What mechatronics taught me about marketing

I trained as a mechatronics engineer, and the main thing it taught me is that machines punish vague thinking. A badly calibrated sensor doesn’t care how enthusiastic you are. Set a tolerance too tight and the part costs three times as much and works no better. When my best friend and I started a writing agency in 2021, I assumed I was leaving all that behind. Instead I found the same physics in softer clothes.

Sensors before actuators

Mechatronics drills the order of operations into you early: measure first, act second. An actuator driven by a bad sensor goes full power in exactly the wrong direction, with total confidence. A lot of the marketing I saw in our SEO audits worked the same way. Actuators everywhere: more posts, more channels, more ad spend. Sensors almost nowhere. There was rarely a baseline, and nobody could say which numbers on the dashboard were signal and which were decoration.

So the first step on any project should be the unglamorous one: instrumentation. Before anyone publishes another post, I want to know which numbers move when the business moves, and which ones just move.

Closed loops beat forecasts

A control loop doesn’t need a forecast. It measures the gap between where it is and where it should be, corrects a little, and measures again. Small corrections, made often. Machines that look impossibly precise are mostly doing just that, many times a second.

In marketing, that means publish, measure, adjust, as often as the channel can respond: weekly for ads or email, monthly for search. A content strategy written in January and still defended, untouched, in August is an open-loop system. Open-loop systems fail quietly. They drift until someone finally looks up and asks why nothing has worked for a year.

Tolerances, not perfection

No engineer specifies a “perfect” shaft. They specify a tolerance: this dimension, plus or minus this much. Anything tighter than the part needs costs more and changes nothing. Somewhere between the workshop and the marketing meeting, we lost that discipline. At the agency I watched writers, me included, spend an afternoon polishing a headline whose click-through rate nobody would ever check, on a page that would have done its job with half the polish.

Precision where it matters, tolerance where it doesn’t. In both worlds, that’s most of the craft.

The translation table

If I had to boil down what engineering taught me into a marketing checklist, it would look like this:

• Measure before you act. Analytics before output, always.
• Close the loop. Every published piece should feed the next decision.
• Set tolerances up front. Decide what “good enough” means before you start, not at eleven at night.
• Watch the interfaces. Systems tend to break at the handoffs: brief to writer, draft to editor, human to AI and back.

That last one matters more every month. The AI systems I build now are mostly interface engineering: deciding exactly where the machine’s work ends and human judgment begins, and making that handoff impossible to skip. In the content pipeline I built in 2025, models do the research and the drafting. At the end, a person checks every article, clicking through the source links and hunting for anything the model made up. No article goes out without that check.

For a while I thought I’d left engineering behind. I hadn’t. I still debug systems all day. The only real difference is that when something goes wrong, a machine shows you an error code. An audience is more polite than that. It just quietly leaves.