Here's something most solicitors know but rarely say out loud: a significant chunk of billable time simply never gets recorded. Not because the work wasn't done. Because nobody wrote it down at the time, and by 5pm the detail is already blurring.
This isn't a discipline problem. It's a design problem. Time recording systems were built on the assumption that you'd stop what you're doing, open the billing module, and log each task as it happened. Nobody actually does that. So instead, the reconstruction happens at the end of the day, or the end of the week, from memory. And memory isn't reliable enough to be the basis of your billing.
The gap is bigger than you think
There's a fairly consistent finding across studies of legal billing: solicitors typically under-record by 15–25% of billable time. Not through any intention to shortchange the firm. Just through the natural fog of a busy day.
The problem compounds at different levels of seniority. A senior partner under-records because they're focused on the client, not the clock. A junior associate under-records because they're not sure if something is billable, or they feel awkward recording three hours for a task they think should have taken two. Both patterns quietly drain the firm's revenue.
The fix isn't nagging people to be better at admin. The fix is making it easier to capture accurate time when the work is actually happening.
What AI changes
When your research lives in the same system as your time recording, something useful happens: you have a record of what you actually did.
If you spend 45 minutes researching a planning law question on a matter, that session exists in the platform. You can see what you searched, what sources you read, how long it took. You're not reconstructing from memory, you're reviewing a log and approving a time entry. That's a meaningfully different and more accurate process.
The same logic applies to document review. If you upload a contract and work through the analysis, that activity is recorded against the matter. The time entry drafts itself from what actually happened, rather than from what you remember happening.
This isn't about AI being clever. It's about removing the gap between doing the work and recording it.
The description problem
The other place firms quietly lose money is in how time is described. "Reviewing documents" gets challenged. "Reviewing draft SPA clause 12 regarding warranty carve-outs; email to client re exposure" doesn't. The second entry took the same time to do but is much harder to dispute because it says something specific.
Vague entries invite the client to push back. Specific entries build a picture of what was actually done for them, which is both more professional and more defensible.
When a research session or a document review automatically suggests a time entry description based on what you actually did, the descriptions tend to be specific. Not because the AI is being helpful for its own sake, but because the activity log is the source material.
One thing worth changing today
If you're not already reviewing WIP, all unbilled time across all matters, before you send invoices, start doing that. It takes twenty minutes and it routinely recovers time that would otherwise have been written off without anyone consciously deciding to write it off.
Most billing systems have a WIP report. If yours does, run it before every invoice cycle. If you find entries you don't recognise, or matters with no entries when you know work was done, that's the gap. That's the revenue that wasn't captured.
AI helps close that gap over time. But the habit of reviewing WIP before billing is free and you can start it this week.
Writford connects research, document activity, and time recording in one place, so the entries exist before you have to write them. See the billing and time recording features if you want to see how it works in practice.