ChatGPT’s new feature lets readers snap a photo of a stack of books and get tailored book recommendations based on the books in the photo. Other LLMs are quickly adopting this feature as well.
There’s a lot of drama around AI, but authors are often thinking about it the wrong way. Stop thinking of AI as a tool to write your book. It’s a tool to sell your book.
Soon, there will be two types of authors: those whose books AI recommends and those wondering why their sales plummeted. Even if you avoid AI, your readers are already using it to find their next read. They used AI this morning to discover new books. Did they find yours? If not, you might be wondering how to get ChatGPT to recommend your book. The answer is AI optimization or AIO.
Where AI Finds Book Information
How do you get Grok, Gemini, Claude, and ChatGPT excited about your book?
Surprisingly, it starts with your website.
Major large language models (LLMs) train on publicly available web data. Social networks like Facebook often block competing bots from harvesting data, making social media tricky for AI. Some bots have access to certain networks, but all have access to the open web. The best place to post public information about your book is on your author website, where you have full control.
While there are other elements of AI optimization, let’s focus on the website first because it’s within your control.
Optimizing Your Website for AI
AI bots function much like search engine bots. If you’ve followed my advice on search engine optimization (SEO), you’re already close to AI optimization. If you need help, take my free course, How to Make Your Author Website Amazing.
Many authors don’t think their website is very important. A few years ago, websites were less critical because readers searched on Amazon or Google. Now, readers trust AI more and search engines less. So, here are some tips to optimize your website for AI bots:
Use WordPress
WordPress is the world’s most popular website platform because it’s highly optimized for search engines and AI bots are similar. It’s easy to make WordPress bot-friendly with a robots.txt file, an XML sitemap, and fast response times. Some robots.txt files slow bots down, which you don’t want for AI optimization. If you’re using WordPress, this is easy, especially with the Yoast SEO plugin, which I recommend. It’s free and handles these settings for you. If you’re not on WordPress, it’s trickier, and you’ll need to figure it out.
Incorporate Structured Data
AI bots love structured data, particularly Schema.org metadata. This might sound complex, but Yoast SEO has a schema tab that simplifies it. Just explore the settings, and you’ll see how easy it is to make your pages more optimized. Structured data helps AI understand your content better.
Use Contextual Content
Bots prioritize contextual content like H1, H2, and H3 headings, bullet points, internal hyperlinks, breadcrumbs, and semantic data. Yoast SEO assists with these elements. While this might feel like a Yoast commercial, the free version is sufficient for most authors.
Think Beyond Google
Historically, Google dominated SEO, but AI uses other search engines. ChatGPT and Perplexity use Bing, Claude uses Brave, and Gemini uses Google. Grok uses a proprietary blend of sources, possibly including Bing, Google, and Brave in its DeepSearch mode. Optimize your website for AI, not just Google.
Optimizing Book Pages
The most critical page for training AI about your book is the dedicated page on your website for that book. Many authors make the mistake of lumping all their books on one page, which creates an SEO mess. Instead, create rich, detailed book pages for each of your books. Include more information on your page than Amazon has on its page. Include:
- Discussion guides
- Sample chapters (AI loves these, as some can’t access full texts)
- Editorial reviews
- Audiobook resources
- Press releases
- ISBN and AISN numbers
- Metadata (Publication Year, Publisher, Publisher City, if it is on a Copyright Page put it on your book page)
- Links to interviews about the book
- Links to all sites selling the book (Amazon, Barnes & Noble, Kobo, etc.)
- Maps, art, bonuses
- Trope lists
Tropes are especially important, as Amazon often lacks this information. Explicitly listing tropes on your book page helps AI recommend your book to readers searching for those themes.
AI Optimized Reputation
Beyond your website, your reputation also influences AI recommendations. LLMs assess “cultural relevance.” LLMs also look at “associations,” which connect similar items. To leverage this, associate your book with popular books your readers love. Here’s how:
Guest Blogging and Podcasting
Guest blog on other authors’ websites and appear on podcasts to gain inbound links. These links boost your site’s cultural relevance and amplify your optimization efforts. ChatGPT reads public podcast transcripts, which are increasingly common thanks to AI transcription. Podcasting is now a powerful way to influence LLMs.
Reviews and Citations
LLMs consider Amazon, Goodreads, and Audible reviews. More reviews are better. For nonfiction, citations from other books boost your cultural relevance. Check out my episodes about getting more reviews if you need help.
- How to Get More Book Reviews With Derek Doepker
- How to Get More 5-Star Reviews With Beta Readers, Editors, and Launch Teams
- How to Get More Book Reviews with Jim Kukral
AI Optimized PR
A big part of cultural relevance for large language models (LLMs) comes down to public relations and what the media says about them. This raises a philosophical question: how do LLMs determine what is true? With so much information on the internet, how does an AI decide what to trust?
Most AIs approach truth in a similar way: they tend to trust mainstream media over alternative news sources. For example, if the Wall Street Journal and a blogger disagree, the AI will likely side with the Wall Street Journal. This makes the mainstream media’s perspective on your book more important than ever. While fewer humans may read these news sources directly, AIs are paying close attention to them.
The exception to this trend is the new Grok 3.5, which claims to operate from first principles rooted in math and science. While we don’t yet know the full details of these principles, they focus on the fundamentals of scientific laws and mathematics.
In the future, we might see AIs with a “worldview selector,” allowing users to choose philosophical or religious lenses like Judeo-Christian, Muslim, Hindu, atheist, or humanist to guide the AI’s reasoning. We’re not there yet, though.
