An inbox and an archive
The moment a system fails is small and easy to miss: you have the receipt in your hand, and you pause. Where does it go? That pause is half a second long and it is the whole failure, because a capture step with a decision in it doesn’t happen reliably.
The fix is structural rather than motivational. You need two containers with two completely different jobs, and most people have one container trying to do both.
Why one container can’t work
A landing zone has to accept anything instantly. Its requirement is that putting something in it costs nothing and requires no thought. Mixed contents, no order, no naming standard, no decisions.
A permanent home has to be orderly. Its requirement is the opposite: consistent naming, no duplicates, nothing ambiguous, nothing half-processed.
Ask one folder — or one drawer, or one photo album — to satisfy both and it satisfies neither. If it
is orderly, capture is slow, so capture stops. If capture is fast, it fills with IMG_4471 and
becomes something you avoid opening, which means nothing is ever findable, which means the whole
exercise was pointless.
There is a third failure that comes from the same root: when there is only one container, you can never tell what has been dealt with. A folder with four hundred things in it, some named and reconciled and some photographed at a till two days ago, has no answer to “what’s outstanding”. So you re-check things, or you stop checking.
What each container is for
The inbox is a holding pen with one rule: nothing lives here. Anything can be dropped in, in any state, with any name. It is expected to be a mess. The single discipline it needs is that it gets emptied — not sorted, not tidied, emptied.
You want exactly one of these per medium. One digital folder. One physical envelope or tray. If there are two, capture has a decision in it again.
The archive is the permanent home, and it is write-once. Things arrive here already named, already annotated, already deduplicated. Once something is in, it isn’t touched again. Crucially, the archive is searched, not browsed — you should almost never open it to look around, which means it does not need to be pleasant to navigate. It needs to be consistent, which is a much lower bar. The naming that makes search work is the subject of finding it again.
The move between them
This is the only ritual in the system, and it is worth defining precisely because a vague one doesn’t get performed.
It happens on a schedule, not on demand. Weekly is comfortable. Monthly works if capture is carrying the annotation properly, which is the point of why batch entry fails: processing can be batched safely, capture can’t.
It is a single pass with a fixed set of actions per item. Name it. Check it has a reason attached. Decide whether the paper original is still needed. Move it. Next. No re-litigating, no reorganising the archive, no starting a new folder structure.
It ends with the inbox empty. That’s the definition of done, and it’s what makes the state of things legible at a glance. An empty inbox means everything is filed. Nine things in the inbox means nine things outstanding. No inventory required.
Anything that can’t be resolved gets its own small holding place. The unidentifiable slip, the
amount that doesn’t match, the thing that raises a question. Two or three of these will accumulate
per pass and they must not be allowed to block the pass — a pending container next to the inbox
keeps them visible without stalling everything behind them.
Keep or bin
KEEP OR BIN — the two containers
· One inbox per medium, deliberately
unsorted
→ KEEP. Its messiness is the
feature.
· Named, annotated file in the archive
→ KEEP. Finished work.
· A folder tree you designed but don't
file into
→ BIN. Unmaintained structure is
worse than none — it implies an
order that isn't there.
· Your email inbox as the archive
→ NOT AN ARCHIVE. It's someone
else's container, it's shared
with everything else, and search
there matches text you didn't
write.
· A second copy left in the inbox after
filing "just in case"
→ BIN. Two copies in two states
is how you lose the ability to
tell which is current.
· Whether the paper original can go once
the file is archived
→ ASK LOCALLY. Acceptable formats
and originals rules vary by
jurisdiction.
The physical version
Everything above works on paper and is arguably easier there, because a tray is visibly empty or visibly not.
A tray by the door is the inbox. A labelled envelope per month in a drawer is the archive — see where paper survives for why the drawer matters more than the labelling. The weekly pass takes the tray, annotates anything unannotated while it’s still recent, and moves it into the current month’s envelope.
The two-container logic is medium-independent, which is the useful thing about it. If you run paper and digital in parallel, you have two inboxes and two archives, and the pass handles both in the same sitting. Which of the two archives is the one you actually rely on is the question in paper, digital, or both.
Where it still breaks
The inbox stops being emptied. Then it is a shoebox with a better name. The failure is recoverable and the recovery is a triage, not a full pass: newest first, because those still have recoverable reasons.
The archive gets reorganised. Someone decides the structure could be better and starts moving things. Half-migrated is the worst possible state. If you must restructure, finish it in one sitting or don’t start.
Capture starts targeting the archive directly because a particular item felt easy to file properly. Reasonable, and it puts the decision back into capture. Everything goes through the inbox, including the easy ones.
What this doesn’t settle
Whether any container satisfies a requirement placed on you: whether digital copies suffice, whether originals must be retained, how long the archive must survive, or whether a particular storage arrangement is acceptable. Those vary by jurisdiction and record type and belong to your tax authority or adviser — how long to keep things covers how to build a scheme without knowing the numbers.
What the split does settle is the half-second pause, which is not a rules problem at all. It is the reason the receipt in your hand ends up on the passenger seat.