A typical content brief contains a target keyword, a word count, a list of related terms and three competitor URLs to reference. A writer handed that document will produce a competent page that reads like a blend of those three competitors, because that is literally what the brief asked for.
Then everyone wonders why the page does not get cited. It does not get cited because it contains nothing the three sources it was built from did not already contain. The brief decided that outcome before a word was written.
The brief is where citability is decided
Editing cannot add original material. An editor can tighten prose, fix structure, correct facts and improve a heading. An editor cannot turn a page with nothing new in it into a page worth quoting, because the missing ingredient is information, and information has to be gathered.
That makes the brief the highest-leverage document in a content operation and the one most teams spend fifteen minutes on. We have written about what assistants are actually selecting for in how AI search engines choose what to cite. This post is about the document that has to put it there.
What a brief must answer before a writer starts
Six things. If the brief cannot answer them, the page is not ready to be written and no amount of writing skill will rescue it.
- The exact question. One sentence, phrased the way a person would actually ask it. Not a keyword. If the page serves several related questions, the primary one still has to be singular.
- The answer, in one or two sentences. If you cannot write the answer in the brief, you do not know it yet. This also becomes the opening of the page, which is the passage most likely to be lifted.
- The claim that is ours. What does this page assert that the existing results do not? A position, a limit, a trade-off, a cost nobody mentions. If the honest answer is nothing, kill the brief.
- The evidence. Named sources with links, our own data, or a person who knows. Mark each one as verified or needs checking. A brief that says research this is a brief that outsources the hard part to whoever is cheapest.
- The first-hand element. What do we know from doing the work that nobody can get from reading? A failure we have seen, a number from our own logs, a pattern across clients, a thing that surprised us.
- Who it is for, and what they already know. A page for a developer and a page for a business owner about the same topic are different pages, not the same page with the jargon adjusted.
The interview is the part people skip
The first-hand element rarely exists in anybody’s head at brief-writing time. It exists in the head of someone who does the work, and it has to be extracted.
Thirty minutes with the practitioner, recorded, produces more usable material than a day of desk research. Useful questions tend to sound like this. What do clients get wrong about this, every time? What did you believe five years ago that you no longer believe? What is the part that always takes longer than people expect? When has this approach failed? What would you tell someone not to bother with?
That last question is the most productive one in the set. Every industry has advice everyone repeats and nobody tests, and a page willing to say an accepted practice is a waste of money is immediately different from the three competitors.
Put the raw quotes in the brief. Not a summary of them, the actual words, because the specificity is the value and summarising sands it off.
Specify the structure, because extraction depends on it
Assistants quote passages, not pages. A brief should therefore specify shape, not just content.
Give the writer the headings as questions or claims, in order. Require that the first sentence under each heading answers that heading directly, with reasoning and caveats after. This feels blunt to writers trained to build towards a point, and it is the single most effective structural habit for both readers in a hurry and machines doing extraction. The same principle applied to questions specifically is in writing FAQ sections that work for AEO.
Name the one table or list the page must contain and say what goes in it. Pages that carry a genuinely informative table get quoted more than pages of equally good prose, because the information is already in a form that can be lifted.
Write the anti-brief too
Half a page of what this piece must not do is worth more than another page of what it should.
List the angles already covered elsewhere on the site, with links, so the writer goes around them instead of through them. Nothing dilutes a site faster than four pages making the same argument in slightly different words, and the fix is a brief that names the overlap in advance rather than an editor noticing it at review. Where overlap already exists, deciding whether to update, merge or delete is the next decision.
List the claims the piece may not make. Anything regulated, anything we cannot evidence, any statistic without a named primary source. Be explicit that an unverifiable number must be cut rather than hedged, because hedged fabrication is still fabrication.
List the phrases that are banned. Every organisation has its own, and the list is worth maintaining.
Internal links belong in the brief, not the edit
Decide which existing pages this one should link to, and which existing pages should be updated to link to it, before writing. Links added at the end are added where they fit the sentence rather than where they serve the reader, and the reverse links never get added at all because nobody owns that step.
Two or three deliberate connections, chosen because the reader genuinely needs that page next, do more than a dozen inserted on keyword matches. This is how a topic cluster actually forms, as opposed to how it looks on a diagram. The strategic version of this is in building a topic map.
What is a waste of effort
Listing thirty related terms to include. Writers either ignore them or stuff them, and both outcomes are worse than not having the list.
Specifying a word count as a target. It makes padding the path of least resistance. Specify what must be covered and let length follow. If you need a number, make it a ceiling rather than a floor.
Pasting competitor outlines into the brief. The writer will anchor on them and produce a fourth version of the same page. If you want to reference competitors, state what they get wrong or leave out, and nothing else.
Briefing from a keyword tool alone. It tells you what has been asked before. It cannot tell you what you know that is worth saying.
Who should write these
Not the writer, because the brief exists to give them what they cannot get alone. Not a keyword researcher, because the valuable part is subject knowledge.
In practice the brief is assembled by whoever runs content, built from an interview with a practitioner, with the research verified before handover. It takes longer than fifteen minutes. On any site where content has to earn attention rather than just exist, it is the best-spent hour in the process.
Briefing an update is a different job
Most content operations have one brief template and use it for new pieces and rewrites alike. That is a mistake, because a rewrite starts with evidence a new piece does not have.
A brief for an existing page should open with what the page currently does. What it ranks for, what it does not, where people leave, which questions it fails to answer, and whether anything in it has become factually out of date.
Then it should state what must not change. A page that has earned links and citations has done so for a reason, and the most common failure in a rewrite is removing the thing that made the page work while improving the parts nobody read. Name the URL, the heading, the table or the passage that must survive.
Then the specific additions. Not make it better, but add a section on this, replace this outdated figure, answer this question that readers keep arriving with.
Whether a page deserves an update at all, or should be merged or removed instead, is a prior decision and we set it out in deciding whether to update, merge or delete.
The brief for a page that underperformed
Worth separating out, because the instinct is to rewrite and the instinct is usually wrong.
Before commissioning anything, establish why it underperformed. There are only a few real answers and they lead to different work. The page was never crawled or indexed, which is a technical problem and no amount of writing fixes it. The page is competing with another page of your own, which is a consolidation problem. The page answers a question nobody asks, which means the brief was wrong rather than the writing. Or the page is genuinely weaker than what it competes against, which is the only case where rewriting is the answer.
Make the brief state which of those it is, with evidence. A rewrite commissioned without that diagnosis produces a better version of a page that was never going to work, and the team concludes that content does not work rather than that the brief did not.
Keep the diagnosis in the brief afterwards. Six months later somebody will propose rewriting the same page again, and the record of what was already tried is what stops the loop.
A sensible sequence
- Confirm the question is one a real person asks, and that the site does not already answer it.
- Write the answer yourself, in two sentences. If you cannot, stop and find out.
- Interview the practitioner for thirty minutes and keep the raw quotes.
- Gather and verify every source. Mark anything unverified as cut rather than optional.
- Write the headings as questions or claims, in order, and name the required table.
- Write the anti-brief: overlaps to avoid, claims not permitted, phrases banned.
- Choose the internal links in both directions and assign the reverse links to someone.
- Hand over, and check the draft against the brief rather than against taste.
The checking step has its own discipline, and we set it out in AI content QA. The deeper argument about why a point of view matters at all is in building a point of view worth citing.
If you want a template version of this to adapt for your own team, get in touch and we will send you the one we use.