Garbage in. Margin out.

We’ve been launching “one-liners” about data like t-shirts out of a cannon.

“Garbage in, garbage out.”
“AI called. It’s not you, it’s your data.” (I have swag that says this.)
“AI doesn’t automate chaos. It automates chaos faster.”

Which is why “Pass-ish” is a bigger problem than most people think. (Let me explain.)

Building a data foundation that sticks doesn’t start with AI. It starts with helping humans understand why it matters. Most teams aren’t careless. They’re practical. If nobody explains the reason behind a standard, it just feels like bureaucracy.

Take meeting names.

A meeting called “Meeting with Dan” tells a human very little (and AI even less).

A meeting called “Contoso_AP_Automation_RequirementsReview” tells a human AND a system exactly what happened.

That one naming convention makes it possible to automatically file the recording, generate the summary, route the action items, and connect project knowledge to the right customer and workstream.

Same meeting. Very different future.

But the bigger issue usually isn’t meetings. It’s operational language.

Humans know:
Complete = Completed = Done
In Progress = In Process = Active
Pass = Passed

But don’t make AI mull it over. Agree on one. Problem solved.

And then somebody creates “Pass-ish.” True story.

Everyone on the project knows what it means. Ish.

But now try to automate reporting, calculate delivery health, flag project risk, or decide whether work is actually actionable.

Does “Pass-ish” mean:
> Ready to proceed?
> Ready with caveats?
> Waiting for approval?
> Failed, but the best we’re going to get? So ... proceed? (yikes)

A human can ping the source. “WTH is pass-ish supposed to mean?”

AI can’t. Neither can your workflow. That’s where the leak shows up.

The “what” is standard naming conventions. The “why” is automating the busywork and enabling outcomes.

Where an actionable backlog actually means something. Where “ready” means ready. Where “blocked” means blocked. And where every status drives the same decision tree no matter who’s reading it.

Here’s the catch: standards only stick when the team understands the why and feels ownership in it. Most data problems aren’t created by bad people — they’re created by good people who assume somebody else owns the process.

Sales owns part of the data. Delivery owns part. Finance owns part. Leadership owns part. Which, uncoordinated, means no one actually owns it. That's a whole separate leak — and the next WTH episode. (Episode 5: Human in the Loop. Accountability Not Found.)

For now: teach the why, standardize the what, and let AI do the busywork — instead of automating your chaos faster.

So… what’s your team’s “Pass-ish”? Every team has one. Drop it in the comments. I’m collecting.