WEEK 45 · DATA

Classify the ten things people actually touch.

A classification scheme too complex to use protects nothing, so start with ten files and a handful of labels you can say from memory.

Submitted by Neil Watkins · MondayMove

May 15, 2026

I helped build a classification matrix worth seven figures, and nobody ever used it.

That program ran for hours, then days, then weeks. We interviewed hundreds of people, or it seemed that way, and everyone had a different way of describing the classification. The taxonomy was broken. Nobody knew where the data was or whether it sat in the right place, and everyone found exceptions to their own statements before the meeting ended.

All that work produced a matrix so vast it was unusable, and paring it down was a literal knife fight. The complexity itself was the barrier. We ended with more than 27 categories, and every one of them had a sponsor passionate about its necessity. Each sponsor insisted their category was so unique it required its own identification, storage, use and protection, and they fought every attempt to put it in what they called a generic category.

This is what kills this type of project.

I don't trust the postmortems on programs like that one to say so honestly. They blame the tool, or the budget, or a business that wouldn't engage. What actually happened is that we tried to describe all of the data before protecting any of it, and every interview made the description bigger.

So flip the order and classify the part people actually touch. File shares, SharePoint sites and databases already record what gets opened, and that record is a ranking your colleagues built for you without anyone having to ask them. The files opened every day are the ones that get emailed, copied, synced to a laptop and attached to the wrong thread. Exposure follows use.

Ten items is small enough that nobody needs a matrix. You need a handful of labels you can recite without looking them up, because if you can't, the person saving a file in a hurry won't apply them either.

I haven't run this exact sort myself. I'd still bet on it over another round of interviews, because it ends the day with ten labeled items you can defend, and the interviews ended with a document.

Access counts have a blind spot. They miss the data nobody opens, and in eDiscovery the data nobody opens is exactly what surfaces at the worst possible moment. That's a different problem with a different method, and it's Wednesday's piece.

So, here's the MondayMove

Pick one data store, sort it by access count, and classify only the top ten files or tables, using no more labels than you can name from memory.

Start small enough to finish.

Friday Follow-Up

The sort will miss the file that matters most.

Sorting by access count is quick, and the trouble starts when you argue over one label and realize what the sort can't see.

Submitted by Neil Watkins

November 6, 2026

MondayMove gives you one concrete action every Monday. FridayFollowUp closes the loop.

Each Friday, a short dispatch on what practitioners actually found when they ran the week's move: where they got stuck, what surprised them, and what to do next. Not sanitized case studies. Field notes. Practitioner to practitioner.

Monday asked you to pick one data store, sort it by access count, and classify only the top ten, using labels you can name from memory.

Nothing has come back from the field yet, so here's where I expect this one to break.

The sort may not exist. Plenty of stores don't keep access history by default, or keep it for a window too short to mean much, and the first hour goes to finding out which kind you have. If your store can't tell you what gets opened, write that down as a finding, because it also means nobody could tell you after an incident.

The top ten will look boring. Expect templates, a logo, a shared tracker and last year's onboarding deck. Classify them anyway, since a boring top ten is the scheme working on easy cases, and easy cases are where people learn the labels.

One of the ten may not belong in that store at all. If so, you've hit Wednesday's problem early, and moving it counts as a result even though it never got a label.

Expect a small knife fight. Two people will disagree about one file, and somebody will propose a new label to settle it. If ten files can't fit your labels, the labels are the problem, and adding one more is how the matrix got vast in the first place.

The sort will miss the file that matters most. Access counts rank what people use, and restricted data parked in the wrong place gets used by nobody. Nothing about this week's move would have found the data in Wednesday's story before opposing counsel did.

That's next week's step. Take the same store and sort it the other way, oldest last access first. Open the bottom ten and ask one question of each, whether it belongs where it sits. This pass skips the labels entirely and looks only for something parked and forgotten, and anything you find gets moved or flagged to its owner the same day.

[NEIL: closing sentence before the sign-off]

Keep going. See what a week can do.

No correct answers here. This is practitioner-to-practitioner. The more honest the responses, the more useful this gets for everyone reading on Monday morning.

See you then.

Discussion