Who controls book sales data?

There is no odometer for the book industry. Music has a relatively clean stream of plays, films report box office nightly, but books never built a single shared meter. Sales happen across dozens of retailers, in print and digital and audio, through subscription programs and direct stores, and each channel treats its own numbers as a competitive asset. What you are left with is a patchwork of partial datasets, a few gatekeepers who sell access to slices of it, and a layer of tools on top that fill the gaps with educated guesses.

Why there is no single number

Start with the structural problem. A book can sell as a hardcover at a chain bookstore, a paperback on Amazon, an ebook on Kindle, an ebook on Kobo, an audiobook on Audible, a library ebook through OverDrive, a direct PDF from the author’s own site, and a Kindle Unlimited borrow that pays by the page. No one entity sees all of those. The retailers that do see large chunks (chiefly Amazon) have no obligation and no incentive to publish them.

So when someone quotes "this book sold X copies," the honest follow-up is always: in what format, through which channel, measured by whom, and over what window? Strip those qualifiers away and the number is close to meaningless. The most-cited statistic of the last few years, that a huge share of books "sell fewer than a dozen copies," came out of trial testimony and was widely misreported. The reality, reconstructed by a BookScan analyst, was that around 15% of the largest publishers’ frontlist titles sold under 12 copies in a year, not half of all books. Even the scary numbers get mangled because the underlying data is so fragmented.

The gatekeepers: who actually controls the data

A handful of organizations control the pieces that exist. Knowing which one holds which slice tells you a lot about why certain numbers are easy to find and others are impossible.

Circana BookScan (print retail)

BookScan is the closest thing the industry has to an official scoreboard, and it only covers print. It records print sales at the point of sale across major chains, online retailers, independents, and mass merchants, and it captures roughly 85% of US print retail. The ownership has changed hands: Nielsen built it, sold the US operation to NPD Group in 2017 (it became NPD BookScan), and NPD became Circana in 2023. Internationally, NielsenIQ BookScan still runs across about 17 territories.

What it does not see matters as much as what it does. BookScan excludes ebooks, audiobooks, library sales, used books, direct-to-consumer sales, and a chunk of Amazon’s own fulfillment. For a print-heavy traditional title it is a solid proxy. For an indie author selling mostly Kindle ebooks and KU borrows, BookScan is close to blind.

Amazon (the black box)

Amazon is the single largest seller of books in the world and the dominant seller of ebooks, and it publishes essentially no sales figures. It does not report ebook units. It does not report Kindle Unlimited reads in any aggregate form. The only public signal it emits is the Best Sellers Rank, a relative ranking updated hourly that tells you a book is selling better or worse than its neighbors, but never how many copies that represents. Every external estimate of Amazon sales is built by reverse-engineering that rank. This is the central reason the data problem is unsolved: the biggest meter in the industry is deliberately switched off to the public.

Bowker (identity and metadata, not sales)

Bowker, owned by ProQuest, is the exclusive ISBN agency for the United States and runs Books In Print, a database of more than 40 million titles with genre, sub-genre, and bibliographic metadata. This is worth understanding precisely because it is easy to confuse with sales data. Bowker controls book identity (what exists, in which category, by whom), not how many copies moved. Genre-classification data starts here, but the velocity, the "what is selling and changing over time" part, does not.

The AAP and Circana PubTrack Digital (publisher-reported)

The Association of American Publishers releases StatShot, an aggregate of sales reported by member publishers across formats. It is real reported data, not an estimate, but it only reflects participating members and is published as industry-level totals, not per-title numbers. For traditional ebooks specifically, Circana PubTrack Digital collects actual units from more than 450 publishers and is estimated to cover about 80% of the traditionally published ebook market. Note the word "traditionally." Self-published and Amazon-exclusive ebooks are outside it almost entirely.

The estimator layer: where the guessing happens

Because the real data is gated or absent, an entire industry of tools grew up to estimate it. These are not scams, and the good ones are genuinely useful, but it is important to be clear-eyed about what they are doing: they are modeling, not measuring.

