Reducing Write-Downs and Write-Offs Before They Happen

Every firm I have watched try to fix its write-down problem starts in the same place, which is the pre-bill. A partner blocks out a Friday afternoon, opens the draft invoices, and goes line by line with a red pen, and by the end of it a meaningful percentage of the month's recorded value has quietly disappeared. The firm calls this billing hygiene. What it actually is, in almost every case, is a very expensive person spending unbillable hours reading the evidence of a problem that happened weeks earlier and can no longer be repaired.

That is the uncomfortable structural fact about write-downs. By the time an hour reaches the pre-bill, everything that determines whether it survives has already been decided. The narrative is already written or already vague. The work was already inside scope or already outside it. Three people were already in that meeting or they were not. The entry was already contemporaneous or already reconstructed from a calendar four days later. The reviewing partner is not making a decision so much as reading a verdict that was rendered at the moment of entry, and the only lever they have left is subtraction.

So this post is not about reviewing invoices better. It is about the four specific places realisation actually leaks, what each one looks like from the inside, and what has to be true at the moment an attorney types an entry for that entry to arrive at the pre-bill in a condition nobody feels the need to cut. Firms that get this right do not have better pre-bill review. They have shorter pre-bill review, because there is far less left to argue about.

A write-down and a write-off are two entirely different admissions

These two words get used interchangeably in most firms and they should not be, because they describe failures at completely different points and they carry completely different information. A write-down happens before the invoice leaves the building. It is the firm looking at its own recorded value and deciding, internally and privately, that some portion of it is not defensible enough to send. Nobody outside the firm ever learns it happened. There is no client conversation, no dispute, no negotiation. The value simply evaporates between the timesheet and the invoice, and unless somebody deliberately records it, it evaporates without leaving a trace anyone can study later.

A write-off happens after the invoice has gone out. The client has now seen the number, has pushed back on it or simply declined to pay it, and the firm has decided that pursuing the balance costs more in relationship damage or collection effort than the money is worth. That is a fundamentally different kind of failure, because the client was involved. A write-off tells you your billing did not survive contact with the person paying it. A write-down tells you your billing did not survive contact with your own partner. Both reduce realisation. Only one of them is a client problem, and firms that lump them together end up trying to solve a client relationship issue when what they actually have is an internal quality control issue, or the reverse.

The pre-bill review is a symptom, not a control

Firms treat the pre-bill as a quality control step, and in a narrow sense it is one, because bad entries do get caught there. But a control that only ever subtracts is not really a control. It cannot add back the specificity a vague entry is missing. It cannot retroactively make an unauthorised task authorised. It cannot un-duplicate three attendees on a call that only needed one. Its entire vocabulary consists of cutting, and cutting is what you do when prevention has already failed.

There is also a hidden cost that almost nobody accounts for. The people doing pre-bill review are usually the highest-billing people in the firm, and the review itself is unbillable. A firm that spends six partner hours a month reading drafts and trimming them is spending real money to reduce its own revenue, which is a genuinely strange thing to build into a process on purpose. The goal should not be a better pre-bill. The goal should be a pre-bill so uneventful that it takes twenty minutes and mostly consists of confirming that things look correct, because the four leaks below were closed upstream where they actually live.

Leak one: narratives that describe activity instead of work

The single largest category of write-down in most firms is the entry that says something happened without saying what. "Attention to file." "Review of correspondence." "Telephone attendance." "Consideration of documents." These are not lies and they are not padding. They are honest descriptions written by someone in a hurry, and they are indefensible for exactly one reason, which is that nobody reading them can tell whether the time recorded is proportionate to the work performed. A partner who cannot tell trims. A client who cannot tell queries. An insurance panel or corporate legal department running line-item review against outside counsel guidelines rejects outright.

What makes this leak so costly is that it is completely invisible as a cause. The firm sees a write-down and attributes it to inefficiency, or to a difficult client, or to an associate taking too long on a task. The real cause was a sentence fragment. The same 1.4 hours, described as review of the third-party disclosure list against the pleaded case with a note of three items requiring follow-up, survives every one of those reviews intact. The work was identical. The record of it was not, and the record is the only thing anybody downstream can actually assess.

