Time recording as the firm's core dataset and its quality problem

A senior partner closes out her week. The last entry reads: "8.0 hrs. Attention to various matters." It gets billed. The client's legal operations team flags it. Three weeks later, after emails and a partial write-down, the firm collects a fraction of that time and the relationship cools.

That single line is the whole problem in miniature. A law firm's time records are its most valuable dataset. They are also, routinely, its dirtiest.

Why time entries are the firm's core dataset

Most law firms sell time. The billable hour, despite decades of predictions about its death, still anchors how the majority of firms price work. That means the time entry is not paperwork around the revenue. It is the revenue, captured as data.

Every time entry is a small record with a predictable structure:

• Who did the work (timekeeper)
• Which matter (the client engagement)
• How long (duration)
• What they did (the narrative)
• What kind of task it was (the task code)

Multiply that by every lawyer, every day, across every matter, and you have the firm's primary operational dataset. It drives billing, profitability analysis, budgeting, staffing decisions, and increasingly, the client's own decision about whether to hire you again.

Here is the uncomfortable part. Unlike a bank's transaction data, which is generated automatically and reconciled to the penny, time entries are typed by humans who would rather be doing anything else. So the dataset that funds the entire enterprise is entered by its least motivated data-entry workforce: the lawyers themselves.

The three quality failures

Bad time data usually fails in one of three ways. The "8.0 hrs, various matters" entry manages all three at once.

1. Poor narrative quality

The narrative is the free-text description of the work. "Attention to various matters" tells the client nothing. It does not say what was reviewed, drafted, or negotiated, or why it took eight hours.

Clients, especially those with in-house legal operations teams (staff who manage outside law firm spend), reject vague narratives. Many large clients enforce outside counsel guidelines: written rules about how their matters must be staffed and billed. Vague entries violate them and get written off.

Compare:

Weak: "Review documents."

Strong: "Review and analyze defendant's third document production (approx. 400 pages) for privilege and relevance; flag 12 documents for further review."

The second entry justifies the time, survives client audit, and can be understood a year later in a fee dispute.

2. Non-contemporaneous capture

Contemporaneous means recorded at the time the work happens. The opposite is "reconstruction," rebuilding your day from memory on Friday afternoon, or worse, at month-end.

Reconstruction is where value evaporates. Research consistently suggests that lawyers who record time contemporaneously capture meaningfully more billable hours than those who reconstruct, because memory quietly discards the six-minute phone call, the quick email, the interrupted train of thought. Those fragments never make it back into the dataset. The work was done. The revenue was not.

Reconstructed entries are also lower quality. You do not remember what those "various matters" actually were, which is precisely how you end up writing "various matters."

3. Missing or wrong task coding

Modern legal billing uses standardized codes. The most common framework is UTBMS (Uniform Task-Based Management System), an open set of codes that classify legal work by task and activity. For example, in litigation, L110 covers fact investigation and development; L120 covers analysis and strategy; L330 covers depositions.

Task codes turn free text into structured, comparable data. With them, a client can ask: "How much did we spend on document review across all our matters last year?" Without them, that question is unanswerable, and the "8.0 hrs, various matters" entry cannot be coded at all because it describes no identifiable task.

You can browse the code sets at the LEDES / UTBMS site, which maintains the standards used across the industry.

The mechanics: e-billing and LEDES

When a firm bills a corporate client today, it rarely sends a paper invoice. It sends a structured data file, usually in LEDES format (Legal Electronic Data Exchange Standard). The client's e-billing system ingests it automatically.

