Legal Dictation Is a Skill, and Nobody Teaches It

Dictation is a skill, and most people are never taught it

Voice is the fastest way to get words out of a lawyer’s head — roughly three times faster than typing for most people, and much faster than that when the alternative is not writing anything at all. Yet dictation has a reputation for producing unusable output, and firms that try it often abandon it within a month.

The reason is almost never the technology. It is that dictating well is a learned skill, and nobody teaches it.

What makes dictated text bad

Speaking as though thinking. Written prose has a shape. Speech wanders, restarts, and qualifies mid-sentence. Transcribed literally, that produces text that is technically accurate and unusable.

No structure declared up front. A dictation that begins in the middle produces a note nobody can file or follow.

Names and numbers unchecked. Every system mishears proper nouns, case numbers and figures. These are exactly the elements where an error matters most and where the reviewer’s eye slides past fastest.

No review step. Dictation that goes straight into the record without being read is how errors reach clients.

A method that produces usable output

Say the frame first. Matter, date, type of note, and who it is for. Ten seconds of preamble makes the result filable — and if the system is capturing a time entry, it makes the entry attributable.

Speak in whole sentences, at a normal pace. Slowing down does not improve accuracy on modern systems; speaking in complete units does. Finish a sentence before you decide it was the wrong one.

Dictate punctuation where meaning depends on it. New paragraph, colon, list items. It takes seconds and removes most of the editing.

Spell proper nouns once, the first time each appears. Client names, counsel names, case names.

Handle numbers deliberately. Say figures in a way that cannot be misheard, and check every one afterwards. Dates, amounts and case numbers are the four-alarm errors.

Read it before it counts. Every dictated note that becomes part of the record, a client communication, or a time entry gets read once by the person who dictated it. This is not optional, and it takes far less time than the dictation saved.

Where it fits best

Dictation is at its most valuable in the gap between work and record — the ten minutes after a hearing, the walk back to the office, the moment after a client call ends.

That is precisely where time entries are otherwise lost. A note dictated in that window becomes a record on the day the work happened, with a narrative specific enough to bill. The same note reconstructed on Friday becomes “attend hearing — 1.5”, which is both less accurate and less defensible.

The same applies to attendance notes. A note dictated immediately after a meeting contains detail that will not survive three days.

What to be careful about

Confidentiality in public. Dictating client detail on a train is a confidentiality problem regardless of how good the software is.

Where the audio goes. If speech is processed by a hosted service, that is client information leaving your control, and it belongs in the same analysis as any other tool — see our note on privilege and AI.

Automatic filing. A system that files a note to a matter based on what it thinks it heard should be showing you its guess, not acting on it silently.

Client-facing text. Dictation is excellent for notes and drafts. Anything going to a client or a court gets the same review as typed text, and the fact that it came out of your mouth is not a substitute for reading it.

The realistic gain

Not “write documents by speaking.” The realistic gain is that work gets recorded on the day it happened, in enough detail to bill and to remember — which, for most firms, is worth considerably more than faster drafting.

A fee earner who dictates a two-minute note after each of six matters in a day has recorded six entries with real narratives, in twelve minutes, at the moment the detail was fresh. The same person doing it from memory on Friday produces six approximations in twenty minutes, and bills less.

How Lawnova handles it

Voice dictation in Lawnova is aimed at that gap: dictate after the work, and the note arrives attached to the matter as a draft time entry with a narrative you can accept or correct. Voice intake does the same for a new enquiry.

The draft is deliberate. The system proposes; a person confirms. That is the only arrangement in which speed and the record can both be trusted.

For where this sits in the wider billing picture, see realisation is decided before the invoice is written.

One last habit worth forming

Dictate the note before you look at your phone. The window between finishing a piece of work and the next interruption is where the useful detail lives, and it closes faster than anyone expects. Firms that record time well are not more disciplined than firms that do not — they have simply made the recording step short enough to survive that window.