Content brief: write it twice, before and after the research

What 19 ranking brief guides ask for, the three fields most of them skip, and our test of what a locked brief changes in the draft.

Written by Avoid Content Team9 min read

What was checked before publishingAvoid Content Team · Nov 19, 2026
Seven steps: researched, written and checked with AI assistance, then read by an editor.
06Line editEdited line by lineEdited the draft line by line
EDEditorAvoid Content TeamRead the whole article before it went live

What is a content brief?

A content brief is the document a writer works from. Its job is to say who the article is for, what the reader should be able to do after reading it, and what the article will be built from. The useful version comes in two parts. An input brief holds the decisions made before any research. A locked brief, written after the research, fixes the angle, names what the article adds to the pages that already rank, and lists the checks the draft has to pass.

Most templates treat the brief as one document filled in once. We think that is why so many briefed drafts still read like the pages they were meant to beat. Below we count what the ranking brief guides ask for and show why that set of fields produces a familiar draft. Then come the two-phase method and a copyable template. At the end is a test we ran on eight drafts to see what the second phase changes.

This guide is for people who write or approve briefs every week, so it skips the basics of keywords and outlines. If you want the full writing process around the brief, our guide to SEO content writing covers it step by step.

What do content brief templates ask for?

We read the 19 content-brief guides that rank on the first page of Google in the US for “content brief”, “content brief template”, “what is a content brief” and “seo content brief”, and marked which fields each one puts in a brief.

FieldGuides that ask for it
Audience19 of 19
Primary keyword19 of 19
Outline or headings19 of 19
Word count18 of 19
Secondary keywords17 of 19
Competitor or search result analysis16 of 19
Angle or unique point of view10 of 19
Acceptance criteria for the draft6 of 19
First-hand material (an expert, client data, a test)4 of 19
A rule that every number is sourced or checked3 of 19
What the reader can do after reading1 of 19

The top of the table is close to unanimous. Almost every guide asks for a word count, usually taken from the pages that already rank. Google’s own guidance on helpful content asks whether you write to a word count because you heard it has a preferred one, and answers “No, we don’t.”

The bottom of the table is where the draft gets decided, and few guides go there. One guide in 19 asks what the reader should be able to do after reading. Six give the editor criteria to accept or reject the draft. Three ask that every number come with a source or be checked before publishing.

Why do briefed drafts still read like everyone else’s?

Look at where the common fields come from. The keyword, the secondary keywords, the headings, the word count and the competitor list are all read off the search results. A brief built from those fields describes the pages that already rank, and a good writer delivers what the brief describes.

A brief built only from the fields the search results supply describes the pages that already rank.

Google draws the line that matters here in its guide to generative AI features in Search. It calls content “commodity” when it is “based on common knowledge, which could originate from anyone, and typically adds little unique insight for readers”, and gives “7 Tips for First-Time Homebuyers” as the example, against “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line”. The same guide asks writers not to “recycle what others on the internet have already said, or could easily be produced by a generative AI model.”

The ranking guides do say that briefs improve drafts. We counted 249 claims across the 19 guides, and 16 of them say a brief changes an outcome, such as fewer revisions, a better first draft or better rankings. Fourteen of the 16 are asserted with no evidence. One cites another vendor’s page, and one is an author’s own story. None of the 19 guides measures what a brief changes in the text.

Two practitioners have written about the gap well, and this guide builds on both. Nathan Ojaokomo, who compared 22 ranking articles on three B2B SaaS topics and found that 13 “followed the same skeleton”, adds a field for what you have that the top results don’t, and says “If you can’t name anything yet, find something before you approve the brief.” The Gatilab brief template is built around one reader outcome and a short list of acceptance checks.

The two-phase brief

The fix is to stop treating the brief as one document and to write it twice, once before the research and once after it.

The input brief, before the research

The input brief records what the client or the editor decides before anyone reads a source. It names the reader and their situation, the business goal, the market, the material the client can give (data, an expert, a product to test), the brand’s voice and rules, and the query. Its last field is a first guess at the angle, labelled as a guess. Nobody knows yet what the ranking pages leave out.

The locked brief, after the research

The locked brief is written once the research is in, and it is the one the writer gets. Now that you know what the top results say and what they skip, it fixes the angle, names what this article adds beyond them, says what the reader will be able to do after reading, and lists the acceptance checks. Its plan gives each point the source it rests on.

A point the research could not support leaves the plan before anyone writes it. That is the cheapest edit in the whole process. Gatilab puts it this way: “a missing source is cheaper to fix in the brief than in the draft.”

