Getting Started with Ooty: AI-Native Marketing Tools in Under 10 Minutes

Ooty brings your marketing data directly into your AI assistant through the Model Context Protocol (MCP). Instead of switching between Google Analytics, Search Console, Google Ads, Meta, YouTube, and Amazon dashboards, you ask Claude a question and get an answer grounded in your real data.

No dashboards to learn. No reports to build. No spreadsheets to maintain. Just ask your AI.

This guide walks you through the full setup -- from zero to asking your first real question -- in under 10 minutes.

Step 1: Sign Up and Choose Your Products

Head to ooty.io and create your account. You'll see the six MCP products -- each one connects Claude to a specific set of marketing platforms.

Pick the products you need:

• Octopus -- SEO and search. Connects to Google Search Console, PageSpeed Insights, Knowledge Graph, and the Indexing API. If you care about organic search performance, this is your starting point.
• Falcon -- Paid advertising. Connects to Google Ads and Meta Ads. Campaign performance, ROAS, cost-per-acquisition, budget analysis.
• Echo -- Social media. Connects to Instagram, LinkedIn, X, and Reddit. Engagement metrics, content performance, cross-platform comparisons.
• Iris -- Video analytics. Connects to YouTube Analytics and YouTube Data API. Watch time, audience retention, subscriber growth, video-level performance.
• Compass -- Web analytics. Connects to Google Analytics 4, Search Console, and Chrome UX Report (CrUX). Traffic sources, conversion paths, user behaviour, site speed.
• Canopy -- E-commerce. Connects to Amazon PA-API, Keepa, and Rainforest. Product research, price tracking, BSR trends, competitive analysis.

You don't need all six. Most users start with one or two products and add more as they see value.

Step 2: Connect Your Accounts

For each product you've chosen, you'll go through a standard OAuth consent screen. This is the same flow you've done before with any "Sign in with Google" or "Connect your Facebook account" integration.

A few things worth knowing:

We request minimum scopes. Compass only asks for Analytics and Search Console read access. Falcon only asks for Ads read access. We never request Gmail, Calendar, or anything outside what the product needs. This isn't just good practice -- it's a GDPR data minimisation requirement.

Your credentials are encrypted server-side. OAuth tokens are encrypted with AES-256-GCM and stored in our database. They never sit on your machine. For the full security architecture, see our architecture transparency post.

You can disconnect at any time. Revoking access removes the encrypted tokens from our servers. You can also revoke from the platform side (Google, Meta, etc.) at any time.

Step 3: Paste One URL Into Claude

Once your accounts are connected, you'll get an MCP endpoint URL and your license key. The setup in Claude looks like this:

Open Claude Desktop, go to Settings, and add your MCP server:

{
  "mcpServers": {
    "ooty-octopus": {
      "url": "https://ooty.io/api/mcp/octopus",
      "headers": {
        "Authorization": "Bearer your-license-key-here"
      }
    }
  }
}

That's it. One URL, one key. No local process to run, no packages to install, no environment variables to configure.

Your license key works across all your Ooty products. Add another product by adding another entry with a different URL path (/api/mcp/falcon, /api/mcp/compass, etc.).

Works on any machine. The same license key works wherever you run Claude. Set it up on your work laptop and your home machine -- same credentials, same data access.

What to Ask First

Once connected, restart Claude and try a real question. Here are good first queries for each product.

If you connected Octopus (SEO):

"Show me my top 10 pages by organic traffic in the last 30 days, and flag any that have dropped more than 20% versus the previous period."

If you connected Falcon (Ads):

"What's my ROAS across all active campaigns this month? Which campaigns are underperforming?"

If you connected Echo (Social):

"What content performed best on Instagram this week, and how does engagement compare to my average?"

If you connected Compass (Analytics):

"What were my top traffic sources last month, and which ones are growing versus declining?"

If you connected Iris (Video):

"Which of my recent YouTube videos have the highest audience retention rate?"

If you connected Canopy (E-Commerce):

"Show me price history and BSR trends for the top 5 products in [your category]."

These aren't template prompts -- they're the kind of questions you'd normally spend 30 minutes answering by switching between dashboards and exporting data. With Ooty, Claude pulls the data directly and gives you the answer.

What Happens Under the Hood

When you ask Claude a question, the flow is:

1. Claude reads your question and decides which tools to call
2. Claude calls the Ooty MCP tool with the appropriate parameters
3. Your request hits our servers (authenticated by your license key)
4. We decrypt your OAuth tokens and call the upstream API (Google, Meta, etc.)
5. The data flows back through to Claude
6. Claude synthesises the results and presents the answer

The entire round-trip takes 400-800ms per tool call. For multi-step questions where Claude calls several tools, the total time is typically a few seconds.

For the full technical details, see our architecture post.

Tips for Getting More Out of It

Ask follow-up questions. The real power of MCP isn't single queries -- it's conversations. Ask "why did organic traffic drop?" and when Claude identifies the cause, ask "what should I do about it?" Claude maintains context across the conversation.

Combine products. If you have Octopus and Compass connected, Claude can cross-reference Search Console data with Analytics data in one conversation. "Show me pages with high impressions but low clicks, and check if any of them also have poor engagement metrics."

Use it for monitoring. Set up a weekly routine: "Give me a summary of anything unusual in my SEO and ads data this week." Claude will pull from all connected sources and flag what needs attention.

Be specific about time ranges. "Last 30 days" is more useful than "recently." Claude will use whatever time range you specify in the API calls.

Troubleshooting

Claude doesn't see the tools: Make sure you've restarted Claude after adding the MCP config. Claude loads MCP servers at startup.

Authentication errors: Double-check your license key. If you've recently changed your subscription, you may need to update the key in your config.

No data returned: Verify that your accounts are connected at ooty.io. If you connected Google Analytics but haven't granted access to the correct property, the API calls will return empty results.

Slow responses: First-call latency is slightly higher (session creation). Subsequent calls should be 400-800ms. If you're consistently seeing slow responses, check your network connection.

What's Next

Once you're set up, a few resources:

• How Ooty Works -- the full architecture and security model
• MCP Security Guide -- the security design behind the scenes
• MCP vs API -- understanding when MCP is the right tool for your needs

Questions? Reach out at hello@ooty.io.