How to clean up a messy project board

How we let the board get that bad

Nothing forces a board to stay clean, and ours proved it. Over a few months it had drifted to roughly a dozen columns, and a good chunk of the open cards had not moved in weeks. Some were finished and never marked done. Some had been quietly abandoned. A few were big ideas that had sat in "in progress" so long that nobody remembered who owned them.

None of that was laziness. Tasks get added the instant someone thinks of them, but almost nothing gets removed. We kept inventing new stages and never retired the old ones. The clearest sign was not the card count, it was the behaviour: people had started asking each other in chat where things stood, because they no longer trusted the board to tell them. Once that happens the board becomes a reporting chore you tidy before a meeting, not a working tool you run the day from. That was the point where we stopped fiddling at the edges and set aside an hour to do it properly.

The pass that actually worked

We did four things, one at a time, and the order carried most of the weight. Clearing the dead cards first meant every later decision was about work that still counted, which made the whole thing quicker and less draining.

Archive the dead cards before anything else

We pulled everything that was not really live, using a rough rule of anything untouched for about a month. We swept for orphaned tasks with no owner or date and archived them, and the ideas we were not ready to drop but not working on either went to a separate someday list. This is where we got burned the first time we ever did this. We deleted a few cards outright that turned out to still matter, and re-remembering lost work is far more expensive than the two seconds it takes to un-archive something. So now the rule is archive, never delete, and let the someday list absorb anything borderline.

Cut the columns down hard

This was the single change that did the most, and it is the one most teams skip because adding structure feels safer than removing it. We looked at each column and asked one question: does a card sitting here mean something genuinely different is happening? For most of our columns the answer was no, so we merged them and roughly halved the number of stages. The board became scannable almost immediately. Every extra column had been a tiny decision the team made on every single card, and taking those away removed a kind of friction we had stopped noticing. If you want the fuller version of this specific step, it is worth seeing what changed when we simplified the board past the cleanup itself.

Split the cards that were too big to move

A handful of cards had been parked in the same column for weeks, and when we looked closely they were not stuck, they were too big. One was basically a whole project wearing a task's clothing. A card like that cannot move because "done" means five different things, so we broke the big ones into tasks that could actually finish. Small tasks move, and a board full of moving cards is one people start believing again. While we were in there we also cleared out the label mess, because old release tags and near-duplicate colours had made the board look busier than it was. A quick pass to audit the tag library and drop anything we could not explain took only a few minutes.

Give every live card one owner and a clear next step

With only real work left, we made sure each card had a single owner. This mattered more than we expected, because the cards that had drifted longest were almost always the ones "the team" owned, which in practice meant nobody did. We renamed the vague ones so the title said what needed to happen rather than just the topic, and that small habit of renaming vague task titles, together with agreeing a simple definition of done, stopped cards from sitting at ninety percent forever. If you only take one idea from all of this, giving each task one clear owner is the one that changes how the whole board behaves.

What the cleanup actually changed

The point of the pass was never a tidier-looking board, it was a board people would trust enough to read instead of asking around. Here is the honest before and after from ours.

Within a day, people were reading the board again instead of pinging each other. We would not claim the cleanup changed how the team worked in some dramatic way, but a board people trust is a board they keep using, and an untrusted one dies quietly no matter how good the tool is.

Why our first cleanup did not last

The mistake that actually taught us something was assuming a clean board would stay clean. It did not. Within about two months ours had started drifting again, because a cleanup buys you a clean board, not a permanently clean one, and we had not changed anything about how it got messy in the first place.

The fix turned out to be small and boring, which is probably why it works. One person now owns a ten minute weekly sweep: archive what finished that week, reassign anything that has drifted, and spot where one person is quietly overloaded before it becomes a stalled card. Making it one person's job rather than a shared good intention is the part that made it stick. The sweep is the real fix. The big cleanup is only ever catching up on sweeps you skipped.

In practice the sweep catches the same three things every week, which is what stops them from piling up into another mess. There are the cards that quietly finished but never got moved, which is most of it. There are the one or two that have not moved in a week, which is the early warning that something is either blocked or too big and needs splitting before it stalls for a month. And there is the occasional column or label that crept back in, which gets merged away on the spot. None of it is dramatic, and that is the point. Ten minutes of small corrections is a lot cheaper than an hour of rescue, and the board never gets far enough from reality for people to stop trusting it.

When a full cleanup is worth it, and when it is not

Match the effort to the state of the board, because the full pass is overkill for a board that is basically fine and not enough for a board with a deeper problem.

Run the full pass when

The board has not been touched in months, you inherited it from someone who left, or the "in progress" column has become a permanent parking lot. That is a real cleanup, so block the hour and do all four steps in order rather than picking at it.

Just do the weekly sweep when

The board is healthy and only lightly cluttered. It does not need surgery every quarter, and ten minutes a week keeps it clean for a fraction of the effort a full pass costs.

Do not bother cleaning when the problem is structure

If tidying one board only pushes the mess somewhere else, or you are running several boards that share the same tasks and owners, the issue is not tidiness. No amount of archiving fixes the wrong work living in the wrong place, and that is a decision about how your projects are set up, not a cleanup.

Clean once, then sweep weekly

Our messy board came back to life in one focused pass: archive the dead cards, cut the columns, split the oversized ones, and give every live task one owner and a clear next step. The order is what made it fast, and cutting columns is what made it readable.

But the pass is only half of it. The thing we wish we had done from the start is put a recurring ten minute sweep on one person's calendar, so the board never drifts far enough to need rescuing again. Clean it once this week, then sweep it weekly, and you will not be back here in three months.