The Hours You Worked and Never Billed

It’s 5:30 on a Friday. A partner opens the time entry screen and tries to rebuild the week from memory, a calendar and a sent folder. Monday’s meeting is easy. So is the brief. The eleven-minute call with a client’s CFO on Tuesday, the three emails about a lease clause, the twenty minutes spent on a question about a K-1 that came in late Wednesday: those are already gone.

What does make it in gets written as “Review file and correspondence, 2.0.” Three weeks later, at pre-bill, that line gets cut to 1.0 because nobody can defend it. Or the client asks about it, and it gets cut then.

None of it feels like a lot in the moment. Added up across a firm and a year, it usually is.

I’ll say up front that I’ve never kept time in six-minute increments. I ran a digital agency for almost 18 years, which meant plenty of timesheets and plenty of hours that never made it onto one. These days I use AI to fix the places where work and money slip through an office. Time capture is one of the clearest examples I know, because the record of the work already exists. It’s just scattered across a calendar, an inbox and a document folder.

Where the time actually leaks

Follow one week of one professional’s time and you’ll find it leaking at a few predictable points.

• Late entry. The further an entry is from the work, the more falls out. Monday’s work entered on Friday is a reconstruction, not a record.
• Small items. Short calls, quick emails, a fast look at a document. Each feels too small to stop and enter. Together they can add up to hours a week.
• Vague descriptions. “Attention to matter” and “review file” invite write-downs at pre-bill and questions from clients, especially clients with billing guidelines.
• Pre-bill cleanup. Partners spending their own evenings rewriting other people’s descriptions so a bill can go out.
• Untracked scope. For accounting firms on fixed fees, the leak looks different. Work outside the engagement never gets recorded, so there’s no basis for a conversation about scope.

Most of this is recall and wording. That’s work AI handles well when it’s pointed at the professional’s own activity, with the professional deciding what’s real.

What AI-assisted time capture looks like

Draft entries from the work itself

The system reads calendar events, sent emails, call logs and document activity, and matches each to a client and matter. When the day wraps up, the professional sees a list of draft entries: the call with the CFO, the three lease emails, the K-1 question. Each draft has a proposed description and an estimated duration based on the activity.

The professional decides what’s true

Nothing is entered until the person who did the work reviews it. They edit the time, rewrite the description, merge items, or delete anything that wasn’t billable work. Five minutes before heading home replaces an hour of reconstruction on Friday.

Descriptions get checked before the bill

Drafts are written in the firm’s preferred style and checked against each client’s billing guidelines, if the client has them. Vague entries, entries without a matter number and block-billed entries get flagged while the work is still fresh, not three weeks later.

Scope gets visible

For fixed-fee engagements, the same records show time spent on work the engagement letter didn’t cover. A partner gets a short list each month to decide what to do about it.

Run the numbers

Here’s the formula in words. Billable hours per week, times the share lost to late or missing entry, times the billing rate, times working weeks per year. Then multiply by the number of timekeepers.

Say, for the sake of the math, a made-up professional bills 30 hours a week and loses 6 percent of real billable time to late or missing entries. That’s 1.8 hours a week. At a made-up rate of $300 an hour, that’s $540 a week. Over 48 working weeks, $25,920 a year, for one person.

A made-up firm with eight timekeepers like that is looking at about $207,000 a year in time that was worked and never recorded. Not all of it would be collected, so multiply by your realization rate. Then add what you currently write down on vague entries at pre-bill, which is a separate leak.

To run it for real, pull three numbers:

1. Average billable hours recorded per timekeeper per week.
2. Your actual leak rate. Here’s a test: take one person, one week, and compare what they entered against their calendar, sent email and call log. The gap is your number.
3. Your average effective rate, meaning standard rate times realization.

Picture a Monday, before and after

This is a made-up scene at a fifteen-person accounting and advisory firm.

Before: Monday morning, the managing partner sends the usual reminder that last week’s time is due. By Wednesday, four people still haven’t entered it. When they do, it’s in round numbers with thin descriptions. At month end, two partners spend an evening fixing entries so invoices can go out, and they quietly write off what they can’t explain.

After: each person got a list of drafted entries at the end of every day last week and spent a few minutes approving and fixing them. By Monday morning, the week is in. Three vague descriptions were flagged on Thursday and rewritten while the work was still fresh. The managing partner gets a short note that two fixed-fee clients ran well past the hours their engagements assumed.

Nobody’s time got padded. It got recorded.

What has to be in place first

• A time and billing system that can take drafts. Check how drafted entries get into it and who approves them before anything is built.
• Clean client and matter numbers. If matters are named five different ways, matching activity to the right one gets shaky.
• Calendars and email in business accounts. Personal accounts and texts on personal phones are blind spots, and they should stay out of the system anyway.
• Billing guidelines written down. The firm’s style for descriptions, plus each client’s rules where they exist.
• A clear word with staff about what’s captured. People should know exactly what activity the system reads and that they control every entry.

Guardrails worth keeping

• The professional approves every entry. No draft goes onto a bill unreviewed.
• Only time actually worked. The system suggests. It never rounds up, pads or invents.
• Client confidentiality comes first. Business-grade tools, data kept out of public model training, access limited, every action logged. Law and accounting firms each have their own duties around client information, and the firm decides what its setup must meet.
• Personal activity stays out. Anything not tied to a client matter gets dropped, not stored.

How you’ll know it’s working

• Billable hours recorded per timekeeper per week, before and after.
• Days from work done to time entered.
• Write-downs at pre-bill, in hours and dollars.
• Client questions or disputes about invoice entries.
• Partner time spent on pre-bill review each month.

Where I’d start

Start with one practice group or one team, and draft entries from the calendar and sent email only. That’s enough to catch most of the short calls and quick emails, it’s easy to measure against the one-week test above, and it shows people the drafts are useful before anything else gets connected.

If you want a rough figure for your own office before talking to anyone, try the AI Opportunity Calculator.