Team NinjaCat
How VML Uses AI Agents to Drive Creative and Commerce Performance
Published:
January 7, 2025
Updated:
January 8, 2026
Table of Contents
Modern marketing runs on data, but insight still takes time. For global agency VML, that tension is very real. Their teams manage massive volumes of performance, creative, and commerce data across some of the world’s most recognizable brands. The challenge is turning that data into action quickly enough to matter.
To keep pace, VML has been building a growing portfolio of purpose-built AI agents for marketing inside the NinjaCat platform, each designed to tackle a specific challenge.
“Our mission is to drive outstanding performance for our clients,” says Erick McNett, Managing Director of Marketing Effectiveness & Analytics at VML. “AI agents help us automate the mundane, sharpen our thinking, and dramatically accelerate our time to insight.”
Two of those agents—Meta Creative Carol and Commerce Funnel Felicity—offer a clear window into how VML is using AI to move beyond dashboards and unlock faster, deeper insights across both brand and performance marketing.
Why VML Built AI Agents Instead of More Dashboards
VML’s teams work with enormous, fast-moving data sets. Social campaigns with hundreds of creative placements. Commerce platforms with tens of millions of sessions. Manual analysis simply doesn’t scale, and traditional reporting tools only show what happened, not why or what to do next. Effective marketing data integration is essential for agencies managing this scale.
“What drew us to NinjaCat was the ability to work with large, complex data sets and pair them with AI automation,” McNett explains. “Instead of spending time pulling reports, our teams can focus on improving performance.”
Using NinjaCat’s AI agent builder, VML began creating agents tailored to specific marketing problems. And yes, they gave them human names.
“It sounds small, but naming the agents actually helped with adoption,” McNett adds. “They’re not abstract tools. They’re part of how we work.”
Meta Creative Carol: Turning Creative Complexity into Clarity
The context:
VML supports a major consumer brand with high-volume, seasonal social campaigns on Meta. These programs are intensely creative: hundreds of assets, multiple formats, and constant pressure to keep engagement high.
The challenge:
Creative performance was being evaluated at too high a level. Campaign summaries couldn’t answer deeper questions:
- Which placements are actually driving engagement?
- What creative themes are resonating?
- When is fatigue setting in? And where?
Enter Meta Creative Carol.
Built inside NinjaCat, Carol is an AI agent designed to monitor Meta campaign performance daily and surface actionable creative insights automatically.
What Carol Does
- Monitors Meta campaign performance across hundreds of placements
- Evaluates placement-level metrics against benchmarks
- Charts top- and bottom-performing placements
- Identifies early signs of creative fatigue
- Detects performance patterns tied to creative attributes (e.g., food shots vs. product-only images, human presence, recipe formats)
- Develops optimization recommendations and estimates potential impact
- Generates a clean, client-ready report summarizing findings and next steps
“Carol lets us go far deeper than we ever could manually,” says McNett. “We’re not just seeing which assets perform—we’re understanding why they perform.”
The Impact
- Hundreds of placements monitored automatically
- Daily creative insights without manual analysis
- Faster creative refresh cycles
- 30% increase in consumer engagement
Most importantly, Carol doesn’t deliver a wall of AI-generated text. The output is a clear, visual dashboard that summarizes insights and recommendations—designed for quick decision-making.
Commerce Funnel Felicity: Protecting Revenue at Scale
The context:
VML also supports a high-volume B2B commerce brand where even small changes in funnel performance can have massive revenue implications. The site sees tens of millions of sessions, across multiple traffic sources, regions, and customer types.
The challenge:
The team needed continuous visibility into funnel health—without relying on reactive, manual analysis. Traditional anomaly detection tools could flag changes, but they often missed context or surfaced issues too late.
So VML built Commerce Funnel Felicity.
Felicity is an AI agent that monitors the full commerce journey weekly, connecting behavioral data with customer feedback to uncover issues others miss.
What Felicity Does
- Monitors funnel conversion data from Google Analytics
- Evaluates inbound traffic quality
- Identifies anomalies such as bot traffic, errant campaigns, and new traffic sources
- Benchmarks each stage of the commerce journey
- Analyzes performance by:
- Source
- Device type
- Geography
- Customer segment
- Compares behavioral data with Medallia feedback to identify experience disconnects
- Develops optimization recommendations and estimates business impact
- Generates a polished summary report for stakeholders
“One of the biggest advantages of an agent is that it sees things we didn’t anticipate,” McNett explains. “Felicity surfaced emerging issues in specific geographic markets that we hadn’t even considered as a scenario.”
The Impact
- 20 million sessions monitored
- Analysis time reduced to 10% of previous effort
- Faster detection of unknown issues
- 53% increase in site conversion rate
“Felicity is diving deeper than we can on a daily basis,” says McNett. “She alerts us to issues early, so we can fix them before they become revenue problems.”
Inside the Agent Workflows
Neither Carol nor Felicity required data engineering or custom development. Both were built by marketers using NinjaCat’s agent framework:
- Define the KPIs and benchmarks that matter
- Connect and unify relevant data sources
- Apply logic to identify patterns, fatigue, or anomalies
- Generate summarized insights and recommendations
- Deliver outputs designed for action—not analysis paralysis
The result is AI that works with marketers, not around them.
What Other Marketing Teams Can Learn from VML
VML’s experience highlights a broader shift in how marketing organizations use AI:
- AI agents don’t replace expertise—they amplify it
- Purpose-built agents outperform generic tools
- Naming and ownership drive adoption
- Insight velocity matters more than insight volume
“These agents have fundamentally changed how fast we can respond,” McNett says. “We’re doing deeper analysis, in less time, with better outcomes.”
Ready to Go Beyond the Dashboard?
VML’s success with Meta Creative Carol and Commerce Funnel Felicity shows what’s possible when AI agents are designed for real marketing workflows, not just reporting.
If you’re ready to move from dashboards to decisions, NinjaCat can help you build agents that think like your team and work at machine speed. Request a demo and start building your own AI agents today.
.png)
