ChatGPT’s little gift to the publishing world

You are looking to boost your GEO visibility, how can you single-handedly increase the chance of your book being understood and surfaced by ChatGPT? The answer is simple, but it is not more keywords, it is developing a TextType 03 description.

What you might already know is that OpenAI identifies books by their ISBN, and has its own separate description field. It isn't pulled from ONIX directly. The long-form description already sitting in your ONIX record is the obvious place to draw from. Under ONIX 3.0, which is what most publishers use now, TextType 03 is still in use. The element holding that code changed its name between versions, but the code itself (and what it means, a long, unrestricted description) is still current. Meanwhile, the actual text sits in its own separate field within that same ONIX record.

However, it is important to note how ONIX and TextType 03 differ as I have been using the terms interchangeably. While ONIX is your whole metadata standard, within it sit  several different text fields, each for a different job. TextType 03 is simply the code for one of those fields. It is for the long, unrestricted description of the book. So when we discuss improving your TextType 03 description, we mean refining that particular text within your ONIX record rather than changing the record as a whole.

So how would you begin writing such a description?

Well, OpenAI shares a specific guideline for building an exemplar description field. Writing this kind of metadata beckons a different discipline altogether, for it favours specificity over style.

While traditional descriptions might look something like this:

‘A sweeping and unforgettable new crime novel from the award-winning author of xx. Full of shocking twists this novel features one detective and a past that refuses to stay behind.’

A TextType 03 description would look more like this:

‘A contemporary police procedural set on the North Yorkshire coast, where Detective Justine Dice connects a cliff side murder to a teenager’s disappearance in 1985. For readers of coastal and cold-case mysteries, the novel weighs up whether justice is even possible when family loyalty is challenged in court testimony. Alternating timelines expose the gap between what happened and what the town remembers’

More specific descriptions like this most effectively lead us to improved levels of GEO visibility. The metadata must contain factual semantic information about what the book is, what it is about, its audience, and what questions it might answer.

What is it?
Historical fiction / psychological thriller / introductory economics textbook.

What is it about?
The actual subjects, themes, people, places, events and issues.

Who is it for?
Readers interested in Tudor history, fans of domestic psychological thrillers, undergraduate economics students, parents of children aged 8–12, etc.

What questions might it answer?
Climate change, grief, Roman Britain, ADHD parenting, investing for beginners, Mediterranean cooking, and so on.

What makes it distinctive?
Setting, period, perspective, methodology, expertise, awards, illustrations, recipes, exercises, primary sources etc.

Prompts such as these directly inform how ChatGPT evaluates content and relates it to meaning, user, intent and relevance, moving a step beyond simply matching keyword queries. Soon to be launching at the Frankfurt Book Fair 2026, Inkara’s GEO health check does this for you. This feature, currently in its final BETA testing phase, reads your existing ONIX and through its analysis automatically enhances your description to an ONIX TextType 03 standard.

Built through the descriptions that generative engines need to understand the book, this description stands alongside, and crucially does not replace the publisher’s beautifully written consumer blurb. While you bolster your GEO visibility, you also retain your sophisticated descriptions for cover and promotional objectives.

Making every book understandable to AI, Inkara’s new feature looks to lend a helping hand with publishers looking to optimise their product’s visibility in AI-generated search engines. Contact us to sign up to our free trial now and expect a bundle of free packages built to ensure your books reach their discoverable potential!

References

https://ns.editeur.org/onix/en/153/03

‍

‍