Optimizely's Head of Content Is Dismantling the Machine Marketers Built

Most marketing organizations are exhausted and don't know why. Julianne DeVincenzo knows why. Eleven months into her role at Optimizely, she is simultaneously supporting a rebrand, a platform migration, and an agent infrastructure overhaul. The machine marketers spent years building eventually started running them, and dismantling it from the inside is the hardest, most necessary work in the discipline right now.

Julianne DeVincenzo is Head of Global Content Strategy at Optimizely, the digital experience platform used by over 10,000 brands to manage content, run experiments, and personalize at scale. She came to the role from Qlik where she worked in demand generation strategy and product marketing for analytics and AI, developing her conviction that most marketing organizations vastly underuse the people they hire. She believes the editorial layer is the last thing AI should touch, and the first thing most marketing leaders accidentally automate away.

"We spent so many years feeding the machine that we forgot we ever loved the work," DeVincenzo said.

That line landed in Copenhagen, where she had four days to prepare a presentation and decided just to show what she was actually building. It resonates because it is a critique of effort applied in the wrong direction, not of the people applying it.

The machine nobody decided to build

Nobody sat down and decided to hollow out marketing. It happened through accumulation. Calendars filled up, task lists overflowed, and teams started working inside the same workflows without ever talking to each other. "It's enormous effort, relentless output, and a full, creeping disconnect between the work being done and the person on the other end of it," DeVincenzo said. Speed became the measure of quality, volume became the definition of good, and most organizations either missed the moment it tipped or are still running the same plays as though it hasn't.

She calls it the butcher counter model. Process replaced collaboration so gradually that most teams missed it happening, and the ones that noticed often felt powerless to reverse it. "The strategy was serving the machine instead of the machine serving the strategy," she said. "And we built that. As marketers, we built it."

First-party testing as competitive intelligence

Optimizely is an AI platform for marketers, which means DeVincenzo's team is, as she put it, ground zero. When she arrived, the digital team had just made the decision to migrate platforms, and working alongside them as a partner rather than a member of that team through that process proved to be the most honest product research available, a live test they could not have designed in a lab. Using the product every day surfaced what no customer research could. "We said things that were true before they were fully true," she said, "and now we are living a little bit in that gap." The external positioning had gotten ahead of the product experience, the communication had been narrating a vision rather than meeting buyers where they actually were, and no advisory board member was going to tell them that.

The process also sharpened her team's advocacy in unanticipated ways. One content marketer was writing about Optimizely's agent platform without actually using it, defaulting to other tools out of habit. DeVincenzo's response was direct. "You can't be a vegetarian selling filet mignon and never having tasted it," she said. Once the team started feeling the friction firsthand, the feedback loop between product and content became immediate, and the distance between positioning and reality got harder to ignore.

The like-minded lunatics playbook

Changing the operating model meant changing minds, and DeVincenzo is precise about how she approached it. She started by finding what she calls the like-minded lunatics, a term she credits to former Qlik mentor Chris Powell, people already feeling the pain of the current system or willing to go on the journey toward something different.

"You find people who are also feeling that pain or are willing to go on a journey to change with you," she said. "You show them ideas, you ask for their feedback, and then you have this conglomerate." By the time leadership heard what she was proposing, someone had already told someone else, and the groundwork had been partially laid. Sales became her loudest internal advocates almost overnight, teams that rarely heard from marketing and were immediately receptive when she ran a listening tour.

The process took roughly seven months and required disrupting workflows, messaging documents, brand architecture, and demand generation strategy simultaneously. "It's hard in a middle management position," she said. "It's disrupting everybody's way of thinking, a lot of meetings about meetings, and teaching people a new style."

What agents should absorb, and what they shouldn't

Before DeVincenzo touched any automation, her team spent three months building the foundation. They mapped messaging architecture across product marketing, strategic narrative, and brand, then trained tone-of-voice agents against that foundation until they were actually useful.

What that looks like in practice is agents handling content ideation grounded in competitive gap analysis and search intent, GEO auditing to track where content surfaces in AI-generated answers, and first-pass structuring so strategists walk into briefing conversations with something to react to. The humans decide which stories are worth telling. That editorial judgment is what determines whether a piece actually lands with a reader.

The em dash story says it best. Someone runs an agent that removes em dashes from editorial copy companywide — an offense DeVincenzo did not take lightly. "Did somebody from product tell Hemingway not to use an em dash?" she said. "Banning punctuation because AI uses it feels like outlawing horses before outlaws ever rode them." Her counter is a weekly gap report mapped against what AI-generated answers are surfacing, with topics routed to the thought leaders best positioned to speak to them. The measure of success is whether the content earns trust, and that standard lives with the humans, not the agents.

What it looks like when it works

If the model DeVincenzo is building reaches its intended state, her content marketing organization will be structurally smaller and dramatically more capable. Agents handle the research and first passes. The humans do the work they were hired to do. And the experts across the organization get the tools and structure to be heard, rather than waiting on marketing to tell their stores. "I want people writing like they're writing for the Wall Street Journal," she said. "On the street, doing interviews, bringing information in, becoming a category-leading source that someone reads and thinks, I trust this." The editorial layer is non-negotiable in her model because taste cannot be automated. It develops slowly, over years of actually caring about the people who read the work.

The organizations that figure this out first will look unrecognizable to the ones still filing briefs and waiting for output. DeVincenzo is not waiting to see which side wins.

"If you feel something, say something," she said. "As soon as you say something, there are ten other people who say, thank god you said that." Change does not get explained away; it gets moved through. And the question she keeps returning to is whether the industry helps design what comes next, or just absorbs it when it arrives.

The views and opinions expressed are those of Julianne DeVincenzo and do not represent the official policy or position of any organization.