Verdict: You can now build a self-updating SEO dashboard inside Claude without touching a dashboard builder. Connect an SEO data MCP (we used SE Ranking's) and the official Google Analytics 4 MCP to Claude Cowork, describe the report you want in plain English, and ask Claude to output it as a live artifact. The artifact lives in its own tab, refreshes with current data when you reopen or click refresh, and costs about one percent of a Pro plan's weekly Claude usage per dashboard build.
Last verified: 2026-06-17 · Tools: Claude Cowork / Desktop, SE Ranking MCP, Google Analytics 4 MCP · Skill level: Intermediate (no code, but comfort with OAuth and connectors helps) · Time to first dashboard: ~10–15 minutes after setup
What changed: dashboards are now a prompt, not a project
For years, building a marketing dashboard meant choosing a BI tool, connecting data sources, writing SQL or wrangling a visual editor, and maintaining it whenever an API changed. In 2026, that workflow is being replaced by something simpler: describe the dashboard you want, connect the relevant MCPs, and let an AI agent assemble it.
The Model Context Protocol (MCP), an open standard Anthropic released in November 2024, is the connector layer. It lets any compatible AI client talk to external data sources through a single, standardized interface instead of bespoke integrations. SE Ranking's MCP exposes 160+ SEO tools, and Google now ships an official GA4 MCP for read-only analytics access. Claude Cowork and Claude Desktop can connect to both and turn the combined data into an interactive artifact.
This matters because the two data sources answer different questions:
- SEO data MCP: Are we visible? — rankings, keyword gaps, backlink profiles, and AI-search visibility across ChatGPT, Gemini, Perplexity, AI Overviews, and AI Mode.
- GA4 MCP: Are people actually coming? — sessions, traffic sources, conversions, and on-site behavior.
Together they close the gap between visibility and outcome. This guide walks you through the setup, a reusable prompt template, and a few practical guardrails.
Before you start: what you'll need
| Requirement | Why it matters | Verified source |
|---|---|---|
| Claude Pro, Max, Team, or Enterprise | Live artifacts are only on paid Claude Desktop plans; free plans don't expose MCP connectors | Claude Help Center |
| Claude Desktop for macOS or Windows | Live artifacts are not available in the browser at launch | Claude Help Center |
| SE Ranking account with API access | The MCP uses the same API credits; available on current plans and trial accounts with 100,000 free API credits | SE Ranking MCP docs |
| GA4 property access | The GA4 MCP is read-only and requires Google OAuth with permissions to the property | Google Analytics MCP |
Cost reality check: Running a single live-artifact dashboard build with two data sources consumed roughly one percent of the weekly usage budget on Claude Pro in internal testing, according to the SE Ranking team. Actual burn depends on model choice (Opus costs more than Sonnet), prompt length, and how many tool calls the agent makes.
Step 1: Connect the SEO data MCP
SE Ranking's MCP is a remote MCP server — centrally hosted — so you don't install Docker or edit JSON configs. The server URL is https://api.seranking.com/mcp and it supports Streamable HTTP with OAuth 2.1 and dynamic client registration.
In Claude Desktop / Cowork
- Open Customize in the left sidebar.
- Choose Connectors → Add custom connector.
- Name it (e.g., "SE Ranking") and paste
https://api.seranking.com/mcp. - Click Connect and sign in with your SE Ranking account.
- Approve the OAuth scopes.
Once connected, Claude will list the available tools — read-only and write tools. For a dashboard workflow, set the read-only tools to "always allow" so permission prompts don't interrupt every data pull. Leave write tools on approval-required to avoid accidental changes.
Alternative for teams: If you share a master SE Ranking workspace, sub-accounts can't OAuth directly because API access is tied to the master account. Until SE Ranking ships per-user API keys, copy the master API key and connect via manual API-key auth instead.
Step 2: Connect the Google Analytics 4 MCP
Google's official GA4 MCP server is also read-only and available on GitHub. The setup is slightly more involved because Google uses its own OAuth consent screen and property-level permissions.
- Install the GA4 MCP server from the official GitHub repository.
- In Claude Desktop, add it as a connector using the server URL or local command shown in the repo's README.
- Authenticate with the Google account that has access to the GA4 property.
- Grant read-only access to the property.
The hardest part is usually permissioning: if the Google account lacks property-level access, the MCP will connect but return no data. Make sure the account has at least Viewer or Analyst rights on the GA4 property.
