Uploaded August 2026 | Updated September 2026, 1 week ago
Step-by-step: how I set up a custom MCP to pull all my AEO + marketing stats with one prompt
If you run marketing for more than a handful of clients, you know the tax: every report means logging into Google Ads, GA4, Search Console, GBP, CallRail, Meta, your CMS… and stitching it together by hand. It eats a day a week.
I fixed most of that by wiring my data sources into a single MCP server per brand, then connecting that to an AI assistant. "MCP" just means a standard way for an AI to read/act on a tool. The trick is bundling many connectors under one brand so the AI can pull everything in one shot.
Here's the exact setup:
Create a brand. In your MCP platform, add a brand (I do one per client). This is the container everything attaches to.
Open its connections. New brands start with zero connections — click into the brand to start adding them.
Add the connectors you actually use. For AEO + local work the high-value ones are Google Ads, Google Analytics 4, Google Tag Manager, Google Business Profile, Google Search Console, and CallRail. Skip the rest until you need them.
Set permissions. Read-only is fine for pure reporting. Use Read + Write only where you want the AI to make changes (e.g. Ads). Be deliberate here.
Choose the right accounts. When you connect each platform, pick the specific account for that brand — watch out for near-duplicates (e.g. the main account vs. the LSA account). Wrong account = garbage data.
Validate. Go to the connections list and confirm each one reads "Connected" and the action count looks right. Fix anything flagged before you trust a report.
Drop the IDs into your project instructions. This is the step most people skip and it's the biggest unlock. In your AI project, paste the platform + account IDs (CMS site ID, Ads account ID, CallRail company/workspace ID, brand ID, Slack channel, etc.). Now the AI never has to guess which account it's touching.
Run one multi-task prompt. Once it's wired, a single prompt does the whole report: "Pull website stats, ad stats, recent work from Slack, and draft the client email." What took a day now takes a couple minutes, and you review a draft instead of building from scratch.
The AEO angle: because the assistant can read GBP, GSC, and call/form data together, you can actually see how AI-surface visibility connects to real leads — not just rankings in isolation.
Full disclosure — I run a marketing agency, so I built this for our own workflow and I'm obviously biased toward it. But the approach is platform-agnostic; the value is in the "one brand, many connectors, IDs in the instructions" pattern, not any specific tool.
Happy to answer setup questions in the comments.
In This Video:
Thanks so much for watching today’s video. Below are a few helpful links and resources — including my website, tools I use (some affiliate), and the businesses I run. My main focus is Develomark & SplashDash, where we partner with local service-based businesses to help them grow.
Subscribe to my YouTube channel:
youtube.com/channel/UCX8gKciOxPDlldyCaCF7fMg?view_as=subscriber
Follow Develomark on Instagram:
instagram.com/develomark
Connect with us on LinkedIn:
linkedin.com/company/develomark
Explore SplashDash AI:
splashdash.com
Follow SplashDash on Instagram:
instagram.com/splashdash.ai
Visit Develomark:
develomark.com
Contact Us:
Develomark
409 Canal St, Plantsville, CT 06479
860-259-3351
ruan@develomark.com
Step-by-step: how I set up a custom MCP to pull all my AEO + marketing stats with one prompt
If you run marketing for more than a handful of clients, you know the tax: every report means logging into Google Ads, GA4, Search Console, GBP, CallRail, Meta, your CMS… and stitching it together by hand. It eats a day a week.
I fixed most of that by wiring my data sources into a single MCP server per brand, then connecting that to an AI assistant. "MCP" just means a standard way for an AI to read/act on a tool. The trick is bundling many connectors under one brand so the AI can pull everything in one shot.
Here's the exact setup:
Create a brand. In your MCP platform, add a brand (I do one per client). This is the container everything attaches to.
Open its connections. New brands start with zero connections — click into the brand to start adding them.
Add the connectors you actually use. For AEO + local work the high-value ones are Google Ads, Google Analytics 4, Google Tag Manager, Google Business Profile, Google Search Console, and CallRail. Skip the rest until you need them.
Set permissions. Read-only is fine for pure reporting. Use Read + Write only where you want the AI to make changes (e.g. Ads). Be deliberate here.
Choose the right accounts. When you connect each platform, pick the specific account for that brand — watch out for near-duplicates (e.g. the main account vs. the LSA account). Wrong account = garbage data.
Validate. Go to the connections list and confirm each one reads "Connected" and the action count looks right. Fix anything flagged before you trust a report.
Drop the IDs into your project instructions. This is the step most people skip and it's the biggest unlock. In your AI project, paste the platform + account IDs (CMS site ID, Ads account ID, CallRail company/workspace ID, brand ID, Slack channel, etc.). Now the AI never has to guess which account it's touching.
Run one multi-task prompt. Once it's wired, a single prompt does the whole report: "Pull website stats, ad stats, recent work from Slack, and draft the client email." What took a day now takes a couple minutes, and you review a draft instead of building from scratch.
The AEO angle: because the assistant can read GBP, GSC, and call/form data together, you can actually see how AI-surface visibility connects to real leads — not just rankings in isolation.
Full disclosure — I run a marketing agency, so I built this for our own workflow and I'm obviously biased toward it. But the approach is platform-agnostic; the value is in the "one brand, many connectors, IDs in the instructions" pattern, not any specific tool.
Happy to answer setup questions in the comments.
In This Video:
Thanks so much for watching today’s video. Below are a few helpful links and resources — including my website, tools I use (some affiliate), and the businesses I run. My main focus is Develomark & SplashDash, where we partner with local service-based businesses to help them grow.
Subscribe to my YouTube channel:
youtube.com/channel/UCX8gKciOxPDlldyCaCF7fMg?view_as=subscriber
Follow Develomark on Instagram:
instagram.com/develomark
Connect with us on LinkedIn:
linkedin.com/company/develomark
Explore SplashDash AI:
splashdash.com
Follow SplashDash on Instagram:
instagram.com/splashdash.ai
Visit Develomark:
develomark.com
Contact Us:
Develomark
409 Canal St, Plantsville, CT 06479
860-259-3351
ruan@develomark.com
