On October 9, 2026, three of the first eight Google results for "content brief template" were creative brief pages for project planning. None of the eight titles mentioned an AI agent, and Google served no AI Overview for the query in that check.
This post gives you a content brief template for the case where an AI agent writes the draft and you approve it. It has seven fields, and a copy-ready version sits near the end.
We use the same structure for every post on this blog. The current workflow saves the brief before the body and requires every evidence item in the brief to reach the article. You will also see the instruction in this article's own brief that keeps it from competing with one of our product pages.
- 1. The term and the queries other pages own
- 2. Search intent and format, read from the titles
- 3. The angle the current results leave open
- 4. Three or more pieces of dated evidence
- 5. An outline with each section's first sentence
- 6. Length and internal links from the profile
- 7. Image slots the human fills
- The copy-ready content brief template
- Which field to fill first on your next post
Each field below says what goes in it and shows the value we used for this post.
1. The term and the queries other pages own
The first field names one primary query for the post and lists the nearby queries it must leave to other pages. Two posts aimed at the same query end up competing for it, which is what our keyword cannibalization check looks for.
For this article the term is "content brief template". DataForSEO reported 210 monthly US searches for it when seodraft measured the term on September 27, 2026.
The brief also carried this instruction from our calendar: "/features/content-brief owns 'content brief', 'seo content brief' and 'content brief generator': link to it and do not target those." So this post goes after the template query and sends the broader term to our content brief feature page.
2. Search intent and format, read from the titles
The intent field records what the searcher wants, and the titles on the results page are the fastest place to read it. Titles that open with "what is" or "how to" belong to someone learning, while "best" and "vs" belong to someone comparing.
Our October 9 read of the first eight results mixed how-to titles with template downloads, such as "How to Write An Optimized Content Brief + Free Template". We filed the intent as informational, with a usable template expected somewhere in the post.
Write the format next to the intent: list, how-to, guide, comparison or a short answer. We picked a list so each field works as a section someone can quote on its own, and seodraft's rules block a post that repeats the template of the last two posts.
3. The angle the current results leave open
The angle field states, in one or two sentences, what your post will say that the current results leave out. You find it by noting what every result covers and then looking for the question none of them answers.
On October 9, the domains behind the first eight results included notion.com, monday.com and keywordinsights.ai. The titles promised templates, steps and examples, and none of them named an AI agent as the writer.
So our angle reads: "A content brief template for posts an AI agent drafts, where the brief carries the only facts the agent may use and the first sentence of every section." The agent reads that line before it writes anything else.
4. Three or more pieces of dated evidence
The evidence field lists every fact the agent may state, each one with its source and the date it was measured. Anything missing from this field is something the agent would have to invent, so the list also works as your fact-check sheet before you publish.
We ask for at least three pieces per post. seodraft's evidence bank enforces it: a draft with fewer than three fails the rules, and so does a draft whose evidence never shows up in the text.
One line from our evidence bank shows the format we use. In a September 17, 2026 test, measuring the query and building its brief cost USD 0.02622 in DataForSEO, read from the account balance before and after the run.
Your own numbers carry the most weight here, such as a response time or a test result. A quote counts only when you fetched the page yourself and can paste the URL next to it.
5. An outline with each section's first sentence
The outline field pairs every H2 with the sentence that opens it, written before any body text exists. Reviewing nine opening sentences is faster than reviewing a full draft, and it shows you whether the post answers the query before the agent writes the long version.
Each opening sentence should answer its heading in one definitive line. seodraft's rules check this on every H2, and a section that opens with a question or a hedge fails.
Here is a pair from this post. The heading "4. Three or more pieces of dated evidence" opens with "The evidence field lists every fact the agent may state, each one with its source and the date it was measured."
6. Length and internal links from the profile
The length and links field copies two numbers from your site profile, so the agent never has to guess them. Our current profile requires at least two internal links and targets 900 to 1,500 words when the topic justifies the length.
List the link targets by URL, and pick them from posts that are already live. seodraft's internal linking gives your agent that list of live posts. In our first week of testing, a real rules run rejected a non-post internal link, fewer internal links than required, and written-out evidence the matcher could not trace. Our founder's agent wrote up that test week.
When the evidence runs out at 700 words, we stop the post at 700 and let the rules flag the length as an advisory.
7. Image slots the human fills
Image slots name each screenshot or chart the post needs, so the agent writes a placeholder and a person captures the real image. Each slot gets an ID, the alt text and a line on what the picture has to show.
For this post we declared one slot: a filled-in brief for the "content brief template" query.
The agent never produces the picture itself, which keeps made-up screenshots out of the post.
The copy-ready content brief template
Copy this block into a Markdown file, a Notion page or your agent's project folder, and fill one copy per post.
CONTENT BRIEF: [working title] 1. TERM Primary query: Queries other pages own (link to those pages): Measured volume, tool and date: 2. INTENT AND FORMAT Intent (informational, commercial, transactional, navigational): Format (list, how-to, guide, comparison, answer): Result titles that told you this: 3. ANGLE What the current results leave out, in one or two sentences: 4. EVIDENCE (3 or more) Claim | source | date measured Claim | source | date measured Claim | source | date measured 5. OUTLINE H2 | first sentence H2 | first sentence 6. LENGTH AND LINKS Target length: Internal links (live URLs only): Product or feature page to link: 7. IMAGE SLOTS slot-1 | alt text | what it shows
Give the agent the filled brief and your site profile in the same message. With both in hand, the agent has the facts, the shape and the first line of every section before it writes the body. The post that grows out of it follows the eight-part blog post template.
Which field to fill first on your next post
Fill the evidence field first, because it tells you whether the post is worth writing at all. If you cannot list three dated facts from your own work or from a page you read, pick another topic or go collect the facts.
Then write the angle and the opening sentences, and hand the brief to your agent after that. Our post on making AI write without sounding generic shows where the brief sits in the rest of the process.
If you want the brief stored next to each post, seodraft keeps it for you: your agent (Claude, Cursor, or ChatGPT on Business, Enterprise or Edu) writes over MCP, the rules decide when the draft is finished, and you approve. It costs $9 a month for one workspace with up to 3 people, and the 7-day trial needs no card. Start your free trial.