Ojaokomo describes the same habit from the other side, as “a plan with open questions, where the answers are supposed to change the plan”. The two-phase brief makes that change visible. You keep both documents, and the difference between them shows what the research changed.

At Avoid Content, every article gets both briefs, and they stay with the published piece, so the editor can see what was decided before the research and what was fixed after it.

The three fields that change a draft

What the article adds beyond the top results

Name the addition and its kind. Good candidates are your own data, an expert’s account, a test you ran, a comparison nobody has made across the existing sources, or a decision rule the reader can apply. If the field is empty when the research is done, the brief is not ready. Writing it anyway produces the commodity article Google describes.

The reader outcome

Write it as one sentence the text alone can deliver, such as “After reading, the store owner can calculate her abandonment rate and choose her first three fixes.” Gatilab defines this field as “the action or decision they should be able to take afterward” and warns that a guide can help someone choose a workflow but cannot promise more revenue. An outcome the article cannot deliver gives the reviewer nothing to check.

Acceptance checks

Five to seven checks the draft passes or fails, short enough to run in one read. For example, the first 60 words answer the question in the title, every number names its source, and each section closes one question the reader has. Gatilab’s reason for putting these above the outline is the clearest we found: “A draft can contain every proposed H2 and still fail both.”

What the locked brief changed in our test

We wanted to know whether the second phase changes the text or only the paperwork, so we tested it on two topics that agencies are asked to write all the time, an employee onboarding checklist for an HR software blog and how to reduce cart abandonment for an e-commerce blog.

For each topic, both briefs were built from the same research, the five pages ranking first on Google in the US that day. One brief followed the standard fields from the table above and was written to be a good brief of that kind, with its own angle. The other was a locked brief, with a list of what the five pages leave out, a fixed angle, the addition, the reader outcome, acceptance checks, and a plan whose points each quoted the page they rested on. One writer, with one set of instructions to use only the research, wrote two drafts from each brief. That made eight drafts, a small test whose limits we give after the results. Each draft was then split into individual claims, 832 in all, and every claim was labelled without knowing which brief it came from. A claim was marked as already stated by the five pages, as new and supported by them (a comparison across the pages, a count, a rule built from their facts), or as supported by nothing in the research.

Topic and briefNew and supported claimsClaims nothing supportsClaims the top pages already make
Onboarding checklist, standard brief11%14%75%
Onboarding checklist, locked brief28%3%67%
Cart abandonment, standard brief8%6%86%
Cart abandonment, locked brief23%2%74%

The two drafts written from each brief landed within eight points of each other on every measure, and we had set the pass mark before the run. The locked brief raised the share of new supported claims two and a half to three times and cut the share of unsupported ones on both topics.

Reading a pair of drafts in full shows what the numbers mean. The cart abandonment draft from the standard brief was a complete reference with twelve fixes, a recovery email sequence and a measurement section, and most of it could have been assembled from the five pages. From the locked brief came a draft that argued an order of work. It put the seven different “average” abandonment rates the five pages give into one table, showed where they disagree on the main cause (39%, 48% and 56% for extra costs), and turned two of the disagreements into rules the reader can apply. Of the two, it was also the less complete, with no section on mobile checkout. So a locked brief changes what the article is for, and it does not make the article longer or more complete.

Its limits matter as much as the result. We ran eight drafts on two topics, written and labelled by models from one family, with one labeller per draft. Most claims in every draft, from either brief, still came from the top five pages, so the brief moves the share and does not make an article original by itself. One defect showed up in the locked drafts too. Because the research was the top five pages and nothing else, the drafts cited those competitors by name, which no client blog would publish. Real work takes the research behind a locked brief past the ranking pages, to primary sources and the client’s own material.

Content brief template

Copy the two parts into your own document. Fill in the input brief before the research, and the locked brief after it.

Input brief (before the research)

  1. Reader and situation. Who searches for this, what they are trying to do, and what they already know.
  2. Business goal. Traffic, leads, authority or citation in AI answers.
  3. Market and language.
  4. Query and related queries.
  5. Client material. Data, an expert to interview, a product to test, past results.
  6. Voice and rules. The brand’s tone, words it never uses, claims it never makes.
  7. Angle, as a guess.
  8. Article type, as a guess (how-to, explainer, comparison, list, case study).

Locked brief (after the research)

  1. Angle. One sentence, chosen after reading what the ranking pages say and skip.
  2. What this article adds. Its kind and exactly what it is, with the source.
  3. Reader outcome. “After reading, [reader] can [action or decision].”
  4. Answer first. The 40 to 60 words the article opens with.
  5. Plan. Sections, the points under each, and the source each point rests on.
  6. Dropped points. What the research could not support, and why.
  7. Acceptance checks. Five to seven, each passed or failed.
  8. Facts to recheck on writing day. Prices, product names, statistics.
  9. Length. The sum of what the points need, not the competitors’ median.
  10. Out of scope. What the article leaves out on purpose.

