231, 527, 794 Public Work Pages: Why They Still Cannot Show Who Has a Content Moat
Why public work-page counts cannot directly prove a content moat, and a reproducible methodology for Sitemap audits.
Introduction
Snapshot date: 2026-09-22. The numbers in the title come from this round of audits of public Sitemaps and on-site pages. They represent relevant URLs recorded under the rules used at the time, not active works, creators, or users.
231, 527, and 794 seem easy to rank, but these three numbers may each mix work landing pages, episodes, characters, language versions, tags, and template pages. Before the counting criteria are aligned, greater numerical precision only makes the misleading impression more convincing.
Step One: Classify URL Types
Label each page as work, episode, character, collection, tag, template, or other. Twenty episodes of the same work should not count as twenty independent works; five language versions of the same page should not be counted repeatedly as supply either.
Step Two: Verify Whether the Content Can Be Experienced
Check a random sample to see whether pages load, contain actual content, allow users to proceed from the entry point to an ending, have broken media, require login, and work on mobile. Appearing in a Sitemap does not mean users can complete the experience.
Step Three: Distinguish Supply Speed from Content Quality
New URLs can indicate publishing speed, but quality requires looking at completion rates, independence, updates, replay, and user feedback. Only some signals are observable externally, so clearly describe the “visible catalog size” rather than claiming “the platform has the richest content.”
Five Types of Metrics That Come Closer to Demonstrating a Moat
- 30/90-day retention of active creators;
- The proportion of completed works rather than drafts;
- Median work completion and replay rates;
- Diversity of distinct themes, structures, and assets;
- The time and success rate for creators to move from their first work to their second.
Most of these data are not public, so competitive analysis must acknowledge what is unknown. Page counts still have value: they can reflect the foundation for public distribution, SEO discoverability, and the supply pipeline; they simply cannot independently prove the quality of the consumption experience or network effects.
A Reproducible Audit Table
For each platform, retain the Sitemap address, crawl time, total URL count, classification rules, deduplication method, invalid-page percentage, and sampling scope. At the next update, report additions, deletions, and changes in type to avoid comparisons across different dates and counting criteria.
A content moat is not about “having the largest inventory.” It means good creators are willing to keep publishing, users are willing to complete and replay works, and those works can be consistently discovered. Public URLs are one link in that chain, not the entire chain.
First, Make the Three Numbers Comparable
Crawling on the same day does not automatically ensure consistent counting criteria. One site may include every episode in its Sitemap, another may list only series landing pages, and another may load works on the client side, leaving its Sitemap incomplete. Before auditing, write an operational definition of an “independent work page” and retain unclassifiable URLs separately instead of forcing them into categories for the sake of tidiness.
For deduplication, use canonical URLs, work IDs, and title–creator combinations, then sample multilingual versions, remakes, and mirror pages for inspection. Automated similarity checks can only identify candidates; they cannot independently establish that two stories are the same. Finally, report raw URLs, the number filtered out, candidate independent works, and the proportion manually reviewed together, so readers can see how the numbers were produced.
Use Stratified Sampling to Check Page Quality
Sample by newer and older works, theme, creator, and page type, recording loading, media, interaction, endings, mobile usability, and the most recent update. Homepage recommendations tend to favor high-quality works, while random sampling more closely represents the inventory; however, external access still cannot establish actual completion rates or creator retention, so explicitly call this a “sample check of whether content can be experienced.”
Additions and deletions between snapshots matter just as much. Many new pages that quickly become invalid suggest that the publishing pipeline and ongoing maintenance may be out of sync; stable counts alongside continuous work updates may suggest that a platform places greater emphasis on developing series. Conclusions should describe observed changes without inferring undisclosed business reasons.
From Public Foundations to Evidence of a Moat
URL volume is best suited to answering whether a foundation for public distribution and search exists. Discussing a content moat also requires the completeness of independent works, user consumption, creators producing subsequent works, and a closed feedback loop on the platform. Internal teams can link the public catalog to median completion rates, replay, and thirty- or ninety-day creator retention; external analysts without these data should list them as variables awaiting verification.
When updating the article, retain old snapshots and calculation-script versions instead of overwriting historical numbers with new counting criteria. 231, 527, and 794 have research value only when their dates, page types, deduplication rules, and invalid-page rules are all traceable; otherwise, they are merely three seemingly precise marketing numbers.
When publishing the report, include explanations of the audit-table fields and allow readers to trace aggregate numbers back to anonymized samples and classification rules. When raw URLs cannot be disclosed, at least disclose sampling, deduplication, and error bounds to provide a way to verify the conclusions.
