What Is a Marketing Engineer? Why I Threw Out Four Job Titles and Kept One

A marketing engineer builds the system that does the marketing, instead of doing the marketing by hand. That is the whole definition. If you typed what is a marketing engineer into Google and landed here, you already have the answer and you can leave. The rest of this is why I threw out four job titles to keep that one, and what changed in the last two years that made the word necessary.

For most of the last five years, my LinkedIn headline was a confession that I didn’t know what I did for a living.

Software Engineer | Digital Marketing Strategist | AI Automation | Data Analysis

Four titles, three pipes, zero clarity. It read like a résumé that had been through a blender. The worst part is that every one of them was true. I’ve been writing software for more than twenty years. I run agentic SEO for law firms. I build automation pipelines. I live in dashboards and query logs. Nothing in that headline was a lie.

It was just useless.

Nobody hires a pipe character. When a managing partner at a personal injury firm reads four titles stacked like that, they don’t think “versatile.” They think “I have no idea which of these I’m buying.” A headline that tries to say everything says nothing, and I was paying for that in every first call.

So I cut it down to two words: Marketing Engineer.

How I ended up with four titles instead of one

Titles accumulate the way browser tabs do. You open one. You never close it.

I started in software. Straight engineering: PHP, databases, servers, the unglamorous middle of the stack. I went to law school from 2000 to 2003 focused on technology and intellectual property, didn’t finish the JD, and went back to building. I am not a lawyer and I don’t play one on a website. Then I ran an agency, Driliquid Studio, and suddenly I wasn’t just shipping code, I was responsible for whether the phone rang. That added “digital marketing.”

Then I sat in CTO seats and CMO seats. Two jobs, two vocabularies, both in the same week. That added “strategist,” because that’s the word you reach for when you decide what gets built and what gets said about it.

Then AI showed up and I added “AI automation,” because I was building agent pipelines. Then “data analysis,” because none of the agent work means anything without measurement.

Four titles. Each one honest. Each one a fossil of a previous era of my career.

Here’s the test that broke it for me. I looked at what I had actually shipped in a given month and asked which title covered it.

I built a production remote MCP server, Crawllr, that client workflows now call directly. Software engineering? Yes. Marketing infrastructure? Also yes. I built AI intake agents that qualify law-firm leads in conversation before a human ever sees them. Engineering, or conversion work? Both, obviously. I built multi-agent content pipelines that audit, brief, draft and route legal content at a volume no single writer matches. That isn’t content marketing and it isn’t software development. It’s a factory.

None of the four titles described the factory. They described the individual tasks the factory replaced.

What is a marketing engineer? The definition I actually use

A marketing engineer is the person who builds the system that does the marketing, rather than the person who does the marketing. Everything else is elaboration.

The output is leverage, not deliverables

Nick Lafferty, founding marketing engineer at Profound, puts it in one line I keep coming back to: the output of a marketing engineer is leverage. He splits the work into two halves. There’s the doorbell, which is automation that watches things so you never sit in a dashboard again. Then there’s the builder, which is capability that wasn’t technically possible two years ago. His example is a machine that turns every webinar the company has ever recorded into articles structured for how AI systems read content, then refreshes them monthly. That’s not a content calendar. That’s a plant.

Why the title arrived in 2026 and not in 2019

Greg Isenberg put a price on the role, arguing that the best marketing engineers will earn $1M a year for the same reason forward-deployed engineers command what they command: they walk in and build the thing that produces pipeline. I have no idea whether that number holds. His historical arc is the part worth keeping.

Traditional marketing was about making people care. Story and psychology. Don Draper.

Digital marketing was about acquiring customers through channels you could measure. SEO, ads, landing pages, funnels.

Growth hacking was about using product and data to build loops. Activation and retention, instrumented.

Marketing engineering is about building a system that learns. That last word carries the weight. A marketing system that gets smarter every week without a human manually feeding it insight was not buildable before agents. Now it is, so the role exists.

Every era of marketing produced a new most-valuable marketer. This is that happening again, on schedule.

The search data says the term is already rotating

I don’t take anyone’s word for a trend, so I pulled the demand. One caveat first, because these are the only hard numbers in the piece. What follows is a single source: US monthly Google search volume as estimated by DataForSEO, covering September 2025 through August 2026, pulled 18 September 2026. Nobody outside Google counts searches. Every tool models the figure instead. DataForSEO builds its estimate from Keyword Planner data blended with clickstream signal, and Keyword Planner itself reports rounded averages that lump close variants together. I have not cross-checked any of this against a second vendor. Read the direction of the curves, not the digits.

• “marketing engineer”: an estimated 320 searches in December 2025, 880 by June 2026, holding in the 700 to 900 range since.
• “what is a marketing engineer”: roughly 40 a month to 170 a month over the same window, about 4x.
• “growth engineer”, the closest adjacent title: an estimated 720 a month in September 2025, down to 480 by August 2026.

That last line is the one I’d underline. The two terms aren’t growing together. Widen the error bar as far as you like on a single vendor’s model and the shape holds: one title is handing off to the other. And the steepest curve is the definitional query. Most of the people searching still can’t define the job. Profound launched a dedicated marketing engineer job board in June 2026 and employers started posting against the title with real salaries attached.