A simplified LEDES-style line item looks like this:

INVOICE_DATE|INVOICE_NUMBER|CLIENT_ID|LAW_FIRM_MATTER_ID|
LINE_ITEM_NUMBER|EXP/FEE|LINE_ITEM_DATE|TASK_CODE|
ACTIVITY_CODE|TIMEKEEPER_ID|HOURS|RATE|LINE_ITEM_DESC

...|F|2026-03-14|L120|A104|JR-0442|2.4|550|
Analyze opposing motion to compel; outline response arguments

Notice what the machine reads: a task code (L120), an activity code (A104, "review/analyze"), hours, rate, and a narrative. If the task code is missing, the file can be rejected on upload before a human ever sees it. If the narrative is vague, an automated rule (or increasingly, an AI review layer) flags it for reduction.

This is the shift that makes data qualitydata qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.View full definition → a hard financial issue rather than a style preference. The client's software now audits your dataset before it pays. Garbage in means cash out.

🎬 [VIDEO: "How Law Firm Billing Actually Works" — youtube.com — a plain-English walkthrough of timekeeping, realization, and client billing for non-lawyers]

Fixing it at the point of entry

The cheapest place to fix data is where it is created. Everything downstream (write-downs, disputes, re-drafting narratives) is more expensive.

Capture contemporaneously. Use a timer, a phone app, or passive-capture tools that log time spent in documents, email, and calls, then prompt the lawyer to confirm and describe them. The goal is to remove the Friday reconstruction habit entirely.

Write the narrative for a skeptical reader. Assume a legal operations analyst who was not in the room will read it. Name the document, the task, the deliverable. Avoid "attention to," "various," and "re: matter."

Code at entry, not at billing. Selecting the UTBMS task code while the work is fresh is accurate. Bulk-coding hundreds of entries at month-end is guesswork.

Never block-bill when the client prohibits it. Block billing means lumping several tasks into one time entry with one duration ("Draft motion; call client; review discovery, 6.0 hrs"). Many outside counsel guidelines forbid it because it hides how time was spent. Break it into discrete entries.

Read the client's guidelines before the first entry. Different clients enforce different rules: no billing for two partners on one call, no block billing, mandatory codes, narrative minimums. The guidelines are the schemaschemaA schema is the formal blueprint that defines how data is structured, named, typed, and related within a database, file, or message.View full definition → your data must conform to.

Why this is a data-governance problem, not a nagging problem

Firms have scolded lawyers about time entries for decades with limited success. The reframe that works is treating time recording as data governancedata governanceData governance is the set of policies, roles, and processes that ensure data is accurate, secure, well-defined, and used responsibly across an organization.View full definition →: the discipline of keeping a critical dataset accurate, consistent, and usable.

That means measuring it. Firms increasingly track metrics like:

• Lag time: days between work performed and time recorded. Rising lag predicts lost hours and worse narratives.
• Narrative rejection rate: the share of entries flagged or reduced by clients.
• Realization rate: the percentage of recorded time actually collected. Poor data qualitydata qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.View full definition → shows up here directly.

When you monitor lag time by timekeeper, you find the reconstruction habit before it costs a quarter of revenue. When you monitor rejection rates by client, you learn which guidelines your firm keeps violating.

Good time data also compounds. Clean, coded, contemporaneous entries become the training set for the firm's own analytics: what a securities litigation matter actually costs, how to price a fixed fee, where a matter is running over budget. "Various matters" contributes nothing to any of that. It is not just a billing risk. It is a permanently missing row in the dataset that would otherwise make the firm smarter.

Key takeaways

• Time entries are the firm's primary dataset and its main revenue record. They are entered by its least enthusiastic data workforce, which is exactly why quality suffers.
• Three failures dominate: vague narratives, non-contemporaneous (reconstructed) capture, and missing or wrong UTBMS task codes. Block-billed "various matters" entries commit all three.
• Clients now audit your data automatically through LEDES e-billing. Missing codes and weak narratives trigger rejections and write-downs before a person reads the invoice.
• Fix data at the point of entry: capture contemporaneously, write narratives for a skeptical outside reader, code as you go, and follow the client's outside counsel guidelines.
• Treat time recording as data governance. Measure lag time, narrative rejection rate, and realization, because clean historical data is what enables pricing, budgeting, and analytics later.