I'm the marketer you call for the messy middle: the problem that's new, has no playbook, and still needs an answer this quarter.

14 years in consumer tech. Today I'm leading a global brand and creative team through a merger and building a creative engine that knows where AI belongs and where it doesn't.

Director, Brand & Creative at Coursera, based in San Francisco. I write The Messy Middle on marketing, AI and leading through complexity. Get in touch.

Some problems I've solved

  1. Deciding where AI fits in our creative process, and where it doesn'tI led a team of strategists, writers and designers as we experimented with AI to find the right places for it in our work. A self-serve creative hub, brand voice and intelligence tools, and production workflows that preserve team time for the highest-impact work.
  2. Growing distribution without spending a dollarI spotted the content already converting on social and turned it into a LinkedIn newsletter one partner and I could run weekly. 1.1M subscribers, all organic, and a format Coursera reused for other audiences.
  3. Feeding the paid creative beastI made launching paid creative as easy as sharing a sheet of URLs, and built a modular story library the paid team can remix on its own. Creative delivery down from months to weeks, and paid tests live in days instead of weeks.
  4. Building awareness through real storiesI wrote the briefs and directed agency and in-house teams on TV spots and documentary-style films built around real learners and learner insights. +7 points brand awareness, $19M in revenue over three years, 23M+ views across paid and organic.
  5. Making AI relevant to a skeptical audienceI researched what people in different jobs needed to hear about AI, wrote messaging for each, and tested it across channels before scaling. The framework now guides how Coursera markets by role.
  6. Finding product-market fit in a seven-week sprintAs the sole marketer on the sprint, I led the research, built the positioning and designed the channel tests. 170+ interviews, two personas with measurable intent, and a playbook for the company to use.
Leadership, AI in creative work, creative systems

Deciding where AI fits in our creative process, and where it doesn't

I established a culture of experimentation on a 27-person team to find the right places for AI in creative work, then shipped the tools and workflows that put it there.

I lead Brand & Creative for Coursera and Udemy across brand strategy, writing, design, video and events, while the two public companies merge. A big question for the team this year was AI: where it makes creative work better, where it doesn't, and how to find out without anyone feeling their craft was on the line.

  • Establish a point of view. We outlined an AI creative philosophy for the company: humans own the story, and AI comes along to support.
  • Make it safe to try. I created “Claude Show and Tell,” a session where anyone on the team could share an experiment, finished or not, and the rest of us could learn from it. Experimenting became part of the job rather than a threat to it.
  • Solve real problems. We found where our capacity or process had challenges and experimented to see if AI could help, so every experiment started from a bottleneck the team already felt.
  • Test, then scale. Everything starts as a pilot with one team or one campaign. We get feedback, fix what breaks, and only when it's working does it become a scaled solution the whole marketing org can use.
Result

I helped the team build systems to power on-brand creative at scale: a self-serve creative hub, brand voice tools, production automation and a co-brand design system for the combined company.

What I'd do differently: set proxy measures for creative's business value earlier (delivery speed, volume absorbed, agency spend avoided) instead of waiting for clean performance data across two companies.

My role: set the direction and the operating model, sponsored each experiment and tool, and owned the merger brand work with the executive team.

Zero to one, owned audience

Growing distribution without spending a dollar

I piloted the idea for a newsletter that turned into an owned channel reaching 1.1 million people every week.

Our consumer team wanted to shift perception and create value for people before they were ready to buy. The catch was we had no product support or marketing dollars to make it happen. I spotted an opportunity to take content that was already converting well, repackage it, and distribute it through LinkedIn's newsletter feature.

  • Start simple. Our constraint was that an issue could be written in under 20 minutes and read in under 5.
  • Test and learn. We produced six months of issues in one two-week sprint and used initial performance to guide what we built next.
  • Respond to reader feedback. The newsletter grew into the topics where we saw the most engagement: resumes, high-income skills, job search tips, and top certificates by domain.
Result

392K subscribers in six weeks. 1.1 million today, 149 weekly editions, 8.25M article views in the last year, a spike every Thursday and no decay. Readers registered and enrolled at a multiple of the rate of regular site visitors. Coursera later copied the format for its enterprise audience.

My role: concept, business case, template and pilot content; co-ran editorial with one partner; built the measurement.

AI-native workflows, creative systems

Feeding the paid creative beast

I made launching paid creative as easy as sharing a sheet of URLs, and built a modular story library the paid team can remix on its own.

Paid campaigns couldn't go live at launch because every title waited on hand-built assets, and the team felt limited in creative testing without foundational assets to build from.

  • Automate the floor. I found an agency that could take our top-performing format and turn it into a scaled system, so all our paid team needs to do is share the links to new products launching to get display, video and social copy.
  • Make it easy to remix. I identified our top story formats, worked with an agency to build foundational assets, and delivered a Figma library and After Effects files so the paid team can launch new creative tests in days, not weeks.
Result

80+ programs can launch from a spreadsheet of URLs. Creative delivery moved from months to weeks, and the paid team remixes foundational assets to launch tests in days.

