AI adoption fails on workflow, not technology
Sarah Kennedy Ellis ran marketing at Marketo through its $4.75B Adobe exit. Now she is VP of Global Demand & Growth at Google Cloud, where the division just posted $58.7B in revenue and sits on a $460B backlog driven by AI demand.
Her team is Google's internal test case for agentic marketing. Not the demo version. The actual day-to-day operation running on AI agents, feeding what breaks back to product teams.
The numbers matter: 70% faster asset production for the Gemini in Chrome launch. Thousands of creative assets, production time dropped from weeks to days. The interesting part is not the speed. It is that conversion rates went up, not down. Personalization at individual level, at scale that was not possible before.
That is the opposite of what usually happens when you scale content.
Why teams are not using the tools you bought
The biggest blocker to adoption inside Google Cloud is not model quality. It is workflow friction and behavioral change required to break existing patterns.
Ellis's framing: the greatest friction in a workflow is the biggest inhibitor to adoption, well over agent quality. Teams spending time on change management and training are getting productivity gains. Teams waiting for a better model are stuck.
If your sales team is not using the AI prospecting tools you licensed, the answer is usually not the tools.
Training that fits into 5 minutes a week
Time is the constraint. Google built training around that reality: AI Boost Bites, 5 to 7 minute videos covering one specific task. Early ones were basic (how to create slides with Gemini). They evolved into multi-agent orchestration across campaigns.
They gamified it. Internal competitions, badges, proof you completed the task. Ellis called badges "gamification from 20 years ago." It worked. The series is now public on YouTube, past a million views.
The top 20% of AI adopters inside Google Cloud marketing are the same people who did the most training. That correlation matters for deployment. The instinct is to buy licenses and wait. What drives adoption is deliberate skill-building.
Where the ROI actually showed up
Ellis's rule for where to deploy agents: high volume plus limited human judgment required for high-quality outcomes. That is the zone where agents earn their keep.
For Google Cloud Next, her team killed the opening video three weeks before the event. It was using AI, but not enough to showcase what the product could do. They rebuilt it: creative director sketched concept on napkin, fed it into image generation, added motion, stitched it with custom agents. 138 Easter eggs referencing Google's history.
The technical wall: resolution. Max output was 4K. The screen at Next is the size of a 737. They upscaled to 12K using a custom Deep Mind model.
The previous version required an agency, bigger budget, much more than three weeks. That path collapsed into an internal team using their own tools. The same project was impossible a year earlier.
What this means for ANZ sales teams
Google Cloud maintains significant ANZ presence through startup programs: $200K in cloud credits, up to $350K for AI startups. The region benefits from Google's global AI infrastructure driving that $460B backlog.
For sales teams selling AI tools or building AI-powered outbound motions, the lesson is not about the technology. It is about workflow integration and training that fits into the time your reps actually have. If adoption is low, look at friction first, agent quality second.
The comp structure for AI-focused marketing and sales roles is shifting. Teams that master agentic workflows are pulling ahead. That skill gap is showing up in attainment data and, eventually, in OTE.
Ellis came from Marketo and Adobe enterprise before Google Cloud. She has lived through platform shifts before. Her read: this one is real, but most teams are stuck on implementation, not capability.