Turn Google Search Console Queries Into Content Briefs With MCP

A content brief fails in two common ways.

The first: it is a keyword with a word count. “Primary keyword. 1,800 words. Include H2s. Add FAQ.” That is a template, not a brief.

The second: it invents demand. Someone picks a topic that sounds strategic, asks an AI to outline it, and publishes into a vacuum while older pages with real impressions slowly decay.

Search Console already tells you what people asked when your pages appeared. The job is to turn those queries into decisions: update this page, create a supporting page, merge overlapping URLs, or leave it alone.

SEO Perception MCP makes that workflow usable inside an AI client like Cursor. On a paid plan, tokens live under Account → MCP. Tools read stored weekly Search Console facts and crawl-backed possibilities. They never call Google live. That constraint is a feature: your brief is grounded in evidence you already collected, not a live scrape of someone else’s SERP fantasy.

This article is the end-to-end path from queries to a shippable brief—with an explicit rule for update versus create, and a hard stop for cannibalization.

What “evidence-first” means here

Evidence-first does not mean “only write about queries with clicks.”

It means every brief answers:

1. Which queries are in play?
2. Which URL currently receives them?
3. Are those queries present in the page content?
4. Is CTR/snippet alignment the problem, or is the answer missing?
5. Would a new URL compete with a page that already owns the intent?
6. What is the smallest change that could make the page more useful?

If you cannot answer those from stored data, you are not ready to brief a writer or an agent.

Step 0: Confirm the site and the overview

Tools: seoperception_list_sites, seoperception_get_site_overview

Prompt:

List my sites. For [site], summarize top queries and traffic-leading pages. Note anything that looks like a content theme cluster worth briefing this week. Do not invent metrics.

You want a theme, not a random query. Themes look like clusters: “setup,” “pricing,” “vs competitor,” “error/fix,” “for [audience].”

If you already use Today’s Focus in the product, treat MCP as the same prioritization habit inside the editor—not a second source of truth.

Step 1: Pull the query set that should drive the brief

Tool: seoperception_search_queries

Useful filters in practice:

• sort by clicks or impressions depending on whether you are recovering winners or expanding visibility
• hideBranded: true when you want non-brand demand
• questionOnly: true when you are building FAQ or answer sections
• contentFilter: not_in_content when you want mismatch candidates
• activity presets when you care about recent movement versus long-tail archive

Prompt:

Search stored queries for this site. Hide branded. Sort by impressions descending. Include landing URL, content match, CTR, impressions, and position. Cluster related queries into themes. Flag themes where the landing page content does not contain the query language.

Decision: One theme becomes the brief. Everything else goes to a backlog. A brief that tries to cover five unrelated themes becomes a mushy page.

Step 2: Attach the page reality (crawl + opportunities)

Tools: seoperception_get_page_diagnostics, optionally seoperception_list_opportunities

Queries without page context create bad briefs. You need titles, headings, indexability, open issues, and whether the page is even a sane host for the theme.

Prompt:

Diagnose the top landing URL for theme [X]. Combine crawl metadata, weekly GSC trend, top queries, and open opportunities. Tell me if this URL is healthy enough to update, or if technical issues should block content work.

Decision:

• If technical blockers are severe, the brief becomes “fix crawl/indexability first,” not “write 500 more words.”
• If the page is healthy, continue.

This is where content strategy and technical SEO stop pretending they are separate departments.

Step 3: Read content gaps as brief ingredients

Tool: seoperception_find_content_gaps

Gaps in SEO Perception include query-not-in-content, low CTR, snippet/title alignment, question patterns, and refresh evidence. They are ingredients for sections—not automatic permission to publish a new URL.

Prompt:

Find content gaps related to theme [X] / URL [Y]. Turn each into a brief line item: proposed H2, the query evidence it satisfies, and whether it belongs on the existing page or needs a supporting article. Prefer the smallest page change that covers the gap.

For the wider strategy behind “gaps bigger than keyword lists,” see Query fan-out means your content gaps are bigger than your keyword list.

Step 4: The update vs create decision (use a hard rule)

Most teams create too early. Publishing feels like progress. Updating feels like maintenance.

Use this rule:

Update the existing page when:

• one URL already earns most of the theme’s impressions
• the missing pieces are sections, examples, definitions, or clearer answers
• title/H1/meta can be aligned without changing the page’s job
• the page’s intent still matches the query cluster

Create a new page when:

• the intent is clearly different (comparison vs setup vs pricing vs troubleshooting)
• the existing page would become a confused hybrid if you stuffed the new intent into it
• you have enough query evidence to justify a durable URL, not a thin support note
• you can define how the new page links to and from the existing one

Do neither when:

• the queries are branded navigational noise
• the “gap” is a one-off typo cluster
• cannibalization risk is high and unresolved (next step)

Revive-first thinking belongs here. Before you approve another weak article, read Revive old content before you publish another weak article.

