Does automatic time tracking actually work, or is it a marketing promise?

It works, with caveats. Passive capture reliably detects events (a call happened, a document was open) but cannot read your intent. A 14-minute call from your mother looks identical to a 14-minute call from a client until you teach the tool who's who. Well-designed tools turn this into a 2-minute daily review rather than letting the timer run wild.

How accurate is passive capture compared to a disciplined timer?

A disciplined lawyer with a timer captures ~70% of billable moments — good enough for steady-state weeks, bad on chaotic ones. Passive capture catches 95%+ of events, trades accuracy for recall (some events are ambiguous and need review), and crucially does not degrade on the weeks when the lawyer is too busy to remember the timer. The failure modes are different.

What exactly does ClaimHour capture?

Call metadata (duration, counterparty phone number, direction — inbound or outbound), email headers (sender, recipient, subject, timestamp, no body), and document edit sessions (file name, matter tag, open and close time). No audio, no email content, no document contents. Everything stays local until you export approved entries to your billing tool.

Will a bar ethics opinion accept passive capture?

Bar ethics opinions on time tracking generally require that entries accurately reflect time actually spent and that the lawyer reviews and attests to them. Passive capture plus a mandatory daily review — which is ClaimHour's model — meets that standard. The tool proposes; the lawyer disposes. We've mapped this alignment to ABA Model Rules 1.5(a) (reasonable fees) and 5.3 (non-lawyer assistance) in our public FAQ.

Buyer's education · April 2026

Automatic time tracking for attorneys: how passive capture actually works

A technical buyer's guide for lawyers considering passive time capture — what it watches, what it misses, how accurate it really is, and the two specific privilege concerns worth thinking about before you install anything.

TL;DR

Automatic (passive) time tracking uses OS-level signals — call metadata, email headers, document open sessions — to detect billable-looking moments without you starting a timer. Done well (ClaimHour, Smokeball AutoTime, Clio Duo), it recovers the 5–10 hours/week most lawyers leave on the timesheet. Done badly, it over-captures and puts privilege-sensitive data on a server you don't control. The difference is what the tool stores, not how clever its AI is.

What "automatic" actually means in this category

Three things can happen passively inside your workflow: a phone call, an email, or a document edit. Each leaves a system-level trace — call log entry, email header, file modified timestamp — that an app can read without reading the content. Passive time tracking for attorneys is just plumbing those three signals into a classifier that guesses "this looks billable" and drops a draft time entry into a review queue.

The sophistication is in the guessing, not the plumbing. A well-trained classifier uses the counterparty's phone number (from your matter's contact list), email domain (from recent matter correspondence), and filename pattern ("Smith_motion_v3.docx" → Smith matter) to auto-assign the event to a client. A naive classifier surfaces every phone call you made and asks you to sort them all. The difference between "magic" and "annoying" is how fast the classifier learns your patterns.

What gets captured, explicitly

Calls

Metadata only: counterparty phone number, call duration, direction (in/out), timestamp. On iOS, ClaimHour reads this from the CallKit log. On macOS, it reads from the Contacts call history when your phone is paired. No audio, no transcription.

Emails

Headers only: sender/recipient, subject line, send time, thread depth. Subject-line keywords are used for matter classification (the word "Smith" in the subject line suggests the Smith matter), but email bodies are not stored, transmitted, or indexed. No message content.

Documents

Edit sessions: filename, parent folder, open time, close time, active-vs-idle seconds. This is how Smokeball AutoTime has worked for years, and how ClaimHour works for Pages/Word/PDF documents on macOS. No document contents.

What's deliberately not captured

Audio or transcripts of any call
Email bodies, attachments, or inline images
Document contents, keystrokes, or diffs
Screenshots or screen recordings
Browser history, unless you opt into a per-matter research timer

This set is the technical definition of a "metadata-only" time tracker. If a vendor's privacy policy reserves the right to read email bodies or stream document contents, it's a different category and a different conversation with your bar association.

Accuracy: passive capture vs. the stopwatch

Two failure modes to separate. Precision is "of the events captured, how many are actually billable." Recall is "of all billable events, how many got captured." Passive tools and timers trade these differently.

Method	Recall (catches)	Precision (correctly billable)	Weekly effort
Paper timesheet	30–50%	High (you filter as you write)	3–5 hours reconstruction
Manual timer (Toggl)	60–75% when disciplined; ~40% in busy weeks	Very high	15–20 min/day clicking
Passive (no review)	95%+	50–70% (personal calls, spam emails included)	0 min — but invoices get messy
Passive + daily review	95%+	95%+ after review	2 min/day review

The "passive + daily review" row is the model ClaimHour is built around. The tool catches everything; a 5pm 2-minute digest lets you dismiss the grocery-store call to your spouse and keep the one to opposing counsel. Precision is restored; recall was never lost.

The two privilege questions worth asking

1. Where does content live?

If the tool needs to read email bodies or record calls to classify them (Billables.ai does, Clio Duo does for Gmail-via-Clio), then your client communications are traveling to a vendor server and being stored in some form, even if only transiently. That's a privilege consideration, a HIPAA consideration if any client has medical records, and a data-breach consideration if the vendor ever gets hacked. ClaimHour's metadata-only model sidesteps all three because there is no content to expose.

2. Who has the decryption key?

Even a metadata-only tool can be privilege-relevant if its database is backed up to a vendor-managed server with a vendor-held key. ClaimHour stores the local database locally (in your Library folder) and exports are pull-based from your machine — we don't hold a key to your data. This is the same privilege profile as a paper timesheet stored in your office safe.

How to evaluate a passive tracker in 15 minutes

Open the privacy policy. Search for the words "email body," "audio," "transcript," "stored on our servers," and "decryption." The tool either promises never to touch those, or it doesn't.
Ask about the review model. Is there a mandatory daily or weekly review? Or does the tool auto-push entries into your billing system without an approval step? Push-through without review is a malpractice risk dressed as a time-saver.
Check export formats. If the tool only exports into its own invoicing engine, you've bought a PMS, not a time tracker. ClaimHour exports to QuickBooks IIF, LawPay, FreshBooks, and CSV.
Ask where data is stored. Local database + pull-based export = no vendor key. Cloud-synced database = vendor key. Both can be defensible, but they're different positions to have to defend.

How ClaimHour fits the category

ClaimHour is the passive capture tool designed specifically around the two privilege questions above. Metadata-only capture, local-first storage, mandatory 2-minute review, native export to the billing tools solos already use. $29–$99/mo standalone — no PMS subscription required. If this matches what you were hoping the category would give you, join the waitlist. Early access opens summer 2026 and first 100 signups get 50% off Pro for life.

Get early access