About 60 hours a month back from chasing client documents.

Month-end close is the whole business at a bookkeeping practice. Revenue recognition, billing, and the owner's month all hang off a single list of documents that was being reconstructed by hand every morning. It was not that anyone was failing at their job. It was that the firm's most important state was living in two people's heads and a shared spreadsheet that had drifted out of date.

The practice had been profitable for years, and the owner knew exactly why it worked: nobody was ever idle, and clients were answered quickly because the owner personally chased them. But the system had no floor under it. When one of the two senior staff was out, the close slipped by days and billing went late, and nobody could say how many documents were outstanding on any given day. They could only say it felt like a lot.

The owner described the first hour of the morning as decision time rather than admin time. That distinction turned out to be the whole project. The chasing was mechanical; deciding which client mattered this week, which conversation needed a phone call instead of an email, and which delay was about to become a billing problem was not.

We started by reading 90 days of shared inbox traffic against the practice's existing tracking sheet and rebuilding the open-item list as structured data. This was the unglamorous part and it mattered most, because it turned what was missing from a recollection into a query. Once the list was real, almost everything downstream became straightforward.

Each morning the agent sorts open items by what is actually blocking the close, drafts a chase in the owner's voice that names the specific missing document and its date, and queues them for review. As clients respond, it files each document to the right client folder, updates the open-item state, and publishes a close-readiness view that the two staff now work from instead of the spreadsheet.

One rule mattered more than the rest: nothing was allowed to send on its own for the first two weeks. The owner reviewed every draft and rejected roughly half of them. That correction phase is where the real rules turned out to live. The office had a voice, a threshold for when a client should be called instead of emailed, and a set of clients who should never get an automated nudge at all. None of that was written down anywhere. It came out of watching the drafts fail.

The 60 hours figure is real, but it understates the change. The practice did not mainly save time chasing documents; it gained a reliable answer to a question it previously could not answer at all. On any given morning the owner can now say exactly how many clients are missing what, what each missing item is worth, and which delay is becoming a billing problem.

That answer is what stopped the close from depending on who was in the office. The two staff who used to hold the client history in their heads now work from the same view, which means a vacation no longer changes the close date. The owner kept the part of the morning worth keeping and handed the rest to a process that does not get tired.

The one thing we would change is the ordering of the second workflow. Month-end billing packet assembly was added in week six and should have shipped in week three; the practice was still assembling packets by hand while the agent was already deep enough in the document flow to do it.

We also left the one-off explanation in. The owner's own framing — that this was about protecting the decision-making part of the morning, not about saving keystrokes — is the reason the staff adopted it. A workflow that only saves time gets abandoned the first week something goes wrong. One that visibly protects judgment gets defended.