Quartz 4

sff-destination-strategy

7 min read

Dungeon Books as the SF/F destination: strategy + gap analysis

The bet

The Emporium online storefront unlocks revenue the physical store can’t reach (we’re size-constrained in-store). Our shot is to be the one-stop destination for sci-fi/fantasy — best place to discover, best place to buy, deepest in this niche.

We’re not going to out-Amazon Amazon. We are going to out-care, out-curate, and out-specialize them in this genre. Amazon’s algorithm buries the long tail; we surface it. Amazon pushes Kindle; we curate physical and special editions. Amazon’s recommendations are collaborative-filter slop; ours come from people who actually read the genre.

What the catalog + storefront work this week unlocks

The pipeline (catalog ingest + enrich + import to Medusa) is the foundation but not the product. It gets books into the storefront with clean metadata. The product is what we build on top:

  • Discovery surfaces (subgenre browse, search, recommendations)
  • Curator surfaces (staff picks, BookTok mentions, guest curators)
  • Series/award/special-edition metadata
  • Reviews/ratings (Hardcover, see below)
  • Community + events (the Guild work already in flight)
  • Loyalty/XP (also already in flight)

Everything in catalog-followon-ideas and ipage-as-api feeds this — but those plans are tactical. This is the doc that says why.

Discovery: what SF/F readers actually want

Things Amazon does poorly that we should do best:

Subgenre taxonomy beyond BISAC

BISAC has five fiction-side categories that map to our store (Science Fiction, Fantasy, Horror, Romance, Literary Fiction, Magical Realism, Young Adult, Non-Fiction). Real SF/F readers browse by tighter labels:

  • Romantasy
  • LitRPG (Dungeon Crawler Carl, Cradle)
  • Cozy Fantasy (Legends & Lattes, T. Kingfisher)
  • Dark Academia (Babel, Ninth House)
  • Cosmic Horror
  • Cosmere (Sanderson)
  • Sword & Sorcery
  • Space Opera
  • Hopepunk / Solarpunk
  • Web-novel-to-print
  • Author-specific universes (Dresden Files, Discworld, Foundation)

These are hand-curated taxonomies on top of BISAC. A book can carry multiple. Probably a vibes table joined to products in Medusa, populated semi-automatically (regex over BISAC + title + description) with human override for the cases that matter.

Series management

SF/F is series-heavy and series management is what Amazon does worst. Our book row has a series string field. That’s not enough. We need:

  • A series entity (id, title, author(s), description)
  • Each book linked to a series with order (1.0, 2.0, 1.5 for novellas)
  • PDP shows “Book 4 of 7” with sibling navigation
  • “Reading order” — publication vs chronological where relevant (Discworld, Cosmere)
  • Series completion progress for logged-in users (“you’ve read 3 of 7”)

This is meaningful infrastructure work but is the feature for SF/F.

Award metadata

Hugo, Nebula, Locus, Goodreads Choice, World Fantasy, BSFA. Mostly a small annual dataset, hand-maintained or scraped. Show as filterable badges on PDP and on category pages (“Hugo winners in Fantasy”). High signal, low effort.

Trope / vibe search

Already in catalog-followon-ideas: “books with dragons + slow-burn romance + enemies-to-lovers” using BISAC + description embeddings + LLM. This is the search Amazon literally cannot do.

Read-alikes

Collaborative filtering breaks at niche scale. What works for SF/F:

  • BISAC overlap (decent baseline)
  • Series sibling expansion (you read Throne of Glass → see ACOTAR, both Maas)
  • Hardcover’s “similar books” data
  • Curator picks (“if you liked X”)
  • BookTok mention overlap

Release calendar

SF/F has tentpole releases — Sanderson, Maas, Yarros, Kuang. An indie should be the place to plan around them. Pre-order driving + release-day events.

Curator picks as first-class

  • Staff picks (Carrie + team)
  • Guest curators (local authors, BookTokers, local indie partners)
  • BookTok mentions as a signal (which mentions, who said it, when)
  • Visible on PDPs and as their own browse surfaces

