Forgot to Track Billable Hours? How to Reconstruct Your Timesheet

Work From Evidence, Not Memory

Memory is the wrong tool for this job. Asked to recall last Tuesday, you will remember the meeting that went badly, forget the ninety minutes of solid work before it, and quietly fill the gaps with what a typical Tuesday looks like. Evidence does the opposite: it hands you hard timestamps and lets you build the day around them.

Modern work leaks timestamps everywhere. Every sent email, calendar event, commit, ticket comment, and chat message records you working at a specific minute. Reconstruction is the discipline of walking those records in order of reliability and turning them into entries you can defend.

The Reconstruction Worksheet

Copy this into a note, one per blank day. Complete the evidence sweep top to bottom before you write a single entry; entries written first tend to go hunting for evidence that confirms them.

TIMESHEET RECONSTRUCTION WORKSHEET Day being reconstructed: ____________ Reconstructing on: ____________  (gap in days: ____) EVIDENCE SWEEP - work top to bottom before writing entries 1. Sent email       times, recipients, topics: 2. Calendar         meetings that actually happened, real durations: 3. Commits/tickets  IDs, timestamps, size of the change: 4. Chat history     active threads, time ranges, who with: 5. Browser/files    docs modified, save times, research trails: ANCHORS First work signal of the day: ____:____   source: ____________ Last work signal of the day:  ____:____   source: ____________ ENTRIES - one per block of work Client/project: What the evidence shows you did: Estimated time:          Rounded down to: Evidence cited: (repeat per block) SANITY CHECKS [ ] Total fits inside the span between anchors, minus breaks [ ] Every entry cites at least one piece of evidence [ ] Memory-only items rounded hard down or dropped [ ] Entries tagged as reconstructed in my own records

The Evidence Trail, in Order of Reliability

1. Sent email. The strongest source. A sent message proves you were working on that matter at that minute, and a substantive draft takes time you can estimate. Check the sent folder, not the inbox; receiving mail proves nothing.
2. Calendar. Meetings that actually happened are ready-made blocks with durations. Strike cancellations and no-shows, and check follow-up email for the ones that ran long.
3. Commits and tickets. Commits, branch pushes, ticket transitions, and comments all carry timestamps, and the diff tells you how big the work was. Read the change, not the count; a one-line fix and a refactor both show up as one commit.
4. Chat history. Threads show when you were actively engaged and with whom, and they catch the support work that never earns a ticket. Weaker, because chat runs alongside everything else.
5. Browser and file history. The gap-filler. File modification times expose deep work that sent no messages, and browser history shows the research trail. Treat it as corroboration rather than proof; an open tab is not effort.

From Evidence to Honest Estimates

Evidence gives you anchors. Estimates fill the space between them, and three rules keep those estimates honest.

Round conservatively, and round down. If the trail suggests roughly 50 minutes, bill 0.75 hours, not a full hour. The shakier the evidence, the harder you round: a block supported only by memory gets cut in half or dropped.

Give the client the benefit of the doubt. Ticket comments spanning two and a half hours do not prove continuous work. If you cannot rule out a break, assume there was one. Any error in a reconstructed timesheet should run in the client’s favor, because you created the uncertainty, not them.

Mark reconstructed entries in your own records. Tag them so future you knows which numbers were measured and which were estimated. The invoice does not need a methodology note, but if a client ever questions a line, you want to open the worksheet and show the trail.

A Worked Example: One Blank Tuesday

Dana, an operations consultant with two clients, opens her timesheet on Thursday and finds Tuesday empty. Her sweep turns up:

• Sent email: a 9:14 am reply resolving a scope question for Client A, referencing two attached documents, and an 11:22 am message summarizing decisions from a morning session.
• Calendar: a 10:00 to 11:00 working session with Client A’s ops lead. The 11:22 summary shows it ran past the hour.
• Tickets: two Client B tickets moved to review, with comments at 2:05 pm and 4:40 pm.
• Chat: a 3:10 to 3:40 pm thread walking Client B’s analyst through a vendor comparison sheet.
• File history: that comparison spreadsheet modified repeatedly between 1:50 pm and 4:35 pm.

Her entries: Client A, scope response and prep, 0.5 hours (drafting plus rereading the documents felt like 40 minutes, rounded down). Client A, working session and summary, 1.25 hours (10:00 to 11:22 by calendar and email). Client B, vendor comparison and ticket work, 2.5 hours (file saves from 1:50 to 4:35 plus the 4:40 ticket comment, trimmed because she cannot rule out gaps in that 2.8-hour span).

Total: 4.25 hours on a day she likely worked six or more. The early morning and the midday stretch left no trace, so they go unbilled. That sting is the real cost of not logging, and it is what fixes the habit.

Estimating Is Honest, Inventing Is Not

A conservative, evidence-backed estimate is ordinary professional practice. An invented line is different. Padding a reconstructed day, even to a plausible total, puts a fabrication on an invoice. That is a trust problem the first time a client compares your invoice against their own calendar, and a liability problem the day a fee dispute makes your timesheets discoverable. One line you cannot support undermines every line you can.

Preventing the Next Blank Timesheet

Reconstruction always loses hours, so the cheap fix is prevention: a work journal you fill in for two minutes at the end of each day, while you still remember it. Plain sentences are enough. That habit is what Journaled is built around: you describe your day in your own words, and it turns the description into structured entries plus a written summary you can edit before saving. The times you actually wrote down land in the entries, it never invents durations, and it exports CSV or JSON you can pull into your invoicing. It will not recover hours already lost; nothing will. It makes sure there is evidence next time.

Common Questions

Can I Bill Reconstructed Hours at All?

Yes. Estimated time is normal in professional billing, provided the estimate is honest and you can explain how you arrived at it. The problem is never reconstruction; it is invention.

How Much Should I Round Down?

To the next quarter or half hour below your estimate, scaled by evidence quality. A calendar-anchored block barely needs rounding. A memory-only block should be halved or dropped.

Should I Tell the Client the Timesheet Was Reconstructed?

Not routinely; invoices do not carry methodology. Keep the worksheet, tag the entries in your own records, and if a client questions a line, show the evidence trail openly.

What If I Find No Evidence for Hours I Know I Worked?

Bill little or none of it. If the amount is substantial, raise it with the client candidly and let them decide. What you cannot support, you cannot honestly invoice.

How Far Back Can Reconstruction Work?

The evidence survives, but your ability to interpret it decays fast. A day or two back is reliable, a few weeks is rough, and a full quarter is mostly archaeology. Reconstruct within the week, and if you write a weekly update, keep it: a dated summary of the week is exactly the evidence a blank day lacks.

What About Deep Work That Sends No Messages?

It still leaves traces: file modification times, document version history, saved drafts, and the research trail in your browser. Corroborate two weak sources before treating the block as solid.