The gap between “term is spiking” and “nobody can define it” is the whole opportunity. It closes.

The part most people are getting wrong

Holly Chen runs growth marketing at Samsara, which reported $1.99 billion in ARR in its most recent quarter. Her marketing organization is 230 people and it is running roughly 90 AI agents. She said the most important sentence in this entire discourse:

“We cannot hire a marketing engineer to AI-fy or transform the organization. Every single marketer needs to be a marketing engineer.”

Why one hire cannot AI-fy an organization

She’s naming a failure mode I watch play out constantly. Leadership knows AI matters, feels behind, and tries to solve it with a job req. One hire, one department, transformed. It doesn’t work, because the people closest to a broken workflow are the only ones who know where it’s broken.

I agree with her, and I want to push on it, because both things are true at once and the money sits in the nuance.

Every marketer needs to become a marketing engineer. Yes. That’s baseline literacy, and Samsara builds it with hackathons and monthly AI power hours. A review layer promotes the good experiments into production.

Somebody still has to own the orchestration layer

At 90 agents you hit problems that bottom-up building cannot solve. How do agents talk to each other? Who maintains them when the person who built one leaves? How does anyone discover that agent number 47 already does the thing they’re about to rebuild? Chen names the orchestration layer herself as the thing on her mind.

That’s an architecture problem with versioning, dependencies, shared context, and a maintenance burden. It is not a marketing problem with AI sprinkled on top. So: everybody builds, and somebody owns the system the building happens inside. Those are two jobs. The second one is mine.

Marketing engineer vs. a marketer who uses ChatGPT

Here’s the line, and it’s sharp.

A marketer using AI opens a chat, asks for ten LinkedIn posts, copies the one they like into a doc, and closes the tab. Next week the model starts from zero. The work evaporated.

A marketing engineer builds a system where that work persists and compounds. Customer language from last month’s intake calls lives in a file the content agent reads. The objections that killed three deals live in the positioning file. The hooks that drove qualified replies are logged with their results, so the next draft starts from evidence instead of from nothing.

On my side of the fence the difference looks like this. The “using AI” version of SEO content is asking a model to write a blog post about a keyword. The engineered version pulls Search Console data, ranks opportunities by buyer intent rather than raw volume, checks the CMS for cannibalization against what’s already published, reads what’s actually ranking, drafts against the client’s real positioning, writes the metadata, proposes internal links, routes the piece for human approval, then logs what happened so the next run starts smarter. Same nominal output. Completely different asset.

That’s the shape of everything I build now. CreditPullEngine is an FCRA-compliant credit report delivery platform running on a direct Equifax contract, a system where the compliance and the marketing are the same object. Medellín.co is a local directory with an AI concierge, structured for retrieval by agents rather than by humans scrolling. The intake agents I build for firms are not chatbots. They’re the first mile of the sales process, running unsupervised, at 11pm on a Saturday.

I do all of it as one operator. No bench, no account layer, no department. Twenty years ago that workload required a floor of people. The reason one person can hold it now isn’t that I got smarter. The leverage changed. That’s what the title is describing: not a skill set, a leverage ratio.

If you run a law firm, here’s the part that matters

You are not hiring a job title. You’re deciding who owns the path from a search to a signed matter, and whether that person can actually touch the systems in the middle of it.

Most agency org charts split that path across four people who never speak, which is exactly how tracking dies in a redesign and LSA calls ring into a voicemail box at lunch. I’ve written the hiring comparison in full, and the full acquisition map lives here. The title is only shorthand for the thing you should be testing in a first call: can they change the system, or only report on it?

Why every marketer becomes this

Because the alternative is competing on the thing that just became free.

AI made average marketing cheap. Not good marketing. Average marketing: competent blog posts and passable ad copy, infinite and instant. If your value was producing that output, your value now approximates the cost of tokens.

What didn’t get cheap is knowing what to point the system at. Chen prefers “judgment” to “taste,” and she’s right, because judgment implies it was earned. It comes from having watched a hundred landing page tests fail. A model can generate a thousand variants. It cannot tell you which one deserves a client’s budget, because it has never lost a client’s budget.

So the job compresses into this: judgment, pointed at systems, at scale. You still need customer understanding and positioning. You now also have to be able to build the thing that executes on it. Agents are becoming a commodity. What you aim them at is the moat.

The headline

I didn’t change my title to sound cooler. I changed it because four titles were describing four historical layers of one career, and the work underneath had collapsed into a single thing.

The engineering didn’t go away. Neither did the marketing or the measurement. They stopped being separate jobs. A marketing engineer isn’t a marketer who learned to code or an engineer who learned marketing. It’s the role that exists after those two stop being different roles.

If you’re staring at your own headline right now, counting pipe characters, the useful question isn’t which one to cut. It’s what all of them were secretly describing.

For me it took four titles and twenty years to find two words. If your firm’s marketing is currently four vendors and nobody owns the system between them, that’s the same problem wearing a different suit. Write me at hi@carlosarias.com and tell me where it’s breaking.