Content Brief: The Template That Stops the Rewrite

Swap the left column for the right one and most rewrites stop happening.

Most briefs are long and still fail that test. They arrive with a word count, a keyword density target and a heading list lifted from whoever ranks first — three fields a writer can follow exactly and still hand back something you have to rewrite.

This is the briefing document inside our content marketing guide. It sits downstream of content mapping — one accepted row becomes one brief — and upstream of the draft. How to actually write is in our content writing guide; this page stops where the writer starts.

The one test a brief has to pass

Hand the same brief to two competent writers who have never spoken to each other. If you would happily publish either draft, the brief works. If you get two different pieces, the brief was a topic. Run it on the last brief you sent — the usual failure looks like this:

• Both cover the same subject. Neither answers the same question.
• One writes for a beginner, one for a practitioner, because the brief said "our audience".
• Both hit 1,800 words. One padded.
• One includes a claim you would never put your name to, because nothing said not to.

Four fields fix almost all of it: the reader's situation, the single question the piece answers, the angle that makes it different from what already ranks, and what the piece must not say. Get those right and the rest is administration.

What most briefs specify, and what actually decides the draft

Briefs are rarely too short. They are precise about the easy-to-measure things and vague about the parts a writer needs a decision on.

The right-hand column is harder to fill. That is the whole reason the left-hand one keeps getting filled instead.

The brief template we use, field by field

This is our own working template — the one-page brief Digital Hangover fills before any piece is assigned, in-house or to a freelancer. Each field is here because a draft once went wrong without it.

One observation from our own briefing, not a study: the field left blank most often is "what we must not say", and it is the one that causes the most rewriting. Writers guess at the boundary, guess generously, and the draft comes back with a claim nobody wants to defend. No measured figure for that — a pattern we keep meeting, stated as such.

A filled brief, so you can see it working

Filled for an invented company. Bindu Books is a made-up Pune accounting-software firm selling to businesses of 10 to 50 staff. The brand is invented; the shape of the brief is what we send.

READER          Owner of a 14-person Pune trading firm. Files GST himself in
                the app. Filed last month's return nine days late and wants
                the number before he decides whether to panic.

ONE QUESTION    "I filed my GSTR-3B nine days late. What will that cost me?"

ANGLE           Every ranking page explains the late-fee rule. None of them
                works a single return end to end with the interest included,
                so readers still cannot tell what they owe. We do the
                arithmetic on one return, in order, with the total at the top.

MUST NOT SAY    - No specific rupee figure unless it is sourced to the CBIC
                  notification, cited by number and date, in the draft.
                - No "consult a CA and relax" hedging in place of the answer.
                - Do not name competitor apps.
                - Do not imply Bindu Books files or waives anything for you.

OUTLINE         H2 The short answer (fee + interest, one worked total)  <-- GAP
                H2 How the late fee is calculated, per return
                H2 How interest on tax owed is different
                H2 The nine-day example, line by line
                H2 What to do this month so it does not repeat

EVIDENCE HELD   CBIC notification (writer to cite number + date) and the GST
                portal help page (cite with access date). Two annotated
                screenshots of the filing screen from our own account.

FIRST-HAND      The worked nine-day calculation, shown as a table with every
                line labelled. That is the thing no competing page has.

TARGET URL      /blog/gstr-3b-late-fee/   Status: NEW page, not a rewrite

PRIMARY QUERY   gstr 3b late fee
                Volume / KD: from our own tool, with the database and the
                pull date written here. Never an undated number.

INTERNAL LINKS  /blog/gst-filing-calendar/ in the last section
                /features/gst-returns/ once, in the short answer

NEXT STEP       One: start the 14-day trial. No newsletter box, no demo form.

AI RULES        Outline stress-test and source summarising: allowed.
                Draft of the worked calculation, the screenshots and every
                figure: human only. See the AI section below.

OWNER           Named editor signs off every figure.  APPROVER: same editor.

DEADLINE        Draft 3 Oct. Published 8 Oct.

That fits on one page. A four-page brief is usually a research dump with a brief hidden inside it.

The content research that has to happen first

You cannot fill the angle field without doing the research, and the research is reading rather than exporting. Read the top ten results the way a buyer would, in the market you are targeting, and note four things.

1. What format is being rewarded. Comparison table, calculator, step-by-step, forum thread, video. If eight of ten are one format, that answers a question you were about to guess at.
2. Who is allowed to rank here. Government pages, banks, big publishers, specialists. If page one is all official sources, your realistic target is the long-tail phrasing beside it.
3. The table stakes. What every result covers. Leave it out and the piece reads as incomplete; covering it wins you nothing on its own.
4. The gap. What none of them does — usually no worked example, no numbers, written for another country, out of date, or the answer is in paragraph nine.

