One test sorts a group standard
from a local choice

Governance lists grow. They start as a page of things the group should standardise and end as a spreadsheet nobody can maintain, because every country adds the items it cares about and nothing ever comes off. I've watched that spreadsheet get written more than once, and the problem is never that the items are wrong. It's that there's no test for whether an item belongs on it.

Here's the test I use. It has two halves, and an item only has to pass one.

Does it roll up, or does it move

First half: does this field appear in a number the group reports? If the board sees a figure that sums across countries, everything feeding that figure is a group standard. Not negotiable, not localisable, not a preference. The definition has to be identical or the sum is fiction.

Second half: does the thing itself move between countries? Assets that transfer, people who transfer, contracts that get novated across entities. If a record crosses a border, both sides have to read it the same way, so its structure is a group standard even when nobody reports on it.

Everything else is local. Not tolerated as local, not local until we get around to it. Local as the correct answer.

Why the answer keeps surprising people

When I've run a real list through those two questions, most items fail both. Approval thresholds, notification rules, the labels on a status field, who signs what, how a workflow is staged, which day of the month a review runs. None of that rolls up and none of it travels. It's operational preference, and a country that's allowed to set it will run the platform better than a country that has to file a request to change it.

The list that survives is short. Usually it's the chart of financial dimensions, the entity and hierarchy structure, the identifiers that reconcile against systems the group already owns, and the handful of date fields that drive group reporting. You can hold that in your head. That's the tell that you've got it right.

The scale that makes this concrete

The platform I architected at Tango served brands operating in more than 100 countries, in 16 languages. Consider what a long governance standard means at that shape. Every item on it has to be defined once, translated 16 times, and then enforced in 100 or more places where local practice already has an answer. A twelve-item standard is achievable. An eighty-item standard isn't a standard, it's a backlog of exceptions that will be granted quietly, one country at a time, until nobody knows what the rule is.

The cost of a group standard isn't in writing it. It's in defending it, in every language, forever.

What to do with the local column

Give it a real boundary rather than an absence of one. Local choice doesn't mean anything goes. It means the group defines the shape of the field and the country fills it in. Status values are locally defined; the fact that there's a status field, and that it's mandatory, and that it has a closed set of values, is group. That's the split I'd hold to, and I'd write it in those terms so a country reading the standard knows exactly where its freedom starts.

Write down which test each item passed

This is the part people skip and it's the part that keeps the list short. Next to every group standard, record which half of the test it passed and the specific report or the specific movement that justifies it. Now the list is falsifiable. When somebody proposes an addition, they have to name the number it rolls into. When the report changes, you can find the standards that no longer have a reason to exist and remove them.

A governance list without that column only ever grows. With it, the list can shrink, and a list that can shrink is one people will still be following in year three.

Sources: The 100 or more countries and 16 languages figures describe Tango Edge, the multi-tenant platform Bill architected and led development of as VP of Engineering. Recorded in docs/bill-thornton-resume-2026.pdf in this repo. Customer names are not published here.