What can you actually ask your ads with Vector MCP?

We built Vector MCP to help marketers make sense of their data. 

That data is at the heart of your ability to understand your buyers, make better decisions, and move faster. The problem is, accessing it requires you to spend hours switching between platforms, digging around dashboards, and stitching together reports. 

And who has the time for that? (In the year 2026? In this economy?)

That’s precious energy you could be using to tackle the parts of the job that actually need your taste and judgement, things like creative and strategy. 

Vector MCP lets you skip the busywork and ask questions about your campaigns and buyer activity in plain language—but the tool is only as useful as the questions you bring to it. 

Below are the prompts we recommend trying to get acquainted with the capabilities of Vector MCP and start getting more out of your ad data and buyer signals. 

Campaign performance

This is where most people start, and where the time savings are most obvious. The questions below replace the weekly ritual of logging into platforms, pulling numbers into spreadsheets, and doing math that should have been done for you already.

1. “How are my campaigns performing this month?”

What you get back: Performance by campaign with key metrics, grouped by objective.

Why it matters: This is the question Kelly Arndt, our Demand Generation Lead, said he asks most. Before Vector MCP, answering it meant going into every ad platform, pulling spend per channel, dropping it into Excel, and doing the math. Now the data comes to him.

But keep in mind, AI platforms will default to treating conversions as the most important metric. If you’re running brand awareness campaigns or thought leadership videos where clicks were never the goal, you need to tell it that upfront. Otherwise it’ll flag those campaigns as underperforming when they’re doing exactly what you designed them to do.

What you’d do next: Identify which campaigns are underperforming relative to their actual objective, not just their conversion rate. Figure out where to shift budget.

2. “Am I on pace with my budget this month?”

What you get back: Spend-to-date versus monthly budget, broken out by campaign and channel, with a variance showing whether you’re over or under.

Why it matters: Kelly used to do this calculation minimum three times a week, sometimes every day. Open every ad platform, pull the spend, drop it into a spreadsheet with formulas, check individual channel budgets and total budget, figure out if he needs to shift money around. It might not be complicated work, but it’s constant and time-consuming. 

Now he has a workflow that pulls pacing every morning and sends it to Slack before he’s at his desk. One thing he didn’t expect: the MCP projected his quarter-end spend at the current rate without being asked. It extrapolated that at his current pace, the total he’d spend by end of quarter was significantly less than his target. That’s not a number he was tracking, but it was enough to flag a real issue.

What you’d do next: If you’re underpacing, figure out which performing campaigns to put more budget behind. If you’re overpacing, pull back before you blow past your number. Kelly described it as playing two games at once, channel-level pacing and total pacing. 

3. “What opportunities do I have?”

What you get back: A read on underperforming campaigns alongside ones producing results, with a view of where budget could move.

Why it matters: This is Kelly’s natural follow-up after he knows what’s working and what isn’t. If a campaign is underpacing and underperforming, that’s budget you can redirect into something with momentum. The pacing data and performance data start working together here.

What you’d do next: Move budget from underperformers into campaigns that are converting. 

Creative performance

Knowing your campaigns are running is one thing. Knowing which ads inside those campaigns are still pulling their weight is another. Creative fatigue is accelerating, and most marketers don’t catch it until performance has already cratered.

1. “Which creatives are fatiguing?”

What you get back: Ads with declining engagement over time.

Why it matters: Creative has a shorter shelf life than it used to. You might get a week or two before engagement starts dropping, and when you’re running dozens of campaigns at once, that’s going to be hard to catch manually. The instinct is usually to blame the offer or the landing page, but sometimes it’s just that the ad has been in front of the same audience too many times. Asking the MCP what’s declining lets you see it early enough to rotate creative before you’ve wasted spend.

What you’d do next: Rotate creative on the fatiguing ads. Pause the worst performers. Reallocate spend to what’s still working.

2. “Pull best performing creative by engagement rate year-to-date.”

What you get back: Creative ranked by engagement, with metrics attached.

Why it matters: Kelly uses prompts like this to look at creative performance over longer time horizons, not just the week-to-week noise. One pull showed him a blended CPL number that was surprisingly low for the overall account. His first reaction was to go to LinkedIn Campaign Manager and validate it, because if that number was real, he wanted to understand why.

What you’d do next: Double down on the creative patterns that are winning. Use what’s working to brief your next round of ads.

