Every firm that has ever bought practice management software has a story about the workflow board that was perfect for six weeks. Then it drifted. Then it was wrong often enough that people stopped trusting it. Then it was a spreadsheet again.
The instinct at that point is to blame discipline — the board would work if people just kept it current. We have watched enough of these die to think the opposite: a board that requires unusual discipline is a badly designed board. Here is what actually causes the drift.
Cause 1: statuses that do not belong to the work
A single global status list has to serve every kind of work the firm does. So it grows: Not Started, In Progress, Waiting on Client, Waiting on Information, Ready for Review, In Review, Ready to File, Filed, Waiting on Signature, Complete, On Hold, Blocked. Twelve statuses.
Now a bookkeeper opening a Monthly Bookkeeping item sees twelve options, of which four apply. "Ready to File" is meaningless for a bookkeeping close. "Waiting on Signature" is meaningless for a payroll reconciliation. Every update becomes a small act of translation, and translation is where accuracy goes to die — different people translate differently, and within a month the same status means three things.
The fix is to attach statuses to work types. An Individual Tax Return has its own status list. Monthly Bookkeeping has a different one. When someone picks a work type, the status dropdown narrows to the statuses that can actually apply. Twelve options become four, all of them meaningful, and the translation step disappears.
The test
Ask a staff member to explain the difference between two adjacent statuses on your current board. If they hesitate, or if two people give you different answers, that pair is the reason your data is unreliable.
Cause 2: updating costs more than it returns
Consider the cost of one status change on a badly built board: filter to find the item, open the record on a new page, change the field, save, get returned to an unfiltered list at the top, re-apply your filters, find your place again. Fifteen seconds of work wrapped in ninety seconds of navigation, thirty times a day.
Staff are not lazy about this; they are correctly economizing. If updating costs a minute and a half and the only beneficiary is a report a partner reads on Friday, batching updates to Friday morning is rational behavior. And a board updated once a week is a board that is wrong for six days out of seven.
What actually removes the cost:
- Change status from the list. A modal from the row, not a page navigation.
- Preserve the view. Opening an item and coming back should return you to the same filtered, sorted, scrolled position — including after a status change.
- Filter state in the URL. So a refresh, a back button or a shared link all land in the same place.
- Saved views. "My open items due this week" should be one click, permanently, not a filter combination rebuilt every morning.
Cause 3: the board shows finished work
A tracking board accumulates. By April a firm's board holds every return filed that season, and the twelve items that still need attention are buried among four hundred that do not.
Manual archiving does not solve this, because manual archiving is another task with the same cost-benefit problem as manual status updates. Automatic archiving does: complete an item, it sinks to the bottom, and a background job removes it from the working list a week later. A week, not immediately, because a mis-clicked "complete" needs to be recoverable without an administrator.
Archived is not deleted. It should sit behind a "show archived" toggle, and it should still be searchable — the history is genuinely useful, it just should not be in the way.
Cause 4: nobody owns it, or everybody does
Two failure modes, opposite in shape. If every item is assigned to the firm rather than a person, the board is a queue nobody feels responsible for. If every item is assigned to whoever created it, the board reflects who opened the work rather than who is doing it.
Default assignment rules per work type fix most of this without anyone thinking about it: new Monthly Bookkeeping work for this client routes to this bookkeeper, new Individual Returns route to the preparer who handled last year's. Set once, applies forever, and the exceptions are handled by reassignment — which people are happy to do, because it takes one click and removes work from their own list.
Cause 5: it does not tell you anything you did not know
The last one is the quietest. A board that only reports what people typed into it is a data entry tax. A board that tells you something is worth maintaining.
"Something" is usually one of: what is overdue, what is due this week, what has been sitting in the same status for three weeks, which client accounts for a disproportionate share of open work, and whether the firm is completing work faster than it is taking it on. None of those require a single extra keystroke from staff — they are all derivable from data the board already holds. But they have to actually be shown, and shown to the people doing the updating, not only to the partner.
The reciprocity rule
Anyone asked to keep a board current should be able to see something useful in it about their own work. A board that only feeds upward will be maintained grudgingly, then badly, then not at all.
A practical setup
- Define work types before statuses. Individual Tax Return, Business Tax Return, Monthly Bookkeeping, Sales Tax, Payroll, Advisory — whatever matches how your firm actually divides work.
- Give each type four to six statuses. If you need more than six, you probably have two work types wearing one name.
- Color the statuses that mean "someone must act". Waiting-on-client is not a problem; waiting-on-client for three weeks is.
- Set default assignments per type. Then let people reassign freely.
- Turn on auto-archive at a week after completion.
- Give every person a saved view of their own open work, and put it on their landing screen.
Do those six things and the board stays current without anyone being asked to be more disciplined — which is fortunate, because asking has never worked.
See how Kveldo's WIP board works →
Read next: the four views a firm needs by February 1 →