The future of publishing is direct sales
Media Transformation Summit: Sarah Arbuthnot on how smart metadata, SEO, and direct channels are reshaping publisher revenue
As AI increasingly shapes how readers find books, the Supadu president argues that publishers need to rethink metadata, websites and search strategies to ensure their titles remain visible, trusted and recommended. Sarah was one of the speakers at the Media Transformation Summit, a digital conference organized by Publishing Signal, powered by knk, in partnership with the Digital Publishing Report (DPR).
Published: 28.7.2026 | Photo / video: Magnific
During the Media Transformation Summit, Sarah Arbuthnot, President of Supadu, led the session Finding Readers in the AI Era: Why Metadata, SEO & GEO Matter for Trade Publishers, highlighting the critical shifts in book discoverability in the age of AI and the publisher's evolving role in helping readers find the right books.
The panel encouraged attendees to look beyond their own organizations. Even the best editorial workflows, production processes, and beautifully designed books achieve very little if the right reader never discovers the right title. As discovery expands far beyond the traditional search box, publishers face a new challenge: how can books remain visible, understandable, and relevant across an increasingly fragmented digital landscape?
For Arbuthnot, the answer begins with understanding that the future of publishing is direct sales. "You need to think like retailers now. You're retailers, you're stores, you're selling products to consumers. Those consumers are readers."
According to Arbuthnot, publishing is undergoing the same transformation that many other industries experienced years ago. Readers no longer discover books in a single place. Instead, recommendations now emerge from online retailers, social media, publishers' websites, search engines increasingly powered by AI, email marketing, and recommendation platforms.
This fragmentation creates new challenges for publishers, many of whom are dealing with shrinking budgets, legacy metadata, changing revenue models, and rapidly evolving consumer behavior.
Why metadata matters more than ever
If discovery is becoming more fragmented, publishers need a stronger foundation.
For Arbuthnot, that foundation is metadata—not simply as a cataloging requirement, but as the information that allows search engines and AI systems to understand what a book is about. At the same time, the publisher's website has become equally important.
Rather than serving simply as a catalog or storefront, publisher websites should become the most authoritative source about their own books. As AI systems increasingly rely on trusted sources to generate recommendations, websites are becoming the reference point that feeds discovery across multiple platforms.
"Your website should be the most trusted version of your content, and metadata is a huge part of your content, because the AI systems are constantly looking for authoritative sources—and we build a lot of our websites in Shopify," she explains.
The insight is that good metadata alone is no longer enough. Publishers need to move beyond static catalog information toward what she describes as dynamic metadata—metadata that evolves alongside reader behavior, cultural trends, and changing search patterns.
That's because readers increasingly search by mood, themes, or emotional expectations rather than traditional categories, using prompts such as "cozy fantasy with found family" or "dark academia for teens." Social media trends can quickly transform niche expressions into major discovery categories, as happened with "romantasy."
This evolution also changes the relationship between metadata, SEO, and the emerging concept of GEO (Generative Engine Optimization).
Metadata tells AI what a book is. SEO helps readers find it through search engines. GEO goes one step further by helping generative AI recommend the book in response to more natural, conversational questions. "AI is not just returning pages, it's building answers," she points out.
Speaking the reader's language
One of the biggest obstacles, Arbuthnot suggested, is that publishers often describe books very differently from the way readers search for them.
Understanding how readers search allows publishers to enrich both static and dynamic metadata, making books easier for AI systems to surface and recommend. "The performance of the metadata is coming from using it dynamically to improve discoverability, improve conversion, improve targeting," Arbuthnot said. "It's not remaining static; it's kind of growing and building its own visibility."
Where should publishers start?
Asked what publishers with limited budgets and thousands of titles should prioritize first, Arbuthnot's answer was straightforward. "Doesn't matter if your metadata is in a spreadsheet, a FileMaker database, or a title management system—do a metadata review," she said.
She also encouraged publishers to strengthen their own websites by creating richer book descriptions, dedicated FAQ pages for books and authors, stronger category structures, better internal search functions, and technically optimized pages.
For Arbuthnot, publishers already possess most of what they need. The challenge is no longer producing more content, but making existing content easier for both readers and AI systems to understand, connect, and recommend.
Sarah Arbuthnot is President of Supadu, the metadata, website and ecommerce platform serving publishers worldwide. With a background in digital strategy and global ecommerce for major brands, she helps publishers grow direct-to-consumer revenue, strengthen their digital presence, and make books in every format more discoverable and accessible to readers.