ICP and targeting

Vector MCP brings in more than campaign metrics. It surfaces company-level engagement data from LinkedIn and contact-level visitor data from Vector Reveal, both tied back to the campaigns that sourced them. That’s what makes it different from connecting Claude or ChatGPT to your ad platform directly, and these prompts tap into that.

1. “Which companies are seeing my ads?”

What you get back: Company-level engagement data from your LinkedIn campaigns.

Why it matters: Most ad platforms show you aggregate performance—clicks happened, impressions were served. But they don’t tell you which companies are actually engaging. Being able to see which accounts from your target list are paying attention is a useful gut check on whether your targeting is reaching the right people.

What you’d do next: Cross-reference engaged companies against your target account list. If a Tier 1 account is engaging with your ads but nobody from sales is working them, that’s a gap worth closing.

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

What you get back: Named contacts at real companies, ranked by engagement, even if they never filled out a form.

Why it matters: Today, ads are measured on two outcomes: someone clicked and filled out a form, or they didn’t. That leaves the vast majority of your ad spend in a black box. Either it was top of funnel and you weren’t expecting a conversion, or it was bottom of funnel and people just didn’t convert.

Because Vector MCP connects your ad data to de-anonymized visitor data, you can ask who clicked your ad but didn’t convert and get actual names back. 

What you’d do next: Hand warm leads to sales with context. Someone engaged with your ad, visited your site, and didn’t fill out a form. That’s a real person at a real company showing intent worth acting on while it’s fresh.

3. “Show me ICP visitors from the last 30 days.”

What you get back: De-anonymized site visitors matched against your ICP criteria, with associated ad engagement data.

Why it matters: Kelly pulled ICP visitors from the last 30 days, matched them against the target account list, and checked how many paid impressions each account had been served. He could see which campaigns delivered those impressions. He found a Tier 1 account with two separate visitors, 368 paid impressions, and no sales owner assigned. Without pulling that data together, there’s no way anyone would have known to follow up.

What you’d do next: Get that account assigned. Alert sales with the visitor context and the ad engagement history. The combination of Vector’s visitor data and LinkedIn campaign data creates a signal you wouldn’t have found in either system alone.

How to get better answers

The prompts above can surface useful data. But the quality of the answers depends on how much context you give. The more your AI platform knows about your strategy, your goals, and what good looks like for your business, the sharper the output gets.

• Tell it what you’re actually trying to do. Not just the campaign objective in the ad platform, but the real one. Tell it which campaigns are for brand awareness, which are for demand generation, what your CPL target is, and what “good” and “bad” look like for your business. For example, it needs to know that your connected TV campaign is measured only on reach. Without that context, the platform will default to flagging anything without conversions as a problem.
• Keep prompts simple and specific. If you want campaign performance, creative data, and site visitor engagement, don’t cram all of that into one massive prompt. Instead, split it into separate queries. The more you ask in a single prompt, the more room there is for the AI to drift or fill gaps on its own.
• Give it your playbook. Every marketer thinks about their strategy differently. You can edit that file or add your own context: how you define your ICP, what your stage progression looks like, what your CPL targets are. Over time, the platform actually starts to think and talk like you, and the outputs get better with each iteration.
• Give it a head start with a skill file. Vector has an optional skill file you can download and upload to your AI platform. It gives the AI some baseline context about how your campaigns are set up and how to interpret your data, so it’s a little smarter out of the gate. You don’t have to use it, but it means less explaining every time you start a new conversation.
• Build it into your routine. The real value isn’t in what you ask once. It’s in the questions that run on a schedule. You can set up recurring tasks in your AI platform that pull your ad performance data, surface contacts engaging with your ads, and cross-reference that against your CRM. Run them daily, weekly, or monthly depending on what you need.

What to do next

You don’t need to be an expert to get value from this. Start with the basics of what’s working, what’s pacing, and what’s fatiguing. Ask those three questions and see what you get. Then build from there.

Want to learn more? Our CEO Joshua Perk recorded a few short videos on getting the most out of Vector MCP:

• How to connect Vector MCP to your AI platform 
• How to use Vector MCP and write better prompts 
• Using a skill file with Vector MCP 
• Best practices and guardrails for Vector MCP 

If you’re ready to start using Vector MCP, sign up for access here →