Last reviewed by Emma Hartley — May 2026
"Can you get ChatGPT to recommend my book?" is fast becoming one of the most-asked questions in indie publishing. The honest answer: you can't make a model recommend you, but you can give it every reason and every signal to. Here's exactly how.
First, understand what the model is doing
When a reader asks for a recommendation, ChatGPT answers from two places: what it learned in training, and what it finds via live web search. To be recommended you need to be recognisable in at least one of those layers — ideally both. That means becoming a clear entity the model can identify, and leaving signals in the places it looks. Everything below serves one of those two goals. (For the bigger picture, see our guide to AI book discovery and AEO.)
Step 1 — Become a recognisable entity
Models are far more confident recommending a book they can pin to a known author identity.
- Complete your Goodreads author profile and Amazon Author Central page. List every title, link them, write a real bio. See Goodreads for authors.
- Create a Wikidata entry for yourself and your books where the facts support it — it's the structured knowledge base many models lean on, and it's free to add.
- Earn a Wikipedia mention only where genuinely warranted — never fabricate notability; models (and editors) punish it.
Step 2 — Make your website machine-readable
Your author website is the one source you fully control. Add schema.org structured data — Book for each title, Person for you, Organization for your imprint. It's invisible to readers and tells machines precisely who wrote what. A book page with clean schema, a clear synopsis and links to retailers is exactly what a web-searching model wants to quote.
Step 3 — Write a description a machine can understand
A book description that opens with atmosphere — "In a world where nothing is as it seems…" — gives a model nothing to grab. Lead instead with the concrete: subject, genre, who it's for. You can be evocative in sentence two. Sentence one should let a machine (and a skim-reading human) know what this book is in plain words.
Step 4 — Leave signals where models look
Models weight some sources heavily. Reddit is among the most-cited corpora across ChatGPT, Claude, Perplexity and Gemini. Genuine participation in your genre's subreddit — answering questions, being mentioned in recommendation threads — feeds directly into what the models repeat. Do it as a real community member, not a spammer (see Reddit & forum promotion). The same logic applies to reputable "best books on X" lists and book-blogger coverage.
Step 5 — Gather concrete reviews
Reviews that describe the book in plain language — what it's about, who'll like it — give models more to quote and more confidence to recommend. A wall of "loved it!" helps sales; a handful of descriptive reviews helps machines.
Step 6 — Measure, fix, re-test
You can't improve blind. Run the free AI Discovery Score to see what each of the four models currently knows about your book. If you want the fixes done for you — a ready-to-paste listing rewrite, the exact schema snippets, a training-data check and a prioritised roadmap — the full AI Discovery Audit is £29.99. Then make your changes and re-test in a few weeks; entity and web signals take time to propagate.
What not to bother with
- Don't try to "prompt-inject" the models or stuff hidden text — it doesn't work and risks your reputation.
- Don't fake a Wikipedia page or buy spammy mentions; models increasingly discount low-trust sources.
- Don't expect overnight results — training data updates on its own schedule; web signals move faster but still take weeks.
The takeaway
Getting recommended by ChatGPT isn't a trick — it's author-platform fundamentals aimed at machines: a clean identity, structured data, a clear description, genuine community presence and concrete reviews. Do those, measure with the free score, and you'll steadily become a book the models reach for.
Frequently asked questions
Can I pay ChatGPT to recommend my book?
No. There's no ad slot inside a recommendation. You earn it by being a recognisable entity with clear signals — that's what the AI Discovery Audit helps you build.
How long does it take to show up?
Web-search signals can shift in weeks; training-data recognition updates on the model's own schedule, so allow a few months and re-test periodically.
Does this work for fiction?
Yes — "books like X" and genre recommendation queries are common. Comp-title clarity and genre signals matter; see find your comp titles.
What's the single highest-impact step?
Becoming a recognisable entity: complete Goodreads, add Wikidata and website schema. Everything else compounds on top of that.
Related guides
- AI book discovery: the new SEO for authors
- Does ChatGPT recommend books?
- Author website essentials
- Goodreads for authors
External references
- Schema.org Book type — structured data for books.
- Wikidata — add a machine-readable entry for your author identity.
About this guide
A practical checklist for authors who want their books surfaced by AI assistants.