Publisher sites usually have deep content and no structure around it. The first build is the author page: a real biography, the complete bibliography with publication dates and formats, links to every edition the press holds, interviews and extracts, and any events. That page can rank for the author's name alongside the retailers, and it feeds every title beneath it. Series pages do the same job for reading order, which is one of the most persistent search behaviors in book buying, and subject or imprint collections gather the long tail that no single title page can hold.
The backlist is where the compounding happens. Titles more than a season old typically have no collection, no internal links and no reason for anyone to reach them, yet together they represent most of a press's catalog and most of its available margin. Grouping the backlist by subject, theme, translator, prize and reading level, then linking those collections from new releases, gives old inventory a route in from search.
Technically, ISBN discipline matters more than anything clever. Editions and formats are variants of one work where the content is the same and separate products where it is not, out-of-print titles keep their URL with an honest availability note rather than being deleted, and Book or Product structured data has to carry the ISBN, format and publication date accurately. Bookshops with a storefront also need location pages and a stock note that does not promise what the shelf cannot deliver.