Currently, many AIs take a shortcut by heavily relying on mainstream media more than human readers do. However, it’s not always the outlets you’d expect, like The New York Times. In fact, The New York Times is at odds with many AI models, blocking their bots and even suing them in court, which limits its influence on AI training data.
Instead, AIs often rely on local television stations and news sites. Why? These sources typically don’t have paywalls or restrictive gateways, unlike many local newspapers, which are often owned by big conglomerates that may block AI access or have licensing deals. For example, News Corp, which owns the Wall Street Journal, has a licensing agreement with ChatGPT, making its content particularly influential for that model but less so for others without similar deals.
Take my local station, KXAN, for instance. Every article is freely available online, so AI bots can easily read and train on that content. This makes local news a surprisingly powerful player in shaping AI perspectives. Sometimes, they’re even more than major outlets like The New York Times.
Mainstream media coverage also helps you secure a Wikipedia page, which every AI trains on. Wikipedia synthesizes mainstream media, so the more press you get, the longer your page. Check my episodes on PR for help getting journalists to cover your book.
- How You Can Create Massive PR for Yourself, Even If You’re A Complete Novice
- How to Conduct Your Own Media Tour (without hiring a PR Firm) With Dr. Saundra Dalton-Smith
- How to Nail a Hostile Media Interview
AI Optimized Book Metadata
Don’t alter your book’s content for AI, but optimize its metadata, which is data about your book. Every LLM emphasized metadata’s importance. In the U.S., Bowker controls ISBNs, where you set keywords, descriptions, publication dates, and more. Use:
- Keyword-rich titles, subtitles, and series names
- Search-optimized book descriptions (my patron toolbox has generators)
- Consistent keywords and categorization
- Appropriate BISAC codes
- Publish Widely (The more sites that list your book, the more data for the LLMs). (Barnes & Noble, Kobo, IndieBound) to create more pages for AI to train on.
- Make sure your paper book is in ALL the databases. (Consider IngramSpark for your wide books.)
IngramSpark is great for wide distribution (but I don’t recommend them for your first book). Wide ebooks also increase your book’s online presence. Listen to my metadata episode for a simple breakdown from a top expert.

AI Optimized Blog
My recommendations for blog optimization are available for Patrons of the Novel Marketing podcast only. You can become a patron today.
AI Optimized Social Media
Most social networks are inaccessible to AI due to data restrictions. However, Reddit and Goodreads are the exception. All AI models train on these because their data is publicly available.
Goodreads is critical for authors, as AI explicitly uses its reviews, ratings, and lists. Its schema.org metadata is bot-friendly, and it serves your book’s data to AI on a silver platter. Ensure your book is fully optimized on Goodreads. Listen to my episode on How to Use Goodreads to Promote Your Book. Reddit is influential, but it is less influential than Goodreads when it comes to authors and books.
Other platforms have limited impact:
- TikTok and Bluesky: Minimal influence on AI rankings
- YouTube: Influences Gemini primarily
- X: Exclusive to Grok
- Facebook/Instagram: Tied to Llama, which almost no one uses.
- LinkedIn: Connected to Copilot, possibly ChatGPT
Focus on investing your time on Goodreads and metadata over social media platforms like X, where content volume dilutes impact.
You’re much better off spending your time on high-impact activities like reading quality books, refining your book’s metadata, and pursuing PR opportunities. Getting the traditional press to talk about you can make a big difference.
Take a close look at your website and ensure its structure is solid. If you’re using Yoast, aim to get all those dots green. Consider hiring a professional to improve the technical elements.
At AuthorMedia Social, we have a job board where you’ll find webmasters who specialize in hourly work for authors. If you built your website using my course, just a few hours with a webmaster can lead to significant improvements. I don’t charge anything for the job board, but I also don’t make any guarantees.

Mistakes to Avoid
- Keyword Stuffing: Don’t overload your book or website with keywords; it won’t help and may hurt.
- Changing Book Content: Keep your book’s content for human readers. Optimize metadata and web pages for AI, not the book itself.
- Expecting Instant Results: AI models take time to train. Even a perfect website may not influence AI for 6–12 months. Grok updates quickly, Claude slowly, with ChatGPT and Gemini updating sometime in between.
The way we create web pages is changing. When you build a page on your website, you now need to consider not only the humans who will read it but also the AI bots scanning it. More and more, tools like Gemini are rewriting your web pages and serving that content to Google searchers instead of directing them straight to your site.
Your novel is still just for human readers, but your website is now for bots, too. At Author Media, we’ve been learning this for years. We’ve developed AI Thomas, an AI engine you can use to ask book marketing questions. It’s trained on my podcast episodes, so its “source of truth” is what I, Thomas Umstattd, say about book marketing.
AI Thomas assumes I know my stuff and does a great job synthesizing answers for authors. We’ve been building webpages specifically to train AI Thomas, helping it provide better answers. We’ve been at this for years, learning a ton along the way, and I hope some of those lessons can help you, too.
Free Resources
For more help, explore these free resources:
- How to Make Your Author Website Amazing. (Part 1: building a website; Part 2: improving it with SEO)
- How to Use Metadata to Sell More Books episode with a top expert
- How to use Goodreads to Promote Your Book episode for optimization tips
By following this advice, you’ll position your book for AI recommendations as readers increasingly ask ChatGPT, Grok, or Claude, “What book should I read next?” Get AI to recommend your book, and your sales will soar.
Thanks for covering so much information on this topic. I’ve been spending more time lately on improving my SEO on my site, especially since I want to start selling my books directly from it. You mention using Yoast for SEO in the article. I have AIOSEO on my site and recently upgraded from the free version to get a few extra benefits. The schema section still has me vexed (it may just be me). Do you recommend one SEO product over another?