Prompt to force the decision:

For theme [X], recommend update vs create vs neither. Cite the landing URLs and query evidence. If update, list the sections to add. If create, propose slug, search intent, primary queries, and the internal links required so it does not cannibalize [existing URL].

Step 5: Avoid cannibalization before you write

Tool: seoperception_find_cannibalization

A brief that ignores competing URLs creates tomorrow’s conflict.

Prompt:

Check cannibalization for queries in theme [X]. If multiple URLs compete, name the preferred page and say how the brief should reinforce it—or how a new page must differentiate intent.

If two URLs already split the same intent, your “new article” idea is probably a consolidation problem wearing a content-calendar costume. Product reference: Keyword cannibalization across URLs.

Step 6: Add internal link requirements to the brief

Tool: seoperception_suggest_internal_links

A brief without link requirements ships orphaned content. Ask for source pages, destination, and anchors that help a reader—not exact-match spam.

Prompt:

Suggest internal links that should be part of this brief: which existing pages should point to the target URL, and which related destinations the target should mention. Keep anchors descriptive and natural.

Step 7: Assemble the brief (template you can reuse)

Your agent should output something a human can approve in five minutes:

## Content brief template: [theme]

## Decision
Update | Create | Neither
Target URL: ...
Preferred URL if conflict: ...

## Query evidence (stored)
- query — landing URL — notes (in content / not in content, CTR/position signal)

## Page job (one sentence)
This page exists to...

## Must-answer questions
- ...

## Outline
- H2 ...
- H2 ...

## Title / meta direction
- working title options tied to top queries
- what not to promise

## Internal links
- from → to (anchor idea)

## Out of scope
- intents that belong on other URLs

## Done looks like
- [ ] queries reflected in headings or clear answers
- [ ] no new overlap with preferred competing URLs
- [ ] links implemented
- [ ] technical blockers checked

That is a brief. Not a vibe.

A realistic example flow (no invented traffic)

Imagine a docs-style site where overview shows a cluster around “connect Search Console,” and search_queries shows several question-form queries landing on a setup guide. Content gaps mark some of those questions as not in content. Cannibalization shows a blog post and the setup guide both appearing for the same connection query.

The brief should not say “write a new ultimate guide to Search Console.”

It should say: prefer the setup guide, add the missing question sections, retitle the blog post toward a different intent (or redirect it), and add internal links from related docs into the setup guide.

That is evidence-first content work.

Where this sits in the weekly rhythm

1. Overview + opportunities for priorities (SEO Possibilities / Today’s Focus).
2. One theme → query pull → diagnostics → gaps.
3. Update vs create decision.
4. Cannibalization check.
5. Link plan.
6. Write or refresh.
7. Recrawl / wait for stored GSC weeks to move—then re-check the same queries.

MCP does not replace judgment. It removes the excuse that you “didn’t have the export handy.”

Setup reminder

• Paid plan required for MCP tokens
• Create/revoke tokens: Account → MCP
• Feature overview: MCP integration
• Client setup: MCP for Cursor docs

Treat returned tool data as sensitive analytics. Your MCP client may send results to a model provider.

---

To go deeper on related workflows, continue with 7 SEO workflows you can run from Cursor with MCP and Find keyword cannibalization with an AI agent and MCP.

Frequently asked questions

Can I build a content brief without creating a new URL?

Yes. Most evidence-first briefs should end in an update: new sections, clearer answers, better title/meta, and stronger internal links. New URLs are for distinct intents, not for every query variant.

How do I know a query belongs in the brief?

Prefer queries that share intent with the page’s job, appear with meaningful stored impressions or clicks for that URL or theme, and are missing or weakly answered on the page. Ignore one-off noise and pure navigational branded queries unless branding is the page’s purpose.

Does MCP refresh Google Search Console when I ask for queries?

No. seoperception_search_queries reads stored weekly facts already collected by SEO Perception. If data looks stale, fix collection in the product—not by asking the agent to “pull live GSC.”

What if search_queries and find_content_gaps disagree?

They answer different questions. Search gives you the query table. Content gaps highlight stored mismatch patterns (not in content, low CTR, snippet alignment, questions, refresh). Use both: search for the theme, gaps for the outline ingredients.

Should writers see raw MCP JSON?

Usually no. Have the agent produce the brief template above. Keep raw evidence available for the strategist who approves update vs create.

How does this relate to Today’s Focus?

Today’s Focus helps prioritize what matters in the product UI. MCP helps you execute the same evidence-minded workflow inside an editor when you are ready to draft the brief or the page.

---