What I'd do differently: we built creative faster than the testing process could absorb it. Plan the testing with the pipeline, or the creative shows up and sits.

My role: wrote the problem statement and both briefs, defined the ten stories, wrote template copy, directed the agency build and the handoff.

Brand films, storytelling

Building awareness through real stories

I wrote the briefs and directed agency and in-house teams on TV spots and documentary-style films built around real learners and learner insights.

Real learners on TV. I put real learners at the center of Coursera's TV work and made it always-on, led by the Professional Certificates spot. It lifted brand awareness 7 points year over year and generated $19M in revenue over three years. The testimonial spots, on TV and in paid digital, have drawn 23M+ views across paid and organic.

Rooting campaigns in insight. The biggest obstacle anyone faces is the story they tell themselves, so we gave that inner voice a new script: a pep talk from a mentor, from TV to social to email. I wrote the brief and shaped the agency's idea.

Organic storytelling. Short films featuring learners selected based on priority products, built for the campaign moment they ran in. I built the process to find, vet and shape each story.

My role: wrote the platform brief; led the storytelling, the sourcing and the match to campaign; directed agency and in-house production.

Positioning, messaging frameworks

Making AI relevant to a skeptical audience

I researched what people in different jobs needed to hear about AI, wrote messaging for each, and tested it across channels before scaling.

Most of the audience couldn't see how GenAI applied to them. A category campaign (“learn AI”) would have been noise. People wanted to know what AI meant for their job, so we built the messaging by role and tested across channels.

  • Establish a framework. One shared problem, three ways to message based on audience research, and clear guardrails: no fear-based marketing, and the learner is always the hero.
  • Test before scaling. I tested role-specific against category messaging on paid social first, then rolled the winners across search, email, on-site and video.
  • Turn learnings into a playbook. The learnings became the company's role-based marketing framework, with role-specific email series, career guides and job search workbooks.
Result

Each role responded to a different message: data professionals to credentials, developers to employability, everyone else to efficiency. The company adopted that as the rule. Role-specific email beat standard targeting with double-digit click-through; role creative won on video completion. The framework carried into the 2025 growth strategy.

What didn't work: job-specific paid creative helped, but not enough to scale without first-party role data. I recommended data to capture in product and marketing channels to power future targeted marketing, and next time I'd build those entry points before the media plan.

My role: accountable owner. Wrote the framework and the test plan, directed channel leads and creative, authored the learnings.

Audience research, positioning

A seven-week sprint to find product-market fit

I needed to find repeatable ways to identify degree-interested learners in our base, and turn it into a playbook other teams could use.

Coursera's executive team wanted to know whether learners in Professional Certificates could be reliably recruited into degrees that give credit for that prior learning. The constraint was that every persona had to be identifiable through on-platform data. I was the marketing lead who led research, built positioning, and designed channel tests.

  • Research fast. 170+ learner interviews in two weeks, live and AI-led, a first for Coursera.
  • Position using their words. 12 personas across four degrees, each with positioning drawn from what learners actually said.
  • Test where they are. Persona-targeted tests on LinkedIn, Facebook and in-platform.
Result

Two personas with statistically higher degree intent, one at the 40% PMF bar. A value prop 85% of learners found appealing. LinkedIn beat Facebook by multiples on qualified traffic at similar cost. Repeatable playbooks and reusable templates to guide future consumer-to-degree marketing.

What I'd do differently: fix the conversion flow the paid tests exposed before spending more on channels.

My role: marketing lead. Ran the paid tests and analysis, co-led interviews and positioning, wrote the playbooks.

About

  • 2012 to 2016Joined Seelio, a seed-stage startup, and built critical functions from zero to one, including marketing, communications, sales enablement and account management, through its acquisition by Keypath Education.
  • 2016 to 2019Joined Coursera early and helped build the consumer brand as it nearly doubled its learner base from 23 million to 45 million.
  • 2019 to 2024Led campaigns and storytelling for the consumer brand as the company went public in 2021, establishing its brand platform and testing entrance into new categories and product offerings.
  • 2025 to nowLead the unified Brand & Creative team for Coursera and Udemy through their merger, guiding a culture of experimentation as we deliver creative excellence and figure out where AI fits in the process.

Where to learn more

I write The Messy Middle, on marketing, AI and leading through complexity.

On LinkedIn you'll find my full history and recommendations from people I've worked with, including:

“[Her] standout quality is her strategic foresight.”Aaron McCullough, former Chief Product Officer, Coursera
“She led with clarity and could achieve even the most ambiguous project with excellence.”Kim Caldbeck, former Chief Marketing Officer, Coursera
“Although she was not a cofounder, she acted every part as one through her passion, commitment, and problem solving skills.”Moses Lee, Chief Executive Officer, Seelio

Get in touch

The fastest way to reach me is a message on LinkedIn.