Content Brief With Claude: Build It From the SERP

Search “content brief with Claude” and you find skills, prompt packs, and npm tools that all promise the same thing. Paste a keyword, get back a tidy brief with a title, an outline, a word count, and a list of headings. It looks like the kind of brief an agency would charge you a hundred dollars for. It reads like one too.

Then you hand it to a writer, they publish the piece, and it lands nowhere.

Here is the problem. Most of those briefs are built from what Claude remembers, not from what is actually ranking today. Claude has never seen the ten pages beating you for your keyword. It is guessing at the angle, the depth, and the questions real searchers ask. A brief built on a guess is a confident outline wearing a brief’s clothes.

A real content brief is almost the opposite of a creative document. It is competitive intelligence. It reverse-engineers the ten results already winning the query, finds the angle they share and the gap they all leave open, and tells your writer exactly how to beat them. Claude can do that brilliantly. It just has to see the SERP first.

What goes into a content brief

A content brief is a one-page spec that tells a writer what to write and why it will rank. Not a vibe, not a topic, a spec. The standard elements are well documented across the SEO field, in guides like Backlinko’s content brief breakdown and Clearscope’s brief template. A complete brief carries most of these.

• Primary keyword and search intent. The one term the page targets, and whether the searcher wants to learn, compare, or buy.
• Secondary keywords. The related terms the page should cover to read as thorough.
• The outline. The H2 and H3 structure, drawn from what ranking pages actually cover.
• Target word count. Pegged to the median of the top results, not a round number you picked.
• Competitor URLs and the gap. What the top ten cover, and the one thing they all skip.
• Title and meta. A headline under sixty characters and a description that earns the click.
• Internal links. The pages on your own site this piece should link to and from.
• Reader profile and CTA. Who this is for, and what you want them to do at the end.

Read that list again and notice something. Half of those fields are judgment, which Claude is great at. The other half are competition, which Claude cannot see on its own. The competitive fields are exactly what turns an outline into a brief. Google’s own guidance on helpful content keeps pointing back to the same thing, content that serves the searcher better than what is already there. You cannot write to that bar without knowing what is already there.

Why Claude writes a confident, wrong brief on its own

Hand Claude a keyword and ask for a content brief and it will give you one in seconds. The headings sound right. The word count sounds reasonable. The intent label looks correct. Everything about it reads professional.

It is also a guess.

Claude’s training data is months old and was never specific to your query. It has no idea that the current top three for your keyword all open with a comparison table, or that the result everyone clicks is a calculator, or that the searchers behind this term actually want a quick yes-or-no answer and bounce off the 3,000-word essays. So Claude defaults to the shape of a generic blog post, because that is the average of everything it has read. Average is what loses in search.

This is the take. A content brief written without the SERP optimizes for an imaginary competitor. It tells your writer to produce a page that beats a page that does not exist. Sometimes the guess gets lucky and matches the real intent. Often it does not, and you find out three months later when the piece is sitting at position 40 and you cannot work out why, because on paper the brief was perfect.

The same trap shows up in the on-page SEO audit world, where a blind audit gives a tidy page a green score it has not earned. Briefs have the identical blind spot at the start of the process instead of the end. Fix it the same way. Give Claude eyes.

The two data sources that make a brief real

Two live feeds turn the guess into intelligence. They do different jobs, so you want both.

The SERP, through an SEO MCP server. The Model Context Protocol is the open standard Anthropic introduced for letting Claude call external tools mid-chat. An SEO MCP server hands Claude tools for SERP results, keyword volume, and difficulty. Once it is connected, “show me the top ten for this keyword and summarize each one’s angle” becomes a real lookup instead of a hallucination. Now the brief can be built from the pages actually winning. The full explainer lives in what is an SEO MCP server.

Your own numbers, through Search Console. If you are refreshing an existing page rather than writing a new one, your real data matters more than any estimate. Google Search Console tells Claude what queries the page already shows up for, where it sits, and what is one edit away from page one. That changes the brief completely, because you stop briefing the keyword you assumed and start briefing the keyword Google already half-agrees you deserve.

The SERP tells Claude what the page is up against. Search Console tells Claude where the page already stands. A brief with both behind it is a different object from a brief Claude typed out of memory.

How to generate a content brief with Claude, step by step

Once Claude can see the data, building the brief is a short conversation. Six moves.

