Everything We Tried First to Get Marketers Using AI Failed

By Mark Boothe, Chief Marketing Officer, Domo

Here is the full list of things we tried to get our marketing team using AI, in order, all of which failed.

Enterprise AI licenses for everyone. Announcement emails explaining what was now available. Documentation. Optional office hours. A self-service AI interface anyone could log into. Enablement materials asking people to go research and experiment on their own.

Adoption from all of that was minimal. As I described it at the time, the vast majority of people didn’t do much with it. Some dabbled. The only real traction was as a writing assistant.

Every one of those failed tactics is standard practice. Every one is what a competent marketing organization does when it rolls out a new capability. And all of them produced the same outcome: deployed, underutilized, eventually abandoned.

I’m walking through this because I think many marketing organizations are currently mid way through that exact list, watching the same nothing happen, and drawing the conclusion that AI was oversold. It wasn’t. The rollout was wrong, and the reason it was wrong is uncomfortable enough that most companies would rather blame the technology.

The Dip Is Normal. The Response Usually Isn’t.

There’s a well-documented pattern with major technology shifts. When PCs arrived in offices, productivity didn’t rise for years. People didn’t know what to do with the machines, the software wasn’t there, nobody had best practices, and in the meantime the computers sat on desks making everyone slower. The value was entirely real and entirely in the future.

That shape is called a J-curve: baseline, then a dip, then a recovery that eventually clears the old level by a wide margin. The dip is where the learning, integration, infrastructure nobody budgeted for exponentially happens. The dip is also where most organizations make their worst decisions.

There seem to be two failure modes. The first is concluding the technology was hype, quietly reducing investment, and waiting for someone else to solve it. The second is over rotating in the excitement phase, deciding AI can carry more than it can, and restructuring around that assumption. We’ve all watched public versions of the second one: a company announces it’s replacing a human function with AI, and some months later is quietly hiring back for the role it eliminated.

Both come from treating the dip as a verdict on the technology rather than a description of where you are.

Why Tools-First Produces Shelfware

The specific reason our early efforts to use AI in marketing failed took us a while to name. It’s a distinction between two orientations.

A consumer of AI asks: what can this tool do for me? A builder with AI asks: what can I build that solves my problem?

That difference decides everything, because consumers wait for perfect tools and builders create solutions out of imperfect ones. Every marketing team on earth currently has access to imperfect tools. Which means the constraint isn’t the software, it’s whether your people are consumers or builders, and most naturally are oriented as consumers. That’s where the work comes in.

You cannot email someone into it. You cannot document them into it. Handing a person a license and telling them to go innovate asks someone hired to execute to spontaneously become an experimenter, with no change to how the organization understands their work and no evidence that failing at it is safe.

What Worked, and What It Costs

The thing that worked is embarrassingly unscalable: sitting next to one person, teaching them one-on-one, and building something that helps them achieve their business objectives.

This might sound a lot like “training,” which is a walkthrough of a tool’s features. But our approach was decidedly different from training. This was more individual enablement, which is getting the skills onto someone’s machine, opening their actual workflow, and building the automation together until it runs.

One example. A member of our events team with no engineering background… His only technical experience was some basic SQL years earlier, and he describes himself as a rookie; he was losing roughly twenty hours a week at peak season to administrative work. Eight to ten redundant forms per project, the same information retyped for different teams, timelines tracked by hand. Now enabled one-on-one with Claude Code, he built a workflow where one submission generates the project timeline, files the campaign request, and triggers assignments to the web, email, and creative teams. That 20 hours is now about two.

He built it by using a coding agent as a coach, letting it explain concepts he didn’t know and validating each step before executing. Other people on the team now come to him to learn how. His answer still is that he’s a rookie too.

His success didn’t come because of his technical prowess. It came because he was AI enabled to make his life easier. He was enabled to solve his own challenges with AI.

The Process That Got Our AI Initiatives Unstuck

In order to be successful, you have to change the hearts and minds of your people. You have to help them understand the vision and what’s in it for them. Changing the mindset will likely take you multiple months at the beginning. The second step, which should be overlapping with step one and will run for several months, is enablement. The third and final step is the implementation of tools. Tools should definitely be your last priority.

Most organizations run that in reverse and then wonder why adoption stalled. Tools without the first two produce shelfware. Tools with the first two become force multipliers, adopted in days rather than quarters.

Twelve months in, 93% of our marketing team members have documented, active AI workflows in their daily work. Nearly all of them have now coded their own agents. Today, the self-reported time savings is twelve hours per person per week. Additionally, average comfort with AI across the team moved from 6.6 to 7.9 on a ten-point scale.

None of that came from the tools. All of it came from the eight months of unglamorous work that had to happen before the tools meant anything (changing mindset and truly enabling team members for success).

That success dip, at the beginning, is going to happen. But make sure you are actually paving the way for future success. The dip will come whether or not you want it. The only real choice is if you get people deliberately bought in and enabled for success. That will only come with one-on-one enablement investments.

About the Author:

Mark Boothe is the Chief Marketing Officer at Domo (https://www.domo.com/), the AI and data products platform. Prior to Domo, Mark held marketing leadership roles at Instructure and Adobe. He worked as an adjunct professor in public relations at BYU and received his MBA from Utah State University.