Our Marketing Transformation Is Now a Business School Case Study

My goal was straightforward, but ambitious: grow faster than the market without increasing spend. At the time, the available playbook was thin: give people AI tools like Copilot and ChatGPT, then watch productivity go up.

We tried it. We asked individuals to experiment with tools and report back. Some people made real progress, but it stayed personal. It didn’t scale, and it didn’t change how campaigns moved through our system. Our campaigns still required more than 50 handoffs. Launches still took 20 days. We were using AI, but we were using it to do the same work in the same order, just slightly faster.

A faster person inside a slow system is still a slow system.

That’s when we realized the tools weren’t the problem. The work itself needed to change.

So we went further: building our own agentic campaign management system. Something fully autonomous that could run campaigns end to end. A “technology first” approach.

That failed too. Without the right context and human input, quality fell apart. The system didn’t know what good looked like.

Both failures taught us something important. You can’t skip the hard work of first systematically discovering, mapping and automating individual tasks. Agents, autonomous workflows, digital coworkers—these things are certainly on the horizon. But transformations fail when they try to get there without doing that foundational work first.

We also learned something about context that I don’t hear people talk about enough.

AI fails without context. But you can’t simply upload documentation into AI and expect high-quality outputs. The AI needs to be systematically trained and evaluated by end users—in our case, my marketing and communications experts—not just by engineers or technical staff.

That distinction matters more than most organizations realize.

We started over with a different approach. We had every marketing function document their tasks. Every discrete piece of work across a campaign lifecycle. Content, social, analytics, communications, product marketing. Fifteen cross-functional workshops. More than 700 tasks total, 50 to 100 per team.

We classified each one: AI-led, AI-assisted or requires human judgment.

More than 500 could be AI-led.

That number was clarifying in a way that strategy discussions hadn’t been. It showed us, concretely, where our teams were actually spending their time. How much of it was coordination. How much was approvals. How much was managing the process of getting things out the door rather than doing the work itself.

When you see that clearly, the path forward stops being abstract.

We didn’t have engineers build the AI assistants. We had marketers build them.

We identified AI champions across the team and had them turn those 500-plus tasks into working assistants. The people closest to the work shaped how the tools operated, embedding real expertise directly into the system.

The reason this matters: you can build a functional AI assistant quickly. Getting it from 80 percent to genuinely useful requires sustained, iterative feedback. And to give that feedback well, you have to be someone who knows what good looks like in that domain.

A content strategist knows what a strong campaign brief reads like. A data analyst knows when an insight is actually an insight. An engineer doesn’t.

My chief of staff Alex Kahn put it this way: “It’s not about the tool. It’s about the substrate beneath the tool. If you don’t start at the task level, nothing else works—not workflows, not roles, not agents. Tasks are the atomic unit. Everything else is an abstraction built on top of them.”

That’s the thing the market isn’t talking about enough. The gap between 80 percent and 100 percent is a domain expertise problem. Not a technology problem. And you close it by putting the people who hold that expertise in charge.

Once we had assistants built by the right people, we orchestrated them through Sapient Bodhi, our agentic AI platform, and connected them into redesigned workflows. The assistants became nodes. Bodhi strung them into flows that could move work through an entire campaign lifecycle with human checkpoints only where judgment actually mattered.

The before and after is stark.

The redesigned workflows reduced campaign launch time from 20 days to three to five days and cut estimated human labor from 100 hours to 14 hours. Manual tasks fell from 80 to 40, while team handoffs were eliminated. People contribute asynchronously now, encoding their context into assistants rather than attending meetings to transfer it.

The measurable impact: time to market improved 50 percent, marketing capacity increased 40 percent. Lifecycle programs scaled seven times over. Creative testing increased 20 times, driving 15 to 25 percent faster funnel velocity at the same spend. Conversion improved 10 to 20 percent from more personalized, intent-based messaging.

But the number that matters most to me isn’t any of those.

It’s the work we were able to do that simply wasn’t possible before. Campaigns we couldn’t have launched. Initiatives that used to be out of reach. The shift from efficiency to actual growth is what this was about.

I’ll end on what I think is the most important.

Automation alone gives you efficiency. On its own, it only delivers efficiencies, not growth. To grow, you have to leverage those gains to transform your operating model.

We tracked time saved and reinvested it deliberately. We rewrote job descriptions for each marketer based on the new model. Campaign managers became journey orchestrators. PR shifted from story pitching to story making. Social moved from channel administration to building genuine influence.

We overhauled hiring, training and promotion around three competencies: story architecture, data fluency and journey orchestration. And five skills we look for in everyone: systems thinking, dot-connecting, judgment, taste and curiosity.

These are the skills AI cannot replicate. They become more valuable as AI handles more of the execution. The goal is not to replace people. It is to give people more space for the work that requires judgment, creativity, relationships, strategy, storytelling and leadership.

Because the real value of AI is not just speed. It is helping us redesign work so people can spend more time on the things that only people can do.