Step 3: Install a library of SEO "skills" (optional but recommended)
A raw MCP gives you dozens of tools, but knowing which ones to chain is its own skill. That's why SE Ranking publishes SEO skills — reusable, open-source prompt recipes that live as Claude personal plugins. They handle common workflows like content briefs, keyword clusters, backlink gaps, and AI-search visibility audits.
To add them in Claude Desktop:
- Go to Customize → Personal plugins → Create plugin → Add marketplace.
- Enter the SE Ranking skills repository URL (
github.com/seranking/seo-skills). - Sync and install the plugin.
Once installed, you can trigger a skill by typing / followed by its name (for example, /seo-content-brief). Skills work best on Opus 4.7 because they chain many tool calls in parallel.
Step 4: Build the dashboard with a single prompt
This is where the workflow flips from "configure a dashboard" to "describe a dashboard." The prompt below combines both MCPs and asks for a finished artifact. You can adapt it to any domain.
Build me a live SEO dashboard for [domain]. Use both the SE Ranking MCP and the Google Analytics 4 MCP.
Pull from SE Ranking:
- My top 5 organic competitors
- My AI search visibility across ChatGPT, Gemini, Perplexity, and AI Overviews
- Share of voice vs. those competitors
- Top 10 keywords where I rank in positions 4–10 (quick-win opportunities)
Pull from GA4 (last 90 days vs. previous 90 days):
- Sessions from AI referral sources (chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com)
- Top landing pages by sessions
- Overall sessions, users, and conversions
Combine both datasets and tell me:
1. Which pages are most likely to benefit from AI-search optimization.
2. Where SE Ranking visibility and GA4 traffic agree — and where they don't.
3. The top 3 actions I should take this week.
Output everything as a live artifact with KPI cards, charts, and a short commentary section. Keep it scannable.
Claude will call the MCP tools, fetch the data, synthesize it, and render the artifact in the side panel.
Step 5: Turn it into a live artifact
A regular artifact is a snapshot. A live artifact re-queries your connected apps when you reopen it or hit refresh.
To create one:
- Ask Claude explicitly: "Create this as a live artifact" or "Output as a live artifact."
- After Claude builds it, look for it in the Live artifacts tab in your Cowork sidebar.
- Reopen it anytime; it will pull fresh data automatically.
- Use the refresh button to force an update.
Important limits: Live artifacts live locally on your computer and do not sync between devices. They also can't be shared with other users yet — sharing is on Anthropic's roadmap. And while live artifacts use your pre-approved connectors without asking again, they can only use connectors you already approved during creation.
What this means for you
For solo marketers and small teams: You no longer need a dedicated BI tool or a dashboard specialist to see how SEO visibility maps to traffic. A single prompt on a Friday afternoon can give you a dashboard you reopen every Monday morning.
For agencies: The same setup scales to multiple clients, but only after you discover the right report shape for each client. We recommend building the first dashboard interactively in Claude Cowork, then moving stable, recurring reports to a scheduler like n8n or Make so they run without manual prompting.
For SEO analysts: This is a complement to, not a replacement for, your existing rank trackers and analytics workflows. Use it for exploratory dashboards and ad-hoc client reports; keep your canonical data in SE Ranking and GA4 for audit trails and historical exports.
FAQ
Can I build this in the browser version of Claude?
No. Live artifacts and MCP connectors for these workflows require Claude Desktop or Claude Cowork on macOS or Windows. The browser does not expose personal plugins or live artifacts at launch.
Do I need to know how to code?
No. The whole workflow is prompt-based. The only technical parts are OAuth sign-ins and, for GA4, installing the MCP server from GitHub. If you can connect a Google account to a third-party app, you can do this.
Does it work with a free SE Ranking plan?
API and MCP access require a current SE Ranking plan with API enabled, or a trial account with 100,000 free API credits. Legacy plans may need an upgrade; contact SE Ranking support if the MCP option isn't visible.
What if I run out of SE Ranking API credits?
SE Ranking is introducing a pay-as-you-go credit wallet that does not expire. As of the June 2026 announcement, pricing starts at $50 for 250,000 credits. Until the self-serve wallet launches, you can request an invoice link via SE Ranking support.
Can I use a different SEO tool's MCP?
The architecture is generic: any MCP server that exposes SEO data will work. At the time of writing, SE Ranking has one of the more complete SEO MCP implementations, but the prompt structure stays the same regardless of the data source.
Is my data safe?
MCP connections use OAuth where possible, and the SE Ranking MCP is read-only on account data by default. For GA4, Google's MCP is also read-only. Still, avoid connecting production accounts on shared machines, and keep write tools on approval-required in Claude.
Discussion
0 comments