A good writer with no SEO background will produce work that ranks, given the right brief. A bad brief turns the same writer into someone assembling a page from instructions, and the result reads exactly like that.
The difference is what the brief tells them to optimise for.
What has to be in the brief?
The question the reader is asking, in the reader's words. Everything else follows from it, and briefs that lead with a keyword instead of a question produce articles that circle the phrase without answering anything.
The working set is small:
- The question. Phrased the way someone would type or say it.
- Who is asking, and what they already know. A brief that does not establish this gets you an introduction explaining what SEO is to an audience of SEOs.
- What they should be able to do afterwards. A decision, a diagnosis, a procedure.
- What we specifically know that a general article would not — the war story, the number from our own data, the thing we learned the hard way.
- Which existing pages to link to, as URLs. Do not make the writer guess at your internal linking.
- What not to cover, because it belongs to another page you already have.
That last item prevents more damage than any keyword instruction. Without an explicit boundary, two writers briefed on adjacent topics will produce two articles competing for the same query.
What should you leave out?
Three instructions that are still standard and still counterproductive.
Keyword density targets. Keyword density has not been a meaningful ranking input for many years, and instructing a writer to hit a percentage guarantees sentences bent around a phrase. Google matches meaning; a writer covering the subject properly will use the relevant language without being told to count.
Exact word counts. A number produces padding, and padding is the most reliable way to make a page worse. Give a range with a reason attached — "this needs to cover four scenarios, probably 1,200 words" — so the writer knows what determines the length.
A list of terms to include. These get inserted mechanically wherever they fit. If a term genuinely needs covering, brief the sub-question it answers instead.
Handing over a competitor analysis and asking the writer to cover everything the top results cover. It produces a synthesis of pages that already exist, which is the definition of a page with no reason to outrank them.
What structural guidance actually helps?
The extraction rules, explained as reader guidance rather than algorithm compliance — which is what they genuinely are.
- Every heading is a question or a claim, in plain words.
- The answer comes immediately underneath, in a sentence or two, before any context.
- Comparisons go in tables. Sequences go in numbered lists.
- No throat-clearing. The first paragraph starts with the substance.
Every one of those makes the page easier to skim, and each is also what makes a passage survive being pulled out for a featured snippet or an AI answer. You do not need to explain the second reason to a writer for it to work.
How do you review what comes back?
Against the question, not against the brief's checkboxes. Read the first sentence under each heading and ask whether it answers the heading. That single pass catches most of what is wrong with weak drafts.
Then check for the thing only you could have supplied. If the article could have been written by someone who had never spoken to your customers, the brief did not give the writer enough, and the fix is upstream rather than in the edit.
One thing worth adding to any template: a line naming the page this article must not become. Writers drift toward the broader version of a subject because it is easier to write, and a brief that says "this is not a general guide to search intent" saves an edit round every time.
Take your current brief template and delete every instruction about keywords. If what remains does not describe a reader and a question, that is what to write next.
Getting content back that reads like a checklist?
We will rebuild your brief template so writers produce work that ranks without being told to write for a machine.
Get in Touch →Does this apply to your site?
Reading about it is one thing. Point the scan at your own site and see whether this applies to you, and what it is worth fixing.
Free and unlimited. No account, no card, and you get every finding rather than a teaser.