Who Actually Owns the Automation After It Breaks

Somewhere right now, an automation that used to work perfectly is quietly failing, and the only person who'd notice left the company eight months ago.

That's not a hypothetical. Industry writers have a name for this exact thing: a zombie bot, an automation that's still technically running but that nobody on the team understands, maintains, or would even know how to fix if it stopped. CIO.com has been writing about it for years. It's common enough to have its own vocabulary, which should tell you something.

The Part Nobody Budgets For

Here's the pitch every automation vendor leads with: look how much time this saves. Here's the part they leave out of the pitch: the build is the cheap part. Multiple sources, including an actual academic multi-case study out of Aalto University, point to the same pattern, that the real cost of automation lives in maintenance, monitoring, and the slow drift of "wait, why does it still do that" that shows up eighteen months later. One frequently cited industry estimate puts licensing at only a quarter to a third of total lifetime cost. I'd treat the exact number skeptically (it's widely repeated but thinly sourced), but the direction is not in dispute anywhere I looked.

The pattern behind it is almost always the same: automations get inherited, not engineered. Someone builds it, someone else half-understands it, a third person patches it during a busy week without touching version control, and eventually there's no one who could explain what it actually does anymore, only what it's supposed to do.

That gap is where things quietly die.

The Checklist to Demand Before You Sign Anything

Before you approve any automation project, get a straight answer on these four things. Not eventually. Before the kickoff call ends.

1. Who is the named owner? Not "the team." A person, with a name, whose job includes noticing if this breaks.
2. Is there a change log? If the logic gets modified six months from now, is that documented anywhere a human could find it, or does it just live in someone's memory until that someone quits?
3. What happens when it breaks? Not "we'll get an error," an actual alert, to an actual person, within an actual timeframe.
4. Is there a review cadence? A calendar date, not a vague "we'll check in periodically" that never happens because nobody owns the checking either.

If a vendor can't answer these cleanly, you're not buying an automation. You're buying a slow-motion mystery with your name on the invoice.

Why This Isn't the Same Problem as Bad QA

I've made the argument elsewhere that "we'll just have someone manually check it" is often more expensive than building the automated gate in the first place. This is a different problem, and I don't want to blur them together. Even a perfectly QA'd automation becomes a liability the moment nobody's assigned to own it. Quality at launch and ownership after launch are two separate failure modes, and most SMBs only ever get warned about the first one.

What This Costs You If You Skip It

Picture the automated version of Weekend at Bernie's, except instead of propping up a body at a party, you're propping up a workflow nobody remembers how to fix, still nodding along at meetings like everything's fine. It's funny until it's your accounts payable process.

The fix isn't complicated. It's just unglamorous, which is probably why most proposals skip it. A named owner, a change log, an alert, a calendar date. That's the whole ask. Any vendor who can't commit to those four things in writing is telling you, politely, that ownership was never part of the plan.

---

Smatthew Cohen is an AI Operator and the founder of Ingenium Vector. Before that he ran a sales firm called Tortoise & Rooster for twelve years, helping boutique manufacturers who couldn't afford the agencies that were ignoring them anyway. He builds things now.

Further Reading

• How to Tell If Your Marketing Agency Is Selling You Smoke: the questions that expose a vague proposal
• What n8n Automation Actually Costs (and Saves) for a 10-Person Firm: the build-cost side of the same math