Record the rest of the result page too — the People Also Ask questions, whether an AI Overview appears, the related searches. Those are the same question in the reader's own words, and they belong in the outline.

Then match format to intent before you write the outline, using our search intent guide. A brief that asks for a 2,000-word guide against a query that wants one number has already failed.

The honest limits of "here is what the top 10 have" tooling

Tools that scrape the ranking pages and hand you a merged outline, a term list and a suggested word count are useful for speed. They are structurally incapable of giving you an angle.

• They produce the average of what already ranks. Follow one exactly and you have commissioned a composite of page one — the one thing that SERP does not need more of.
• Extracted headings prove a heading exists. They cannot tell you why, or that it is there because everyone copied the same 2019 post.
• Term and entity lists are correlation. A word appearing on ranking pages is not a requirement, and a suggested count for it is a density target renamed.
• Suggested word counts are the mean length of the current top ten. Google is direct about this in Creating helpful, reliable, people-first content (last updated 10 December 2025): "Are you writing to a particular word count because you've heard or read that Google has a preferred word count? (No, we don't.)"
• They cannot see what is missing. Absence has no data — and absence is where your angle is.

Use the tool for table stakes. Do the gap yourself, by reading — about twenty minutes per brief, and the twenty minutes that decides whether the page was worth writing.

What the brief must say about AI drafting

Most teams now have some AI in the drafting process, and a brief that says nothing about it reads as permission for all of it. Pin down four things, per piece.

1. Which stages are allowed. Summarising sources you supplied, stress-testing an outline, tightening a paragraph you wrote — fine. Drafting the section that is your whole angle — usually not, because that section is the reason the page exists.
2. What is human-only, named. Every figure and its source, every quote, anything about a client, the first-hand element, screenshots and worked examples. If a model cannot verify it, a model must not assert it.
3. Who owns accuracy at publication. One named person who has checked each claim against the source in the evidence field. Not "the content team".
4. Whether the page discloses how it was made. Google's guidance about AI-generated content (last updated 10 December 2025) frames this as context, not compliance: "Sharing information about how a piece of content was created can help give your readers more context," added in a way that makes sense for your audience. Decide once and keep it consistent.

The risk the brief is really managing is volume without value. Google's spam policies (last updated 28 August 2026) define scaled content abuse as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users," and describe it as "typically focused on creating large amounts of unoriginal content that provides little to no value to users, no matter how it's created."

That last clause is the operative one. The method is not the problem; unoriginal at scale is — which is why the angle and first-hand-element fields carry more weight in an AI-assisted workflow, not less.

The working rule: a human owns the judgement — the angle, the boundary of what must not be said, every fact, and the final read that asks whether this page says anything.

The SEO fields to carry, and the two to drop

A brief should carry SEO instructions. Four of them, and all four are decisions rather than quotas.

Two fields should come off the template permanently.

Keyword density. No search engine publishes a target, nothing checks your draft against one after publication, and a percentage reliably produces sentences a person would not write. Put the query in the title, the H1 and the first liftable answer, then use the reader's language.

LSI keyword lists. The term borrows from latent semantic indexing, an information-retrieval technique that predates modern search engines, and Google has never documented an "LSI keyword" as a ranking input. What survives from the idea is not a word list: name the real entities the piece must mention to be credible — the specific law, product, city, role or form number. That is a research instruction, not a vocabulary quota.

If you would rather hand this over than build it, our content marketing services include the briefing itself, not just the writing. Directionally, content marketing in India runs about ₹25,000 to ₹1,50,000 a month depending on volume and depth — a market range, not a quote.

Where to go from here

Take the last brief you sent and rewrite it against this template. Fifteen minutes, and you will almost certainly find the "must not say" box empty.

• This week. Cut word count and density from the template. Add reader-and-trigger, the one question, the angle, and what not to say.
• Before the next brief. Twenty minutes reading the SERP by hand, then one sentence naming the gap. If you cannot write it, the piece is not ready to brief.
• Once. Write your AI rules and disclosure decision into the template so they are not re-argued per piece.
• Every brief. Apply the two-writer test before you send it.

One expectation to set: better briefs show up as less rewriting within a fortnight, but organic results from the pages they produce typically take about three to six months to become meaningful. Judge the brief on its draft; judge the page on traffic later.