Buying: what Amazon structurally can’t or won’t do

  • Special editions (deluxe, signed, sprayed-edges, sprayed-page-edges). Currently a major BookTok purchase driver. Our CSV already differentiates these by title suffix (“Deluxe Limited Edition”, “Standard Edition”) — needs structuring as a real attribute, not a string match.
  • Author signings + virtual events with niche authors. Web infra needed: events calendar, ticketing/RSVP, livestream embeds.
  • Subscription boxes — monthly curated SF/F. Needs subscription/recurring billing in Medusa.
  • Bundles — “Sanderson Stormlight starter set”, “ACOTAR complete,” “Hugo winners of the decade.” Medusa supports bundle products; modeling work.
  • Pre-order rewards — signed bookplates, exclusive bookmarks, sprayed-edge upgrades. Need attribute support + workflow.
  • Loyalty/XP that means something. Already in flight via the Guild work. Reading challenges, badges, leaderboards.
  • Knowledgeable human curation visible on every PDP — staff pick callouts, curator notes, “we recommend pairing this with…”

Reviews + ratings: Hardcover (not Goodreads)

Goodreads is dead for new builds. Amazon shut new API access in 2020; existing keys grandfathered, no new tokens issued. Only ways in: illegitimate scraping or unofficial reverse-engineered libraries that break monthly. Plus the obvious tension — positioning as the indie alternative to Amazon while depending on Amazon’s data product is bad optics if it ever leaks publicly.

Hardcover is the right answer.

  • Documented GraphQL API (https://docs.hardcover.app/api/)
  • ISBN-keyed lookups
  • Free for personal / small commercial use
  • Built explicitly as the indie/BookTok Goodreads alternative
  • Smaller dataset (orders of magnitude fewer reviews per book) but growing fast in exactly our readership
  • Roadmap actively courts integrations

Plan:

  • Per-book Hardcover lookup keyed on ISBN, cache rating + a handful of top reviews
  • Show on PDP as “X readers on Hardcover gave this Y stars” with a few quoted reviews
  • Refresh cache weekly or on-demand
  • Worth pairing with StoryGraph (also has API, BookTok-aligned) as a second signal — but start with Hardcover only

What this isn’t: trying to BE Hardcover. We surface their reviews to inform purchase decisions, not to compete with their social-reading product.

Gap analysis: what’s in the plans vs what this strategy demands

Already planned (catalog-followon-ideas, ipage-as-api, medusa-inventory):

  • ✓ Catalog pipeline (foundation)
  • ✓ Recommendations engine (BISAC + co-purchase baseline)
  • ✓ Search-by-vibe
  • ✓ Inventory wiring (Medusa stock locations)
  • ✓ Discord bot for back-office
  • ✓ MCP server / AI agent surfaces

Not yet planned but required by this strategy:

  1. Hardcover integration — per-ISBN review/rating cache, PDP display, weekly refresh
  2. Subgenre taxonomy — vibes entity joined to products, semi-auto + human-curated
  3. Series management — series entity with order, PDP sibling nav, reading-order metadata
  4. Award metadata — small annual dataset, badge display on PDP + filter on category pages
  5. Special editions as structured attribute — extract “Deluxe Limited”, “Standard”, “Signed”, etc. from title strings or detail page; surface as variant/attribute
  6. Curator picks system — admin-facing, staff/guest pick with note, visible on PDP
  7. Release calendar — storefront surface + pre-order workflow
  8. Bundle product type — Medusa supports it, need to model
  9. Author signings / events infra — events calendar, RSVP, virtual event embeds
  10. Subscription boxes — recurring billing, curated monthly selections

Suggested order

Roughly priced by leverage vs effort. The first few unlock real differentiation; the later ones are dependencies on demand.

Tier 1 (next 1-2 months) — fundamental SF/F destination features:

  1. Hardcover reviews integration (a week)
  2. Subgenre/vibes taxonomy (1-2 weeks)
  3. Series management (2-3 weeks, more if it touches storefront PDP redesign)
  4. Recommendations engine baseline (already in plan, ~2 weeks)

Tier 2 (months 2-4) — depth + curation: 5. Award metadata + badges (a week) 6. Special editions as structured attribute (1 week) 7. Curator picks system (2 weeks) 8. Release calendar + pre-order workflow (2-3 weeks)

Tier 3 (months 4+) — community + commerce expansion: 9. Bundle products 10. Author signings / virtual events 11. Subscription boxes

Tier 1 is the unfair-advantage stack. By the time those four ship, we’re objectively the better place than Amazon for an SF/F reader to discover + buy. Everything after is depth.

What we’re explicitly not building

  • A social network for readers (Hardcover, StoryGraph, Goodreads already exist; we link out)
  • A free book database for general use (we’re a store, not a public catalog service)
  • Mass-market fiction beyond SF/F (focus is the moat)
  • Anything competing on price (we lose that fight by definition; we compete on curation + depth + experience)

Cross-references

Graph View

Backlinks