How we run a 200-article month without losing the plot

Two hundred articles in a month sounds like a claim rather than a workflow. It is neither impressive nor reckless on its own — it depends entirely on what kind of pages they are and who reads them before they go live. This is the process we use across our own sites, written out in enough detail that you could copy it.

Volume is a workflow problem, not a writing problem

When a team cannot publish enough, the instinct is to find faster writers. In practice the delay is almost never in the writing. It is in the space around it: a brief that has to be rewritten because it was ambiguous, a draft that sat in an inbox for six days, an SEO check done from memory at 5pm on a Friday, a publishing step that only one person knows how to do.

Measure it once and the picture usually holds. Writing a 1,200-word draft takes two to four hours. Getting that draft from idea to live page takes eleven days. The gap is the workflow, and the gap is where automation earns its place.

The first month we tracked it, our median time from approved keyword to published post was nine days. Actual writing accounted for less than four hours of that.

The brief does most of the work

A vague brief produces a draft that has to be rewritten, whether a person or a model writes it. Ours has six fields and none of them are optional: target keyword, search intent, audience, format, required sections and the one thing the page must convince the reader of. That last field is the one people skip, and it is the one that separates a page with a point from a page that summarises the internet.

Because the fields are fixed, they can be filled in bulk. A research session produces forty keywords; an hour of work turns them into forty briefs; the pipeline turns those into forty drafts overnight. The leverage is in standardising the brief, not in the drafting.

What a good brief field looks like

• Target keyword: one phrase, not three variations.
• Intent: informational, commercial or transactional — this decides the structure.
• Audience: specific enough to change the vocabulary.
• Required sections: the two or three things the page must cover to be useful.
• Angle: the claim the page makes that a competitor's page does not.

Where the review gate sits

We do not review every page. We review every page in the categories where being wrong costs something. Pricing comparisons, anything with a health or legal angle, and anything that names a specific product all go to an editor. Format-driven pages — category descriptions, glossary entries, straightforward how-to content on subjects we have covered fifty times — publish on schedule and are audited weekly in batches.

The distinction is not about trusting the tool. It is about where an error surfaces. A wrong sentence in a glossary entry is embarrassing; a wrong sentence in a page recommending a purchase is a refund and a reputation problem.

Three checks that catch most problems

Over two years, three checks have caught the majority of issues that would have reached readers.

First, the claim check. Any sentence containing a number, a date, a price or a superlative gets flagged for verification. Models are confident about specifics and specifics are what readers quote back at you.

Second, the duplication check. Every draft is compared against everything already published in the same workspace. Publishing two pages that answer the same question splits your own traffic and confuses the site's structure.

Third, the “so what” check. An editor reads only the introduction and the closing section. If those two paragraphs do not say something a reader could act on, the whole draft goes back. It takes ninety seconds and it is the single most useful quality filter we have.

Cadence, not sprints

The temptation with a pipeline is to publish everything at once. We have tried it. A hundred pages in a week produces a spike in indexing activity, a mess in your internal linking, and nothing much afterwards. A steady rate — ours is eight to ten a day across six sites — is easier to review, easier to attribute, and easier to stop when something is not working.

Steadiness also protects the editorial side. An editor can read ten introductions a day indefinitely. Nobody can read a hundred in a week and still be useful by Thursday.

Where to start if you publish four a month

Do not start at two hundred. Start by writing your brief template and using it for a month, by hand. If the briefs are good, the drafts get better immediately regardless of who writes them. Then automate the drafting for one format only — the one you write most often and enjoy least. Keep the review gate on for everything for the first month.

After that month you will know two things: which formats are safe to run unattended, and how much of your bottleneck was actually the writing. In our experience it is about a fifth. The rest was the workflow, and that is the part worth fixing first.

Keep reading