Power(rangers) your MCP: what two demand gen rangers taught us at Booooo-t Camp

Every Monday morning, somewhere, a demand gen marketer is doing the same thing:

Logging into LinkedIn Campaign Manager.
Exporting a CSV.
Doing the same in every other platform.
Pasting it all into Google Sheets.
Running vlookups to join it together.
Making it look presentable.
Then doing it all over again next week.

It’s a ritual that’s costing you time you don’t have.

At our latest Booooo-t Camp, we called in two rangers to show us a better way.

Kelly Arndt (Vector’s Demand Gen Lead) and Richard Meyer (founder of AGNB Growth) suited up and walked us through exactly how they’re using Vector MCP to replace the spreadsheet grind with something that actually scales.

Here’s what they covered.

But first, what even is an MCP?

MCP stands for Model Context Protocol.

It’s a way for an LLM to connect directly to your tools—your ad platforms, your CRM, your site visitor data—so it can go get the information you need and actually do something useful with it.

Richard explained it with a parenting story:

“I asked my five-year-old to pick up his room. He came back 15 minutes later: ‘Dad, I picked up my room.’ I went to check, and sure enough, everything—down to the Cheez-It crumbles on the floor—was on his bed. He physically picked everything up. That’s not what I meant. I meant clean. That gap between what you say and what you mean? That’s what MCP solves.”

In short, an MCP is the shared language between you and your tools. And when that tool is Vector—with contact-level ad performance and site visitor identification in the same place—the gap closes somewhere really useful.

The before

Before Vector MCP, Kelly and Richard were living the same story. 

Kelly would log into every platform, pull the data, export it, reformat it, load it into Google Sheets, and spend time making it look nice enough to share. Repeat every week. Eventually he graduated to uploading giant CSVs into ChatGPT just to get everything in one place.

Richard ran custom reports from every ad platform, joined together with vlookups and SUMIFS. “Not as time consuming, but also not automated.” He eventually built his own MCP at his previous company, which was fully functional, but “super token heavy.” He also tested a third-party one from an attribution tool. That one was alright, but tended to be hallucinatory. 

Then they both got their hands on Vector MCP.

How Kelly uses it

Kelly’s setup starts before he ever runs a query. He builds markdown context files first: his demand gen strategy, budget allocation by quarter, what good looks like for each campaign type. That context is what turns a generic output into a useful one.

From there, he runs on a rhythm:

• Daily pacing checks: Are campaigns under or over-delivering before it compounds into wasted spend?
• Weekly creative performance: What’s resonating in the new batch? What should be paused?
• Monthly system view: Best performers, how campaigns are working together, where to push harder.

Where it gets interesting is the site visitor data layered on top. He queries Vector MCP to pull target accounts from HubSpot, match them to site visitors, and bucket them by tier with contact names, titles, and LinkedIn URLs, then exports it as a CSV for sales. He also pulls CTV impressions served to accounts currently in the pipeline.

One query. No Google Sheet required. And not just reporting, but triage, optimization, and sales handoffs, too. All without leaving Claude.

How Richard uses it

Richard built a Claude project loaded with everything relevant to his demand gen strategy. ICP definitions, campaign goals, and where things stand. He works in Claude Code with an exported markdown file as the foundation. With that context loaded, the MCP goes and gets the data. Think of it as Alpha 5 running diagnostics while you focus on the mission.

He uses it to:

• Monitor LinkedIn campaign performance: Not just top-line numbers, but early signs of creative fatigue before they become expensive
• Match campaign performance back to contact-level data: Seeing which companies are actually coming from his ads, checking ICP fit in real time, and adjusting targeting based on what he’s actually seeing rather than what he assumed going in
• Surface what he’d otherwise miss: The LLM flagged a 66% click-through rate on one influencer campaign that was, in his words, “far and above better than what was in the rest of the account.” He didn’t go looking for this. It surfaced on its own.

His longer-term goal is agents. Automated checks running in the background, flagging creative fatigue and targeting drift, so insights show up without him ever having to pull a report:

“What’s underwhelming about AI for first-timers is it doesn’t understand you yet. You have to train it. Once you do that, and once it has real data to work with, that’s where you get the speed and the precision.”

What makes Vector MCP different

Both Kelly and Richard have used other MCPs. Richard built one himself, used a third-party one, and found both wanting in different ways. Token-heavy, hallucinatory, or just missing the data that actually matters to a demand gen person.

Here’s what they said sets Vector’s apart:

• The data doesn’t hallucinate. Richard’s self-built MCP worked, but was prone to drift. The third-party one he tested was “sort of hallucinatory.” Vector’s approach to querying and loading data is specifically designed to prevent the context-window decay that causes those problems.
• It’s token-efficient. The MCP Richard built at his last company was “super token heavy.” Vector’s is lighter, which matters when you’re running complex, context-loaded queries.
• Site visitor data is built in. This is the one you won’t find in any generic MCP. Other MCPs exist, but Vector’s also gives you visitor ID, meaning you can stitch ad performance and site visitor identity together in the same query and match both back to your CRM.
• Your GTM lens comes pre-applied. Your ICP, target account tiers,and pipeline stages are all inside your Vector account already. So when you ask a question, you’re not getting raw platform data back. You’re getting data that already knows who you care about.

The takeaway

Your buyers are visiting your site, engaging with your ads, and moving through your pipeline. That data is there.

The old workflow required you to export it, clean it, stitch it together in a spreadsheet, make it look pretty, and hope you caught the signal before it cost you.

The new workflow allows you to build your context once, connect your MCP, and let your campaigns tell you what you need to know before you even have to ask.

You don’t have a data problem. You have a workflow problem.

Time to fire up Vector MCP. 

Want to see it in action?

Kelly and Richard walked through real prompts, live outputs, and their exact workflows during Booooo-t Camp.

Watch the full recording here:

Remember, an MCP is only as powerful as the data behind it. Vector gives your LLM something no generic MCP can: contact-level ad performance and website de-anonymization in the same place, queryable alongside your ICP and pipeline data. Now, instead of stitching data from five platforms into a Google Sheet, you ask one question and get the full picture.

Want to see what Vector MCP could surface for your campaigns?

Sign up for access →