Gatilab warns that “a brief becomes harder to use when every article inherits every possible field”. Drop a field when it does not change the work for that article.

An example: the two briefs behind this guide

This guide was written from the same two documents. The input brief named the reader as a head of content who approves briefs every week and still gets drafts that retell the ranking pages. Its angle, as a guess, was that a brief has to change the article. Its plan was an explainer and how-to with a template.

Three things changed after the research. Reading Ojaokomo’s guide showed that the general point was already made well, so the locked brief narrowed the angle to the two phases and to measuring what the second phase changes. Our count of the 19 ranking guides gave the opening evidence. We fixed the type as a how-to, because the reader’s job is to write a brief and check a draft against it. We also dropped a planned point from the locked brief, a figure on how many content marketers use briefs, because the page it came from did not show how it was measured.

How to check a draft against the brief

  1. Read the reader outcome first, then read the draft as that reader. Ask whether you could do what the outcome promises using only the text.
  2. Run the acceptance checks one by one and mark each as passed or failed. A draft that fails one goes back with that check named.
  3. Recheck the volatile facts on writing day and again before publishing. Ojaokomo’s field for this lists “the prices, vendor names and owners, statistics and live examples”, because a brief can wait weeks before anyone writes from it.
  4. Trace every number to its source. If a figure is not in the plan’s sources, ask where it came from before the draft moves on. Our article on fact-checking AI content describes that check in detail.

Frequently asked

How long should a content brief be?

Long enough to hold the decisions that change the draft, and short enough to read before writing. Gatilab advises keeping the acceptance list short enough for a reviewer to use in a single read. A locked brief with a plan and its sources runs longer than an input brief, and that is expected, because the plan carries the research the writer would otherwise redo.

What is the difference between a content brief and a creative brief?

A content brief plans one article. A creative brief plans a campaign, an ad or a design, and covers the message and what each channel gets. Searches for “content brief template” return both, so check which one a template is before you copy it.

Can a tool write a content brief?

A tool can fill in the fields that come from the search results, such as the keywords, the headings of the ranking pages and their median length. Those are the fields almost every guide already asks for. The fields that change the draft, meaning the addition, the reader outcome, the acceptance checks and the sources for each point, depend on research done first, so they belong in the locked brief.

Sources

6 sources

Methodology: every quotation above was checked against the page it is cited from on November 19, 2026, and each quoted sentence appears on that page. The count of ranking guides and the test are our own, run on November 19, 2026. The counts come from saved scripts, and the test’s setup and limits are given in the section above.

  1. 01Google Search Central. “Google’s guide to optimizing for generative AI features on Google Search.” Last updated July 10, 2026. https://developers.google.com/search/docs/fundamentals/ai-optimization-guide.Supports: the definition of commodity content and its two example titles; the advice not to recycle what others have said or what a generative AI model could produce.
  2. 02Google Search Central. “Creating helpful, reliable, people-first content.” Read November 19, 2026. https://developers.google.com/search/docs/fundamentals/creating-helpful-content.Supports: Google has no preferred word count.
  3. 03Gatilab. “Content Brief Template for SEO Teams.” October 10, 2026. https://gatilab.com/content-brief-template/.Supports: the reader outcome field; acceptance checks over the outline; a missing source is cheaper to fix in the brief; not every article needs every field.
Show all 6 sources
  1. 04Ojaokomo, N. “Content Brief Template: The One I Use, With a Real Brief Next to the Article It Became.” November 10, 2026. https://nathanojaokomo.com/blog/content-brief-template/.Supports: 13 of 22 ranking articles on three B2B SaaS topics sharing one skeleton (his count); the field for what you have that the top results don’t; a brief as a plan with open questions; facts to recheck on writing day.
  2. 05Avoid Content. “What 19 ranking content-brief guides put in a brief.” Our count of the guides on Google’s first page (US) for four brief queries, November 19, 2026.Supports: the field table; 249 claims, 16 of them about outcomes, none measured.
  3. 06Avoid Content. “Standard brief against locked brief.” Our test on two topics, eight drafts, 832 claims labelled blind, November 19, 2026.Supports: the results table and the reading of one pair of drafts.

Pipeline · 9 min read

Put your own topic through the pipeline.

One article free on sign-up. No credit card. Agencies start with a pilot.