We built an MCP for marketing. Here’s what it actually does.

Every marketer has more performance data than they know what to do with. 

The problem is that getting from raw numbers to an actual decision still takes an hour, a spreadsheet, and three platform logins. 

The answers already exist. They’re just buried across ad platforms, analytics tools, reports, and spreadsheets. You might not be short on data, but you sure are short on a reliable way to access and use it. 

We built an MCP to fix that. It gives marketers a way to ask questions about their campaigns and buyer activity and get real answers back, in seconds, without digging through yet another dashboard. 

The real problem with dashboards

Dashboards weren’t built for how marketers actually work. They were built for how ad platforms want you to work. They show you what keeps you spending, not what’s actually working. They’re not telling you the full story. 

No single platform knows your strategy, your pipeline, or what actually turned into revenue. And without that context, any analysis is going to be off. It’ll default to optimizing for conversions and tell you your brand awareness campaigns are failing. It’ll flag thought leadership videos as garbage because they got zero clicks, even if clicks were never the point.

You can’t tell LinkedIn Ads what your strategy is. But you can tell Claude. You tell it which campaigns are focused on brand awareness, which are driving demand, what your target CPL is, when you’d start worrying, what “good” looks like for each campaign type. And because it lives in Claude, your ad data sits next to your CRM data, your strategy docs, whatever else you’ve connected. 

The marketer controls the full picture for the first time. 

More context → better insights → sharper strategy. 

What we built (and what we didn’t)

As we built this, we made some very intentional decisions about what we included and what we didn’t. 

What’s included

‍V1 gives you full access to LinkedIn Ads, including campaign performance, creative performance, bids, and spend. You also get de-anonymized site visitor data from Vector Reveal, so you can see who visited your site and tie them back to the campaigns that actually brought them there. All access is read-only.

What’s not included

‍You can’t edit campaigns or automate budgets from within the MCP. There’s no full cross-channel attribution yet. We’re starting with LinkedIn and Vector site visitor data. More integrations are on the roadmap, but we wanted to get the foundation right first.

Read-only is intentional. There are other MCPs out there that let you modify bids and toggle campaigns. We’re not doing that yet because we think it gets the order wrong. Anyone can turn off a campaign. The hard part is knowing which one to turn off, and that’s harder to do well at scale than people think. 

We’d rather take the reporting and analysis off your plate so you have more time for the strategic calls that actually require your taste and judgment. 

What this looks like in practice

Our team is using this right now to answer questions like: 

“Am I on budget this month?” 

Our Demand Generation Lead, Kelly Arndt, used to answer this manually. Open every ad platform, pull spend by channel, drop it into Excel, and do the math, minimum three times a week. Now Claude pulls pacing by campaign, channel, and total every morning and drops it in Slack before he’s at his desk. It even projected his quarter-end spend at current rate without being asked.

“Which creatives are fatiguing?” 

A year ago you could run the same ad for a quarter. Now performance drops in a week or two, and when you’ve got dozens of campaigns going, catching that manually is basically impossible. Most marketers start thinking the offer’s stale, or the landing page needs work. But more and more, the answer is just that the creative is fatiguing faster than anyone expected. Being able to ask Claude what’s declining and get an answer before you’ve even logged into the platform changes how quickly you catch it.

“Who engaged with our ads but didn’t fill out the form?” 

You ask, and you get back actual people at actual companies, ranked by engagement. That’s what happens when you can query your de-anonymized visitor data alongside your ad performance. Most platforms can tell you something happened. We can show you who.

“Which ICP buyers have visited my site in the last 30 days?”

Kelly pulled ICP visitors from the last 30 days, matched them against our target account list, and checked how many paid impressions each account had been served, and saw which campaigns delivered them. He found a Tier 1 account with two separate visitors, 368 paid impressions, and no sales owner assigned. That's a warm account nobody was working. It took a few questions in Claude.

Why we built this, and why the data matters

An MCP is only as good as the data underneath it. Most marketing data is aggregated at the company level, or inferred and stitched together from third-party sources. Ours is tied to actual people, based on real engagement with your ads and your site, and captured directly. That matters when you’re trying to trust the answers you’re getting.

We’re also approved API partners with the platforms we connect to, and we cache data properly so you don’t get rate-limited or banned. If you’re spending real money on ads, how your MCP accesses your accounts matters. We’ll go deeper on the trust and safety side in a future post.

Right now, most AI tools in marketing help you write ads or launch campaigns faster. That’s fine, but we kept coming back to a different problem: marketers spend so much time figuring out what’s going on that they barely have time to act on it. That’s what we built this for. 

The early adopters will probably be marketers who are already comfortable with AI, but we’re building this to be approachable for anyone who’s curious enough to try it.

Want in? Sign up for early access →