What happens
when taste
gets a shelf.

The challenge

Print ARC distribution is expensive, review rates are low, and the buzz it does create is scattered across platforms that never convert into pre-orders.

What PaperBoxd will do

Planned, not shipped: pre-release shelves gated to early readers, with reactions posted during the read rather than after finishing. There is no publisher-facing product today.

The evidence

This digital-first, social-proof approach is the model NetGalley has made the industry standard for getting secure digital review copies to reviewers, influencers and librarians ahead of publication. NetGalley has since extended it toward consumers with Booktrovert, letting publishers run direct-to-reader giveaways and discovery campaigns.

The challenge

Video algorithms reward trending audio and fast cuts, which quietly suppresses the deep-dive recommendation content book creators are actually good at.

What PaperBoxd does

Keep the videos, move the canonical list to PaperBoxd — named by hyper-specific mood rather than genre, since a list here is just a shelf with a point of view.

The evidence

Niche, identity-driven creator recommendation is measurably powerful: a UK study reported by State of Digital Publishing found 38% of young people rely on BookTok for recommendations over family and friends, and 68% said it made them read a book they would not otherwise have considered. The same dynamic pushed Colleen Hoover past 20 million copies sold globally, and Penguin Random House's creator campaign for Last Night at the Telegraph Club produced a sevenfold sales increase.

The challenge

Readers discover books socially, then hit friction working out whether their local library has a copy. The intention dies somewhere between the recommendation and the catalogue search.

What PaperBoxd will do

Planned, not shipped: surface holds and instant digital lending inline on the shelf, so borrowing is a tap rather than a separate errand. No library catalogue is integrated today.

The evidence

Demand for frictionless digital lending is enormous and still climbing. OverDrive, which builds Libby, reported 820 million digital library checkouts in 2025 — up nearly 11% on 2024 — with 9.8 million new Libby installs in that year alone.

The challenge

The sharpest drop-off in reading apps happens during a slump. Readers who can't finish a book stop logging entirely rather than admit the streak is broken, and the app loses them along with the honest data.

What PaperBoxd will do

Planned, not shipped: a DNF status beside Reading and Finished, taking one line of reason and staying off your books-read count. There is a DNF tab on the profile today, but it lists everything you have started or queued and not finished — the same books as the procrastination wall — rather than the ones you decided to put down. Nothing marks a book abandoned yet, and there is no reason field.

The evidence

This mirrors how The StoryGraph grew against Goodreads — mood-based stats and non-judgemental tracking, including importing Goodreads did-not-finish shelves wholesale, with machine-learning recommendations weighted on mood rather than star ratings. Its own public roadmap carries community requests for granular DNF metrics so readers can see abandonment rates before committing.

The challenge

Independent bookshops compete with a recommendation engine that runs 24 hours a day on somebody else's platform, and they see none of the signal it generates about their own neighbourhood.

What PaperBoxd will do

Planned, not shipped: give the shop the local signal — what readers within a few miles are shelving and finishing, refreshed weekly. There is no shop-facing product or location data today.

The evidence

Community-driven recommendation converts at retail scale: TikTok reported that more than 50 million #BookTok-recommended books were sold across Europe in 2025, worth roughly €800 million in its key book markets — demand that arrives in shops already primed rather than needing to be sold from cold.

The challenge

Compulsory reading logs measure compliance, not reading. Students learn to report what the log rewards, and teachers lose sight of what anyone is actually reading.

What PaperBoxd does

Self-selected books, a private list shared with the class, and a leaderboard that ranks on streak or diary entries instead of pages — so a poetry reader isn't losing to someone speed-running airport thrillers. A Friends tab narrows the board to people you follow, though scope and metric can't yet be combined. These are the existing lists and leaderboard used deliberately; there is no separate classroom product.

The evidence

The research consistently favours self-selection: independent, self-chosen reading is associated with higher engagement, better literacy outcomes and stronger retention, and large-scale PISA analysis across 164,233 secondary students in 24 countries found reading self-perception and metacognitive knowledge robustly correlated with digital reading performance.

The challenge

Most trackers treat an audiobook as a second-class edition or refuse to count it, which quietly tells a third of readers that their reading doesn't count.

What PaperBoxd will do

Page progress reads as a percentage today, so editions of one title stay comparable. The rest is planned, not shipped: a Print/Digital/Audio format exists in the data model but there is no way to set it on the web yet, and logging minutes like pages is still to come.

The evidence

Audio is no longer a niche format. The Audio Publishers Association put US audiobook sales at $2.43 billion in 2025, up 9% year on year, with 58% of American adults — an estimated 157 million people — having listened to an audiobook and publishers reporting over 750,000 active titles, a 43% jump on 2024.

The challenge

A decade of ratings, dates and review text is the single biggest reason people stay on a tracker they've stopped enjoying. Switching means either abandoning that history or trusting an importer to carry it across intact.

What PaperBoxd does

A free importer on every account: shelves and star ratings, matched on ISBN first and title + author as a fallback, with unmatched rows skipped and reported back as a count plus a sample. Dates read, review text and a review step for uncertain matches are all still to come.

The evidence

Goodreads' CSV export is the de facto migration path off the platform, and the ecosystem has grown around it — The StoryGraph's importer is the best-known destination. Keeping migration free on every account, with no tier attached to it, is the design decision here.

The challenge

The genre dropdown is a filing system, not a way anyone describes what they want to read next.

What PaperBoxd does

One open field instead of the dropdown, answered with books rather than a filtered category list.

The evidence

Mood-first discovery is the axis competitors are converging on — The StoryGraph's whole recommendation pitch is mood and pace over genre and star rating — and it maps onto how social discovery already works, where readers describe a feeling and the community answers with titles.

The challenge

Star ratings compress everything interesting out of a reading experience. Two five-star books can have nothing in common and one-star reviews are frequently the most useful signal in the dataset.

What PaperBoxd will do

Planned for the embeddings-based recommendation work, not shipped: model the free-text reason alongside the rating, cluster on mood and pace rather than category, and let abandonment be a signal rather than missing data.

The evidence

The industry is moving the same way: mood- and pace-based tagging is StoryGraph's core differentiator, and the scale of social recommendation — 50 million BookTok-driven sales across Europe in 2025 alone — shows readers already navigate by feeling when a platform lets them.