Automateed platform data
Automateed Book-Project Statistics: September 2026 Snapshot
Our September 5 capture counted 80,768 book-project records across four Automateed workflows. This report shows what those records measure, how they were counted, and where they cannot answer questions about authors or publishing.
Captured September 5, 2026 at 00:31 CEST (2026-09-04T22:31:47.968Z UTC). The underlying database was queried read-only in one consistent snapshot. This is Automateed's own platform data, reviewed with AI assistance; it has not been externally audited.
What the format counts measure
A record means a project exists in one of the platform's creation tables. We did not filter for a finished draft, publication, payment, or a successful sale. The same person or manuscript can have several records. These totals therefore describe workflow activity, not books available to readers.
| Workflow format | Project records | Share of all records |
|---|---|---|
| eBooks | 72,673 | 90.0% |
| Novels | 4,335 | 5.4% |
| Storybooks | 2,308 | 2.9% |
| Coloring books | 1,452 | 1.8% |
| Total | 80,768 | 100% |
eBooks account for 90.0% of these records. This reflects use of the ebook workflow on one platform. It does not show which genre sells best, whether drafts are original, or whether a particular format gives an author a better return.
Country attribution comes from account IPs
Country-level geolocation resolved 75,861 records to 217 country or territory labels. Another 4,907 records (6.1%) had no usable country result. The account IP stored when we ran the query was joined to each project; we did not measure the location of an individual creation session.
A VPN, travel, shared account, or outdated account IP can affect the label. The figures do not identify nationality, author residence, buyer location, or 217 sovereign states. Shares below use only the 75,861 attributed records as their denominator.
| Country label | Total | eBooks | Novels | Storybooks | Coloring books | Share |
|---|---|---|---|---|---|---|
| United States | 13,174 | 11,365 | 1,050 | 424 | 335 | 17.37% |
| India | 7,746 | 7,112 | 299 | 221 | 114 | 10.21% |
| Brazil | 5,613 | 5,350 | 86 | 100 | 77 | 7.40% |
| Nigeria | 3,981 | 3,692 | 216 | 64 | 9 | 5.25% |
| United Kingdom | 3,553 | 3,083 | 260 | 142 | 68 | 4.68% |
| Indonesia | 3,040 | 2,913 | 77 | 42 | 8 | 4.01% |
| Germany | 2,253 | 1,934 | 110 | 88 | 121 | 2.97% |
| France | 2,191 | 1,977 | 147 | 34 | 33 | 2.89% |
| South Africa | 2,132 | 1,903 | 144 | 57 | 28 | 2.81% |
| Pakistan | 1,724 | 1,575 | 51 | 70 | 28 | 2.27% |
The ten largest labels are shown. The remaining labels account for 30,454 attributed records.
Topics are an older, separate title sample
We also counted the existing classification table: 63,552 title records across 41 labels. Their stored classification dates run from January 1 to March 16, 2026 (UTC). We did not classify new titles for this report. The five largest labels remain the following; the download contains counts for all 41.
| Assigned label | Title records | Share of classified titles |
|---|---|---|
| Self-Help & Personal Development | 6,873 | 10.8% |
| Health & Fitness | 6,104 | 9.6% |
| Business & Entrepreneurship | 4,581 | 7.2% |
| Religion & Spirituality | 3,869 | 6.1% |
| Psychology & Mental Health | 3,497 | 5.5% |
The classifier chooses one leading topic from the title alone, without reading the manuscript. It reads a shared project table without restricting the format. Current matches include 59,561 ebook records, 3,602 novels and 209 uploaded books; 180 previously categorized projects are no longer in that table. Calling this an ebook-only sample would be inaccurate.
The model is directed to choose a specific label where possible; ambiguous or multilingual titles can be misclassified. We have not measured classification accuracy with a human-labeled sample. These percentages measure stored title labels, not reader demand, topic quality, sales, or niche profitability.
Methodology and checks
The format and geography queries ran in a read-only, repeatable-read transaction with a consistent database snapshot. We counted ebook rows with an ebook or legacy missing type, combined 113 rows from the novel table with 4,222 novel-type rows from the shared ebook table, and counted storybook and coloring-book rows. We did not deduplicate multiple projects that may describe the same work.
All dates were included, including 5,349 legacy records with no creation timestamp. This is a count of records present at capture, not a historical time series. Deletions, migrations, and missing dates prevent a current date-filtered query from reconstructing what the database contained on a past date.
For geography, we joined the same project populations to their associated accounts and applied the platform's IP-cleaning rules and geoip-lite version 1.4.10. We checked that attributed plus unresolved records equals the format total, and that each displayed country's four format counts sum to its total. The provider's underlying geolocation database build date was not verified.
For topics, we grouped the existing classification records by label, checked their distinct title-ID count and stored date range, then separately reconciled their current project types. The 41 label counts sum to 63,552 distinct stored title IDs. This older, mixed-format population has a different denominator and collection period from the project and geography tables.
The queries, source-code fingerprint, capture timestamp, totals, and limitations are in the downloadable aggregate. The review was performed with Codex using read-only database queries and code inspection. It verifies the reported extraction and arithmetic; it is not an external audit of the platform, a human assessment of every classification, or a representative survey of publishing.
No account identifiers, IP values, email addresses, private titles, or book text are published. The download contains aggregate counts and methods only. We will create a new dated snapshot for any future update instead of silently replacing these figures.
Download and integrity check
The checksum verifies the downloaded file's bytes. It does not independently validate the source data.
810b1754e5fa7d4143941ab52fa8993f2fd7271aaec091ffece532bad8411e13Earlier reported figures
Automateed's July 7 PRWeb announcement reported 77,636 projects. This page previously displayed a July 15 platform-reported total of 77,742. We retained those figures in the qualified historical archive. The original dated database exports have not been recovered, so those totals are not independently reconstructed or a validated baseline for growth.
The previously stated per-format explanation for the 106-record difference has been removed because the available press release does not substantiate it. The new September snapshot stands on its own observed extraction. The older topic percentages are now identified by their actual stored classification period, and geography is correctly described as account-IP attribution.
Report and methods reviewed September 5, 2026. To report an error, see our editorial policy and correction process.