What a defensible narrative actually contains

A narrative that survives review does four things, and they are all mechanical rather than stylistic. It names the specific document, party or issue involved rather than a category. It says what was done to that thing, meaning drafted, reviewed against, revised, advised on, and not the passive constructions that hide the verb. It states the output or the decision that resulted, because an entry that ends in a result is far harder to argue is unproductive. And it stays as a single task, because block billing that fuses four activities into one 3.2 hour line gives the reviewer no way to test any individual component and so invites a cut to the whole block.

The practical implication is that narrative quality is a system property, not a discipline property. If your time entry form is a single free-text box with no matter context visible, no prompt about what task type this is, and no view of what the last entries on this matter looked like, you will get vague narratives no matter how many times the practice is reminded. Casely ties every entry to the matter from the moment it is created and converts unbilled time into an itemised draft in one click, which matters here for a reason that is easy to miss: when an attorney can see the itemised draft their entries will actually produce, the writing changes. People describe work differently when they know they are writing the invoice rather than filling in a timesheet.

Leak two: work performed outside the agreed scope

The second leak is work that was genuinely done, genuinely useful, genuinely well recorded, and genuinely not something the client ever agreed to pay for. This is the write-down that feels most unfair to the person who did the work, because there is nothing wrong with the entry at all. The problem sits at the boundary of the engagement, and it usually happened gradually. A client asks one adjacent question. Then another. Six weeks later a third of the recorded time on the matter relates to a workstream that appears nowhere in the engagement letter, and the partner reviewing the pre-bill knows perfectly well that sending it will trigger a fight.

The mechanism that closes this leak is making scope visible at the point of entry rather than at the point of billing. If an attorney can see, while logging time, that the matter is at a defined stage and that this task belongs to a workstream the client has authorised, the out-of-scope work either does not get done or it gets escalated for a scope variation before the hours accumulate. Casely's matter stage tracker is a clickable stepper that is configurable per firm and per practice area precisely so that stage means something specific to how your matters actually run, and a matter sitting at a stage that does not contemplate the work being logged against it is a signal worth acting on in the same week rather than at month end. Where the extra work is legitimately a separate engagement, connecting it as a related matter with the reason stated keeps the original matter's economics clean instead of burying a second job inside the first one's numbers.

Leak three: duplicated staffing that nobody authorised

Three fee earners attend a client call that one could have handled. Two attorneys separately research the same point because neither knew the other was on it. A senior associate redrafts a document a junior has already drafted, from scratch, rather than marking it up. Each of these produces recorded time that is entirely real and entirely unbillable in the client's eyes, and the partner reviewing the pre-bill will cut the duplicate every time, because sending a client three sets of hours for one meeting is how you lose the client.

Duplicated staffing is a coordination failure, not a billing failure, and it is worth being honest about which of the three patterns above is happening in your firm before you try to fix it. Genuine over-attendance at meetings is usually cultural and gets fixed by someone saying out loud who is required and who is optional. Duplicate research is an information problem, and it is solved by everyone working a matter being able to see, without asking, what has already been done and by whom. Redrafting from scratch is often a supervision or trust problem wearing a workflow costume. The write-down looks the same in all three cases, which is exactly why a firm that only sees the write-down and not the cause keeps solving the wrong one.

Leak four: late entries and the credibility discount

An entry written four days after the work carries a discount that has nothing to do with its accuracy. Even when the reconstructed figure is exactly right, it reads differently, because reconstructed entries cluster on round numbers, arrive in batches with identical timestamps, and describe work in the vague register that memory produces. A reviewing partner cannot articulate why the batch of eleven entries all logged at 6pm on Friday for work spread across the week feels soft, but they trim it anyway, and a sophisticated client reviewing an invoice with the same pattern reaches the same instinct faster.

There is a second and larger cost, which is that late entry and vague narrative are the same failure viewed at different distances. Nobody writes a precise narrative about Tuesday on Friday, because the precision is genuinely gone. So lateness does not just create its own write-down risk, it manufactures leak one at scale. Fixing entry timing is therefore the highest-leverage change available, because it is the only one that closes two leaks with a single behaviour, and it is also the one most dependent on the tool being reachable at the moment work ends rather than only from a desk. Casely is cloud-native with nothing to install locally, so an entry logged from a phone in a corridor after a hearing is the same first-class entry as one typed at a desk, which is the entire point. The window in which an entry is accurate is measured in minutes, and a tool that is not available inside that window is not a capture system.

Every reduction needs a reason attached to it, permanently

Here is the practice that separates firms that improve from firms that simply repeat the same month forever. When time is reduced or removed from a draft invoice, the reduction has to carry a stated reason, and the record of it has to survive. Not a note in someone's head. Not a deletion. A recorded reason, attached to the entry, readable six months later by someone who was not in the room.

