The thing under the thing.

I have a bad habit of looking for the plumbing.

The habit makes me mildly annoying in meetings and actively dangerous around dashboards.

Someone will show a chart with a nice upward line, and while everyone else is nodding like responsible adults, I am already thinking about the form field, the source logic, the handoff, the duplicate account problem, the sales rep who changed the lead source because the dropdown looked wrong, and the one campaign nobody wants to admit was tagged as "Other" for 11 months.

This is how my brain works. I wish it had chosen something cooler. Classic cars. Regional Italian cooking. The history of maritime disasters. Anything with better dinner party upside.

Instead, I notice the thing under the thing.

A campaign is rarely only a campaign. It is a test of whether the audience definition is real, whether the message has proof behind it, whether the website knows where to send people, whether sales knows what to do next, whether the CRM captures the truth, whether leadership can read the result without squinting.

A deck is rarely only a deck. It is a fossil record of how well the company understands itself.

A dashboard is rarely only a dashboard. It is a live animal with a complicated digestive system and several places where the food can go bad.

This used to feel like a liability. Earlier in my career, I thought the job was to make better visible stuff: better pages, better copy, better campaigns, better presentations. I still care about all of that. Taste matters. The visible layer is where people decide whether they trust you enough to keep reading.

But the visible layer can lie.

A beautiful page can route the wrong person to the wrong team. A sharp campaign can generate leads nobody follows up on. A board deck can make the quarter look orderly while the pipeline is quietly turning into soup. A case study can sound impressive while the proof behind it sits in six different folders, two abandoned docs, and somebody's memory of a Slack thread from 2023.

At some point, the surface stopped being enough. The machinery underneath had become too deranged to ignore.

That is probably the cleanest explanation for how I ended up calling myself a GTM Engineer, a phrase I like and mistrust in equal measure. It sounds a little invented because it is. So are most useful job descriptions before the market catches up to the work.

The work itself is old and very human. Find the gap between what the business says it wants and what its systems make possible. Then build the bridge.

Sometimes the bridge is a lead routing rule. Sometimes it is a claims register. Sometimes it is a 35-metric funnel model, which sounds like a cry for help because in some ways it is. Sometimes it is a prompt library with enough governance around it to stop AI from writing like a haunted McKinsey intern. Sometimes it is a simple naming convention that prevents 4 people from wasting 40 hours.

The common thread is that the thing people are arguing about is usually not the thing.

Sales says the leads are bad. Marketing says sales is slow. Finance says the attribution is mush. The CEO says the story is unclear. The team says there is too much work. The dashboard says everything is fine, which is exactly what a guilty dashboard would say.

Underneath, there is usually a missing definition, a broken handoff, a fuzzy owner, a weak proof standard, or a workflow that depends on one heroic person remembering to do something at the right time.

Heroics are expensive. They also do not scale, except as folklore.

I have a lot of affection for the heroic phase of a company. Everybody has 6 jobs. The calendar is an active crime scene. Someone is making edits to the deck at midnight while eating cereal over the sink. You can get weirdly close to people under those conditions. There is a little foxhole camaraderie in it.

But you cannot run a mature function on adrenaline and cereal.

Eventually the work needs lanes. It needs standards. It needs a way to tell the difference between urgent and merely loud. It needs a system that keeps working when the person who knows the whole backstory is on vacation, sick, burned out, or finally done pretending the shared drive makes sense.

This is where I like to work.

I like the moment when the mess becomes visible enough to fix. I like taking a process that lived in everyone's head and putting it somewhere the team can inspect it. I like naming the ugly part. I like a good version-controlled deck, which is a sentence that would have embarrassed me deeply at 25.

I like when the system gets quiet.

Quiet is underrated. A good operating system does not announce itself all day. The lead goes to the right owner. The claim is approved before it ships. The source data holds. The handoff happens. The weekly review surfaces the problem early enough to matter. Nobody gets a medal because nobody had to sprint across the building with a metaphorical bucket.

That is the kind of work I trust.

A lot of my career can be summarized as discovering that the machine room is where the human stuff either gets protected or destroyed. Bad systems turn reasonable people into territorial weirdos. Good systems give them back a little dignity. They make the work less mysterious. They reduce the amount of guessing, defending, chasing, and apologizing.

That may be why I care about the plumbing so much.

The surface is where the company performs. The plumbing is where the company tells the truth.