names generative AI used this way as its first example. Google's AI content guidance is equally clear that appropriate use of AI is not against its guidelines. The problem is volume without value.

Otterly tested this in 2026 with two new sites of roughly 1,000 AI-generated posts each. Otterly reports that Google removed one after about 14 days and the other after about 90, with no manual action in Search Console. Google's AI features then almost never cited them, while Bing kept them and Copilot kept citing them. It's one vendor-run test on a low-stakes topic, so treat it as an illustration, not proof. The direction matches Google's written policy: losing the Google index means losing AI Overviews and AI Mode too.

Write fewer pages that answer real buyer questions with real detail.

Week 1: see what AI says. Write five buyer prompts from the list above, adapted to your topic. Run them on ChatGPT, Perplexity, Gemini and Google. Note every wrong price, refund term and module. The free checker on this site shows whether engines name you for three buyer questions.

Week 2: fix access. Test your sales page as each AI crawler. Read the platform's robots.txt. Load the page with JavaScript off. Check for snippet controls.

Week 3: fix your own facts. Make price, refund window, syllabus, cohort dates and certificate status identical on the sales page, terms page, platform listing and personal site. Put them in plain HTML. Remove expired launch pricing.

Week 4: fix third parties. Email the owners of roundups, affiliate pages and listings with outdated details. Check Google Search Console and Bing Webmaster Tools separately for index status.

Then re-run the same prompts. Answers vary run to run, so look for patterns. If you sell software alongside the course, the SaaS pricing and access guide covers similar ground.