This is the same discipline Casely applies elsewhere for exactly the same reason. Every document carries a comment field recording what changed and why, so the history of a file is legible rather than inferred. In trust accounting, corrections are voided and stay visible rather than being deleted, because a ledger you can quietly erase is not a ledger anyone can audit. Billing reductions deserve identical treatment. The moment a write-down becomes a deletion, the firm loses its only opportunity to learn that the same timekeeper, the same task type, or the same client keeps generating the same cut, month after month, which is precisely the pattern that should be driving a training conversation, a staffing change, or a repricing.

Turning write-down reasons into the only report that changes behaviour

Once reasons are recorded, the report you want is not total write-down value. Total value tells you the size of the wound and nothing about its cause. The report that changes behaviour is write-down value split by reason, then by timekeeper, then by matter type, looked at every month by the person who can actually act on it. When you can see that sixty percent of last quarter's reductions were narrative quality and that they concentrate in two people, you have a training problem with a name and a fix that takes an afternoon. When you can see that most of it is out-of-scope work on one client, you have a scoping conversation to have with that client, and you can have it with evidence rather than a feeling.

This is also where the shape of your billing arrangements starts to matter. A practice area that keeps producing scope write-downs on hourly matters is often telling you it should be priced differently, and moving it to a flat fee removes the entire category of dispute rather than managing it. Casely supports hourly, flat-fee, contingency and blended billing natively rather than treating anything other than hourly as an exception, and it exports LEDES 1998B for the institutional clients whose e-billing systems will reject block billing and vague task descriptions automatically, without a human ever forming an opinion about them. Those clients are, in a sense, doing you a favour. They enforce at the invoice what a well-run firm enforces at the entry.

The one thing a reviewing partner should still be doing

None of this argues for abolishing pre-bill review. It argues for changing what it is for. A reviewer who has to repair narratives and hunt for duplicate staffing is doing data cleanup. A reviewer looking at entries that are already specific, already in scope, already contemporaneous and already free of duplication is doing the thing only a partner can do, which is exercising commercial judgment about this particular client, this particular relationship and this particular moment in the matter. That judgment is legitimate and sometimes it will still produce a reduction, and that is fine. A deliberate commercial decision recorded with a reason is a different object entirely from a defensive cut made because nobody could tell what an entry meant.

It is worth adding the obvious caution here, because this is a legal-adjacent question rather than purely an operational one. What counts as a reasonable fee, what a bill must disclose, and how billing records must be maintained are all governed locally and they genuinely differ. In the United States that runs through each state's adoption of the professional conduct rules on fees. In England and Wales it runs through the regulator's transparency and costs rules and, in contentious work, the costs assessment process. Canada handles it provincially and Australia through its legal profession legislation, and none of these map neatly onto each other. Treat everything above as billing operations guidance, and confirm the disclosure and record-keeping specifics with your own regulator before changing what your invoices contain.

Where this actually leaves you

The reframe worth taking away is that realisation is decided at the keyboard, not at the pre-bill. Every one of the four leaks here is closed upstream or not at all. Narratives get specific at the moment of typing or they never do. Scope gets caught while the work is being considered or it gets caught after the hours are spent. Duplication gets prevented by visibility into the matter or it gets discovered in a draft invoice when it is far too late to undo. And late entries manufacture every other problem on the list, which makes entry timing the first thing to fix rather than the last.

That is why the system you record time in matters more than the rigour of the person reviewing invoices. If entries are tied to the matter from creation, if stage and scope are visible while the work is being logged, if the draft invoice is one click away so people write knowing what the client will read, and if every reduction leaves a permanent reason behind it, the write-down rate falls for structural reasons rather than through anyone trying harder. If you want the mechanics of the capture side, legal time tracking software covers how entries get created and tied to matters, and legal billing software covers what happens from unbilled time through to the itemised invoice and LEDES export.

Start with one month of honest measurement before you change anything else. Record a reason on every single reduction, no exceptions, and at the end of the month sort them by cause. Most firms are genuinely surprised by what comes back, because the story they have been telling themselves about difficult clients and inefficient associates usually turns out to be a narrative quality problem and a scoping problem wearing a disguise. You cannot fix what you have been deleting. Casely is free to start on, so there is no reason the measurement has to wait for a budget cycle.