Back in 2001, I was installing Ethernet cables at companies on the East Coast. At one of those offices, the manager refused to use use email and barely used a computer. I remember thinking that was strange, because I’d already been on the internet for over a decade at that point. But he was near the end of his career, so he was going to keep doing what worked for him until he retired.
A few years later, I noticed smaller versions of the same thing.
Some people were still hesitant with spreadsheets, some people still hesitated to use email, and any new application they faced was like them trying to program a VCR. This was in the early 2000s, and it was obvious there was a gap between people who adopted technology and people who didn’t. What stands out to me now is that I saw that gap inside marketing teams too, even on very successful online marketing teams.
—
Right now we’re at another one of those moments, except this one is much bigger. AI and LLMs let people generate complex things at a level that would’ve required serious expertise just a couple years ago.
That’s the good news.
The bad news is that generating something is not the same as maintaining it.
If you’ve written software before, you already know this.
There’s a saying in software that I find more and more accurate over the years: when you’re 90% done, you’ve got about 90% more to go.
When LLMs can do that first 90% in an afternoon, it starts making that other 90% seems like it can _also_ be done in an afternoon. It’s incredibly deceptive.
—
The details are where the pain starts. Integration is difficult. API changes are annoying to maintain. Troubleshooting network issues is crazy-making. And debugging is a whole other layer of challenge. And for anyone who has never built or maintained software before, they usually don’t have a good feel for how quickly this stuff turns ugly.
So now we have insane productivity gains at the same time that we’re accumulating technical debt. If you don’t deal with that debt early, it compounds. Six months from now, or twelve months from now, something breaks, and now you’ve potentially lost weeks trying to get it working properly again.
—
LLMs help, obviously. They’re great at finding problems and helping solve them. But as the code base grows, they get more confused too. They start making changes they shouldn’t make. And if the person reviewing those changes doesn’t really understand the system, that’s where things go sideways fast.
I’ve seen this enough now that I think a lot of teams are underestimating the risk. Once the complexity gets beyond what you can hold in your head, and beyond what the model is handling cleanly, you’re in a very bad place. You can absolutely trash a year’s worth of work in one night if you don’t know what you’re doing.
—
That’s why I believe marketing teams are going to need a fourth core role. Traditionally, you’ve got some version of an account manager or team lead, an analyst that handles a lot of data management and reporting, and a creative who could be a copywriter, designer, etc. That’s been a pretty normal structure for a long time.
I believe this long-standing structure for marketing teams will be evolving for the first time in decades and adding a new role.
That new fourth role is a marketing engineer.
That person smooths out operations for everyone else. They collect and structure data better. They pull information together from third-party systems. They improve reporting. They design processes and workflows so the team moves faster without building garbage behind the scenes.
For the creative, that might mean better systems for finding content opportunities. For the account manager, it might mean scripts and internal tools that reduce busywork so they can spend more time on client relationships. For the analyst, it might mean getting hard-to-reach data delivered in the exact format they need.
—
They also prototype internal tools. That looks different on every team. You might use something like n8n to move quickly at first, and that’s fine for prototyping. But long term, a lot of teams are going to need actual scripts, actual code, actual systems running somewhere reliable.
So when I say marketing engineer, I don’t mean somebody playing with prompts. I mean somebody who understands systems, automation, data flow, maintenance, failure points, and how to build useful internal tooling without burying the team in technical debt.
I think that role becomes extremely important over the next few years. Not because marketing is becoming engineering, but because modern marketing teams are going to break if nobody owns the engineering layer inside the work.
I have a lot more to say about the responsibilities, etc, of a marketing engineer, but I’ll expand on those ideas in future videos.
—
To help grow this role on teams, I’ve created a new site and YouTube channel, MarketingEngineer.io. I’ll have at least a couple new videos each week helping you as a digital marketer to understand the engineering that influences your campaigns, and also to learn how to avoid common engineering pitfalls, while increasing your teams productivity as we bring AI and LLMs into every part of the digital marketing campaign and management.
This will be anything from improving your copywriting workflows (for example, did you know that the most important tools for a copywriter are a microphones and an AI code editor?), to understanding how for-you-pages and home feeds rank content, to more technical talks on how neural networks and LLMs work behind the scenes, so you can make even better use of them for your clients.
So, check out MarketingEngineer.io, subscribe to the channel on YouTube, and follow me on LinkedIn to continue evolving at the intersection of marketing and engineering.