About The Books Insider

The Books Insider is an independent audiobook data project. We measure audiobooks and publish what we measure: listening hours, finish-time math, narration speed, genre length statistics and narrator data, computed the same way across a dataset of 8,560 audiobooks in 11 genres. The site answers the questions listeners actually search (how long is this audiobook, when will I finish it, who narrates it, is it long for its genre) with numbers we compute ourselves, never with copied marketing blurbs.

Every one of the 2,187 book pages carries the same instruments: a finish-time calculator (your minutes per day × playback speed → a real finish date), length percentiles against the book's genre, and narrator information when it can be reliably credited. The whole dataset is summarized on the audiobook statistics page and open to reuse with attribution.

Independence

The Books Insider is reader-supported. We participate in the audiobooks.com affiliate program and may earn a commission when you start a trial through our links, at no extra cost to you. That relationship pays for the servers and the data work, and it ends there:

• Commissions never influence our data. Every duration, rating and statistic is computed by the same pipeline for every book, whether or not a book earns anything. No number on this site can be bought, nudged or sponsored.
• Coverage is popularity-led, not payout-led. The books here are the ones readers rate and reach for, selected by public reader-popularity signals. Every book links to the same partner, so there is no incentive to favour one title over another.
• No paid placement, no fake reviews, no invented authorship. We publish computed findings and clearly labeled estimates, nothing else. Ratings are shown with their counts and never adjusted.
• Disclosure is on every page. Affiliate links carry rel="sponsored"in the markup, each page's footer states the relationship, and the full picture is on the affiliate disclosure page.

How we measure listening time

Two kinds of running time appear on this site, and every figure says which one it is.

• Real durations. For 1,773 audiobooks we hold the retailer's own listed narration time, read from the store product page's structured record after the page's author was verified against ours. A store duration becomes the book's running time on this site only when it fits that book: not an abridged, dramatized or partial edition, and within half to double our page-count estimate (1,640 books today). A duration that fails the check usually belongs to a different edition, so the page keeps its labeled estimate instead. Real durations are never labeled as estimates.
• Estimated durations. Where no verified real duration is available, we estimate from the book's page count and label it as an estimate:

listening hours ≈ pages × 275 words per page ÷ 9,300 words heard per hour

275 words per page is a standard typesetting average; 9,300 words per hour corresponds to the ~155 words-per-minute pace common in commercial audiobook narration. Real narration pace varies, so treat estimates as good to roughly ±10%. Every estimated figure on this site is marked “est.” or “estimated”; if it isn't marked, it's a real duration.

How accurate are the estimates?

We check the formula against the real durations we hold, and publish the result as a statistic that refreshes with every dataset build: Across 1,773 audiobooks holding both our page-based estimate and the store's real listed duration, the median estimate is 10.4 hours against 10 measured, a 4.3% gap (3.0% on the averages); book by book, 58% of estimates land within 20% of the real time.

The estimation formula is validated in aggregate; individual estimates stay labeled as estimates. The check itself lives on the statistics page at #estimate-accuracy-median.

The instruments

• The Audiobook Finish-Time Calculator. Pick any audiobook we measure or enter a running time, set minutes per day and playback speed, and get the day you finish. The same instrument sits on every book page.
• The Books Insider Finish-Time Index. Days to finish at common listening paces, for the median audiobook and for every genre: at 30 minutes a day the median audiobook takes 18 days. Published at #listening-time.
• The Books Insider Narration-Speed Index. The median narration pace in words per minute, measured from real store-listed durations paired with page counts: currently 154 wpm. Published at #narration-speed, with the assumption-free minutes-per-100-pages twin.

How the other stats work

• Genre percentiles (“longer than N% of romance audiobooks”) are computed within each genre cohort of our dataset, not against some unstated universe.
• Finish times are arithmetic on the book's hours: minutes per day and playback speed → days to finish and the calendar date.
• Narrator data appears only when a narrator is explicitly credited in our source metadata. When we can't credit one reliably, we show nothing rather than guess: a wrong narrator is worse than none.
• Ratings come from Open Library's reader ratings and are shown with their counts. We never fabricate or adjust ratings.
• Rankings on the best-of pages score every qualifying book by its listener rating weighted by rating count (a Bayesian average); a list exists only when enough books genuinely qualify. No editorial opinion enters a ranking.

Data sources

• Open Library: bibliographic records, page counts, subjects, reader ratings, cover images.
• Apple's public iTunes Search API: audiobook editions, descriptions and narrator credits.
• audiobooks.com product pages, read only after the page's author is verified against ours: real narration durations, narrator credits, series positions and edition markers, from the page's own structured record.

The dataset is refreshed monthly and re-checked weekly. Each page states when its numbers are computed estimates; the current dataset (build October 5, 2026) covers 8,560 titles with an average length of 9.8 hours.

Contact

Questions, corrections, or a data error you've spotted:support@thebooksinsider.com. Corrections to computed numbers are treated as bugs and fixed in the next refresh.