BSR-to-sales calculators

Publisher Rocket, the Kindlepreneur and TCK calculators, BookBeam, Helium 10, and a dozen others all do the same core trick. They take a book’s Amazon Best Sellers Rank and convert it to an estimated daily sales figure using a curve calibrated against real sales data that a handful of publishers shared. Publisher Rocket, for example, built its model by tracking how BSR moves and comparing it against actual daily sales from partner publishers.

The vendors quote accuracy figures ranging from a claimed 6% error to a more candid 15 to 20% deviation. Treat all of these as marketing numbers. The model is most reliable for high-ranked books in popular categories where there is a lot of calibration data, and least reliable in the long tail, exactly where most books live. It also leans on rules of thumb like "paperbacks sell at roughly 30% of the Kindle rate at the same rank, hardcovers at 15%," which are averages, not facts about your book. And the entire curve can shift the day Amazon tweaks how the rank is computed.

Bookstat / the old Author Earnings reports

The most ambitious attempt to map the ebook market was Author Earnings, run by the analyst known as Data Guy, later commercialized as Bookstat. It scrapes Amazon rankings at scale and extrapolates them into market-wide estimates, and it captures the online and self-published side far better than BookScan ever did. It was influential enough that Circana eventually brought the dataset into its own stack. But it is still extrapolation from public ranking signals, the methodology has always been contested, and access is a paid, industry-facing product, not something a working author casually checks.

Genre and trend tools

This is the part most authors actually want: what genres and sub-genres are rising or falling over time. K-lytics is the best-known specialist here, producing deep reports on Amazon Kindle categories covering market size, sales velocity trends, price distribution, Kindle Unlimited penetration, and the traditional-versus-indie split within a niche. Publisher Rocket and BookBeam offer category-level views too.

Useful as they are, understand the source: every one of these is built on Amazon category rankings and the same BSR estimation underneath. They are excellent for relative, directional questions ("is paranormal romance heating up, is this sub-niche crowded, what price points dominate") and poor for absolute ones ("exactly how many copies does the average book in this category sell"). The trend line is more trustworthy than any single number on it.

What an author can actually access

Here is the practical map, sorted by how real the data is. The pattern is simple: your own first-party numbers are exact, everything about other people’s books is an estimate, and the gap between those two is where most of the confusion lives.

Your own books: exact, and underused

The single most reliable book-sales dataset you will ever touch is your own, and most authors barely look at it. KDP gives you exact units and exact KU pages read. Author Central hands you free BookScan print figures for your own titles. Each store you sell on reports your sales precisely. Royalty statements are the slow but authoritative confirmation. None of this is estimated. The catch is that it only covers your books, so it tells you how you are doing, not what the market is doing.

The wider market: triangulate, never trust one number

For everyone else’s books and for genre trends, you are in estimate territory, and the right habit is triangulation. Cross a paid estimator (Publisher Rocket, K-lytics) against free directional signals: Amazon category bestseller lists and Amazon Charts, Publishers Weekly and The Bookseller for the traditional side, library hold queues in Libby/OverDrive, Goodreads "want to read" counts, and Google Trends for search interest. Each one is weak alone. Agreement across several is a real signal. A single confident-looking number from one tool is not.

What this means for you

Three things to take away. First, when a tool promises you "real book sales data," ask where the number comes from. If the answer is anything other than first-party retailer reporting or a licensed Circana feed, it is an estimate dressed as a fact, and that is fine as long as you treat it as a direction rather than a measurement. Second, the data you most undervalue is your own: it is the only exact picture you have, and it is sitting in your KDP and Author Central dashboards right now. Third, for genre trends, the shape of the curve over time is trustworthy even when the absolute height of it is not, so read the trend, not the digit.

You cannot control the market’s data. It is fragmented by design and gate-kept by the people who profit from holding it. What you can control is the quality and completeness of your own writing data: how much you produced, when, at what pace, and whether that output is trending up or down. That is the one dataset nobody gate-keeps from you, and it is the one that most directly predicts whether the book gets finished at all.

Related reading:

• How much do authors actually earn? The income data, reconciled.
• Indie vs traditional vs hybrid publishing. Where the money and control sit.
• Who writes? Author demographics. How many writers there really are.
• State of author AI 2026. The consent and data fight.