L5 · Net-new content, where humans stay mandatory in AI-era SEO

A good writer on your team produces a handful of pages a month worth putting a name on. That number has not moved in a decade. Every content plan you have written was really a plan for spending it.

L5 · Net-new content is where the Quattr Method draws its brightest human line. AI diagnoses the gap, ranks it, and writes the brief. People write the words that carry your brand's name. That split is policy. It is not squeamishness about machines.

It is also the only work primary on two questions at once. R2 · Retrieved, the second question: do you enter the consideration set, the pool a search engine or an answer engine like ChatGPT draws from at all? And R3 · Referenced, the third question: are you the source answers are built from?

A refresh gate stands in front

Net-new does not start until your refresh backlog is under control and the gap analysis has run. The refresh and content quality article carries the arithmetic. A decayed page is already indexed and already linked, so it proves a lift in days. A new page starts at zero.

The gate is unpopular exactly once per engagement. That is the week somebody wants to announce a content velocity number. It stops being unpopular the first time the refreshed cohort's receipts land.

Scaling net-new before the refresh economics prove out is the most expensive way to learn how your content process behaves. The gate exists so proof funds the writing instead of hope. Capacity is the second reason. Writer hours are the scarcest thing you have, and the gate spends them where the evidence points.

Content briefs inherit their evidence

The brief is where AI does the most useful work on this page. Gap analysis names demand your competitors already proved they can win, and the keyword-gap-audit skill ranks it. A skill is a packaged analysis you run by name in your assistant.

Prompt coverage gaps do the same job on the AI side. They name the questions where answers never cite you, and the ai-prompt-coverage-gaps skill shows who gets cited instead and which of their pages do the answering. Being cited means an AI answer naming one of your pages as a source. That is harder than a mention, an answer naming your brand without linking a page.

A brief built that way carries its own justification. It names the demand it serves and the page that wins it today, the intent it must satisfy, the questions it must answer cleanly, and the internal links that will feed it at launch. Your writer starts with the case made.

The human-in-loop matrix, stated plainly

Your content policy has three tiers, and what separates them is what a failure costs. Autonomous: modules and links at scale, with checks watching them. Supervised: refreshes and contextual work, with a person approving. Human-mandatory: net-new editorial, brand-sensitive material, and anything where trust in your brand is the product.

Those failures are not the same size. Telemetry catches a weak module and you replace it the same week. A weak brand statement, or a wrong claim on a page where trust is the product, costs something no rollback recovers. The line sits where the downside stops being cheap.

Two rules ride along. Anything generated stays grounded in your brand documents, so the machine writes from your facts rather than its own priors. And a customer who wants no AI-drafted content gets exactly that, with no argument. Both are contractual in spirit, because they make the program auditable by the people whose names go on the pages.

Write pages an answer can lift

Because the question above this one is Referenced, you write net-new pages for passage retrieval. Self-contained sections. Definitions that survive being quoted alone. The answer in the first sentence rather than the fourth.

None of that is machine-hostile writing dressed up as craft. The structure that gets a passage lifted into an answer is the structure a human scanner rewards, so the discipline costs you nothing on the traditional side. A page enters the consideration set on its demand match, and earns citations on how cleanly its passages answer.

The cold start is real. It is also manageable. Internal linking routes authority to the page at launch, and the coverage-gap check tells you within a few replays whether it started earning citations. A miss is a finding too.

Judgment is the scope here, not the caveat. The method hands your writer better evidence than they have ever had, then gets out of the sentence.