Step one. Confirm the keyword is winnable. Before you brief anything, ask Claude for the difficulty of the target keyword and read it honestly. There is no point briefing a perfect page for a term you cannot rank for this year. Our guide on checking keyword difficulty without Ahrefs covers how to read those scores without overtrusting them.

Step two. Pull the SERP. Ask Claude to fetch the top ten results for the keyword, then summarize each one’s angle, rough word count, and format. You are looking for the pattern. Are they listicles, how-tos, comparisons, tools? This is the single most useful step in the whole process.

Step three. Find the shared angle and the common gap. Ask Claude what every top result covers, and the one thing none of them does well. The shared angle becomes the spine of your outline. The gap becomes your reason to exist. A page that covers the consensus and fills the gap is how you beat an established result without more backlinks.

Step four. Pull secondary keywords. Ask for related terms with volume, then group them into subheadings. This is keyword research feeding straight into the outline, so the brief covers the cluster, not just the head term.

Step five. Read Search Console, if this is a refresh. Ask Claude how the existing page is doing and which queries it already ranks for. Often you discover it ranks for something you were not even targeting, which reshapes the entire brief around a keyword you would have missed.

Step six. Assemble the brief. Ask Claude to write the full brief from everything above. Title, meta, intent, the outline with the gap built in, a word count pegged to the competitor median, the internal links, and a one-line reader profile. You leave the chat with a document a writer can execute, not a topic to go research.

The whole loop runs about ten minutes once the data sources are wired up. Steps two through five are lookups. Step six is Claude doing the synthesis you would otherwise grind out across a dozen tabs.

Content brief template (and which parts need data)

Here is the brief as a template, with an honest column for what Claude can fill on its own versus what needs a live feed. Read the right-hand column top to bottom. The rows marked no are the ones that make the brief worth more than an outline.

A brief that only fills the green rows is the outline every prompt pack already gives you. The red rows are the work, and they are the half that needs a feed.

What this workflow still will not do

An honest read of the limits, because the prompt packs never list them.

It writes the brief, not the article. A great brief makes the writing faster and sharper. It does not write a good page by itself. Claude can draft from the brief too, but the brief is the high-leverage artifact, so treat it as the deliverable.

Estimates are not Google’s ledger. Live SEO data is decision-useful and directionally right, not exact. Your Search Console is the exception, because that data is genuinely yours. Treat volume and difficulty as a strong signal, not gospel.

A brief cannot close an authority gap. If the top ten all sit on far stronger domains, the cleanest brief in the world will not get you there on its own. Sometimes the honest answer is that the keyword is not winnable yet, and a good brief should say so rather than send a writer at a wall.

Claude can be confidently wrong about strategy. It is excellent at pulling the data and shaping the brief. It is not infallible about the call. The brief is a strong first draft of a plan. You still make the decisions.

Generating content briefs with ContextBolt SEO

Full disclosure, since you are reading this on the ContextBolt blog. I build one of these.

ContextBolt SEO is a hosted SEO MCP server made for exactly this workflow. You subscribe, you get one MCP URL, and you paste it into Claude Desktop, Claude Code, Cursor, or Codex. From then on Claude can pull the SERP, related keywords, and difficulty inside any chat, with no DataForSEO account to register, no deposit, and no credentials to manage. Ask “build me a content brief for ‘best crm for freelancers’, pull the top ten and tell me what they all miss” and the brief comes back grounded in the real SERP, not a guess. It is $35 a month for 1,000 lookups at launch pricing, which is a quarter of an Ahrefs plan, month to month, cancel any time.

Two things make it fit the brief workflow specifically. It remembers every lookup across sessions, so when you brief a second piece in the same cluster the related research is already there at no extra credit. And it saves each finding to a ./seo-findings/ folder in your project as markdown, so your briefs and the data behind them live where you work instead of in a chat you will lose. The free Google Search Console connection sits right alongside it, so refreshing an old page reads your real numbers in the same conversation. All of that runs automatically and none of it counts against your lookups.

If you would rather not pay for data at all, you still can. Read the SERP by hand, lean on free keyword checkers, and use your Search Console, which is the only paid-grade data that comes free. Our Ahrefs alternatives round-up covers the cheaper picks. The method is the same either way.

A content brief with Claude is not a prompt you paste and trust. It is Claude reading the real competition, reading your own numbers, and handing you a plan a writer can actually execute. Wire up the data, ask the right questions, and the brief stops being a guess.