How We Work

Last updated: January 2026 · Author: Giovanni Picaro

Most editorial sites do not publish their workflow because the workflow is the secret sauce, or because the workflow does not exist. Ours is documented because we want readers to know exactly what stands between an idea and a published article, and because the structure is itself a quality signal.

Every long-form review and feature on Scanhua moves through five stages before it is published. Shorter news pieces follow a compressed version of the same path. There is no automated pipeline, there is no AI in the loop generating drafts, and there is no rotating cast of freelancers we have never met. Four people, working through five stages, every time.

Stage 1 — Reading and research

Before anyone writes anything, the writer assigned to a piece reads the work in question. Completely, when the work is finished. To the most recent meaningful checkpoint when the work is ongoing — the end of an arc, a major narrative pivot, a volume number that the publisher itself has marked as significant.

For a finished series of, say, ten volumes, this stage typically takes two to three weeks. For a long-running shōnen, the writer assigned to the piece must have read the work in full before claiming it; we do not assign series to writers who have not done the reading.

Research material accompanying the reading includes: the original-language source where one of us reads the language; available academic literature on the work or its author; previous serious English-language criticism (which we read in part to know what has already been said and not say it again); and the publisher’s own promotional and editorial material, treated as a primary source for marketing claims rather than as ground truth.

The writer keeps notes throughout this stage. The notes are not yet a draft.

Stage 2 — Outlining and angle

Before drafting begins, the writer produces a one-page outline that includes: what the article is fundamentally arguing (every review has a thesis, even if it is “this is exactly as good as you have heard and here is why”), what evidence supports that thesis, what counterarguments deserve serious consideration, and what the reader should take away if they read only the introduction and the conclusion.

The outline is shared with at least one other team member before drafting begins. This step is the single biggest reason our published articles do not rehash points the writer has not actually thought through. If the outline does not survive a quick review by a colleague, the article is not written.

Stage 3 — Drafting

The first draft is written from the outline. We aim for a clean first draft within a week of outline approval for review-length pieces, longer for major features. We do not rush this stage; we also do not let it stretch indefinitely.

Drafts on Scanhua are written by humans. We discuss this in detail on our AI Usage Policy page; in summary, we do not use language models to generate drafts of editorial content, and we do not “polish” AI drafts into publishable articles. AI is allowed for narrow operational tasks like translating Japanese terms into English equivalents or organizing notes into tables, and even those uses are disclosed when they are non-trivial.

Stage 4 — Editing and fact-checking

Every long-form article is edited by at least one team member other than the writer. The editor’s job is threefold: line editing for clarity and economy, structural editing for argument and pacing, and a fact-check pass on every concrete claim.

Fact-checking on Scanhua means literally going back to the source material and confirming that each claim in the article is supported. If a review says “the third arc of Vinland Saga spans volumes 8 through 14,” the editor confirms this against the published volume index. If a feature claims that “Inio Asano was born in 1980,” the editor confirms this against the author’s verifiable bio. If a manga is described as “the bestselling shōjo of 2019,” the editor finds the source for that ranking and links to it.

This stage is also where conflicts of interest are surfaced if they have not been already. The editor asks the writer: have you been provided this work for free by the publisher? Have you previously written about this author in a paid context? Is there any commercial relationship the reader should know about? Anything yes goes into the article’s disclosure block.

Stage 5 — Publication and post-publication

Publication is the visible part. What matters more is what happens afterward.

Every article on Scanhua is open to corrections. Reader emails identifying errors are taken seriously and acted on quickly — usually within 48 hours of receipt. Corrections are logged at the bottom of the article with the date and a description of what was changed; we do not silently revise. Our full Corrections Policy describes the process in detail, including the difference between a correction (factual error fixed) and an update (new information added in light of subsequent events).

Comments on articles, where enabled, are moderated by the editorial team. Spam and abuse are removed. Substantive disagreement is welcome and frequently leads to article corrections or follow-up pieces.

News pieces and short articles

Not everything on the site is a long-form review. News pieces and short articles follow a compressed version of the same workflow:

• Source verification — we do not reprint press releases. Every news piece is built from at least two independent sources, and the sources are linked.
• Editorial framing — we do not write breathless announcements. We summarize what happened, link the primary source, and add the editorial context that makes the news matter to our readers.
• Single-pass fact-check — faster than the long-form process, but not absent. The editor reviews names, dates, and verifiable facts before publication.

What this all costs

Producing a single major review takes between 40 and 80 hours of work across the writer and the editor. A short news piece takes between two and four. We publish less than larger sites because we cannot move faster without abandoning this process, and we will not abandon this process.

Reader-supported revenue (ad networks, occasional affiliate links disclosed per the Affiliate Disclosure) covers the operational costs of running the site. The editorial labor is done by people who have other work and who treat Scanhua as a labor of love.

The audit trail

For every long-form article, we keep an internal record of: the original outline, the editor assigned, the fact-check pass, and any post-publication corrections. We do not publish this record (it would be enormous and unreadable), but it exists, and we are willing to share specific records with serious inquirers for legitimate reasons — researchers studying editorial practices, journalists investigating media transparency, publishers concerned about a specific review.

This is what editorial accountability looks like in practice. Not slogans, not promises, but documented process.

Related pages: Editorial Standards · Methodology · Corrections Policy · AI Usage Policy · Sources & Citations