UTBMS and ABA Billing Codes: A Practical Guide for Law Firms

Two lawyers can perform nearly identical work and still need to code it differently. One client may accept the standard litigation task and activity codes. Another may allow only a subset, use a revised set, add custom values, or require a particular combination for that matter.

That is the practical challenge behind what people often call “ABA billing codes.” Selecting a code is not simply matching a narrative to a universal list. It is a client-specific decision: identify the actual work, use the code catalog authorized for the matter, follow the applicable billing guideline, and have a person approve the result.

This guide explains that operating model. It is not a replacement for the client’s engagement terms, outside counsel guidelines, or authoritative code definitions.

What UTBMS means

UTBMS stands for the Uniform Task-Based Management System. The ABA’s UTBMS toolkit describes a family of code sets used to organize legal work for budgeting and billing. It includes sets for litigation, bankruptcy, counseling, projects, workers’ compensation, mergers and acquisitions, criminal law, and real estate. The LEDES Oversight Committee maintains additional sets.

“ABA billing codes” is useful search language, but it is imprecise. The history published by the ABA explains that the original litigation initiative was a joint effort involving the ABA Section of Litigation, the American Corporate Counsel Association, corporate law departments, law firms, and Price Waterhouse. The ABA describes the code sets as a methodology that can be modified for the engagement, not a rule that every client must adopt unchanged.

The existence of a code therefore does not make it required. The client’s instructions and the matter’s configuration determine whether UTBMS applies and which values are acceptable.

Task, phase, activity, and expense codes

The terms describe different dimensions of the record:

• A phase is a broad grouping of related legal work.
• A task code describes what subject or stage the work concerned.
• An activity code describes what the professional did.
• An expense code categorizes a cost rather than a timekeeper’s work.

For one source-verified example, the ABA Litigation Code Set places dispositive motions under task L240 and identifies drafting or revising as activity A103. Drafting a summary-judgment motion might therefore suggest that pairing under the original ABA set. But the example is not a universal answer. The client may use another set, restrict a value, require a custom mapping, or classify the work differently under its guidelines.

The narrative still matters. A task and activity pair categorizes an entry; it does not explain the work’s purpose, subject, or result. The code, narrative, matter, and underlying work record should corroborate one another.

UTBMS is not LEDES

UTBMS and LEDES are related, but they are not synonyms.

UTBMS supplies classifications. LEDES supplies electronic data-exchange formats that can carry billing information. For example, the LEDES 1998B specification describes a 24-field, pipe-delimited e-billing format.

A file can be structurally valid while containing a code the client does not accept. It can also contain a valid code beside an inadequate narrative, unapproved timekeeper, disputed rate, or prohibited activity. File validation answers whether the invoice is packaged correctly. It does not prove substantive compliance with the engagement terms.

The distinction is useful when a bill fails. A malformed field calls for an invoice or integration fix. A wrong task code calls for a coding fix. A disputed necessity or staffing decision calls for human review. Treating all three as “UTBMS errors” obscures the actual control that needs attention.

Why a valid code can still be wrong

Law firms usually work across three layers that change independently:

1. The industry code set. Multiple UTBMS families and revisions exist. The UTBMS code-set portal includes standards created by different bodies and for different practices or jurisdictions.
2. The client’s policy. An outside counsel guideline may require a subset, reserve catch-all values, add custom codes, or govern allowed combinations.
3. The firm’s system mapping. The billing system may store its own identifiers, names, and matter assignments for those values.

A timekeeper can choose a recognizable code from an online list and still miss the client-specific catalog held in the billing system. Conversely, the label shown on screen can look correct while its exported identifier points to the wrong record.

The safest source of truth is the accepted catalog for that client and matter, connected to the authoritative guideline and the billing-system identifiers that will actually be exported. A generic cheat sheet can help someone learn the vocabulary, but it should not govern a live invoice.

Common coding failures

Most recurring problems fit a small number of patterns:

• Task and activity are confused. The task does not describe the subject of the work, or the activity does not describe the action performed.
• The narrative and codes disagree. The entry says the lawyer researched an issue while the activity indicates a communication, for example.
• A broad fallback becomes the default. Repeated use of “other” can reduce the information UTBMS is meant to provide and may conflict with client rules.
• Coding happens during prebill. Billing staff reconstruct work after the context needed to distinguish similar tasks has faded.
• Client variations are ignored. A standard value is valid in the abstract but unavailable or redefined for this client.
• Distinct work is combined. One entry spans tasks or activities that the applicable guideline expects to be separated.
• The integration mapping is stale. The visible label and exported source identifier no longer agree.
• Automation is accepted as fact. A plausible suggestion is exported even though the evidence is incomplete.

None of these mistakes necessarily causes a rejection in every e-billing system. They are review risks whose consequences depend on the client’s rules, the invoice platform, and human judgment.

A client-aware coding workflow

Coding works best as a controlled process rather than a memory test.

1. Register the accepted source

For each client, record which code set or custom catalog governs, where it came from, and who owns its interpretation. Keep the applicable outside counsel guideline beside that decision. Do not activate a catalog merely because its name resembles a standard one.

2. Scope the catalog

Define whether the catalog applies across the firm, to a client, or only to a matter. Document which fields are required, optional, or prohibited. When instructions conflict, route the question to the billing administrator or relationship owner rather than asking every timekeeper to improvise.

3. Preserve billing-system identity

Synchronize the allowed task and activity values from the system that will receive the approved entries. Preserve the source identifiers as well as the labels. A mapping should be tested end to end: source catalog, time-entry display, reviewer choice, export, and billing-system result.

4. Suggest while the work is fresh

An assistive system can use the captured work, matter context, narrative, applicable guideline, and approved prior entries to narrow the candidate codes. The useful output is not an unexplained declaration of “the correct code.” It is a constrained suggestion that can show alternatives or leave the field blank when evidence is weak.

5. Check consistency before approval

Validate that required fields are present, the values belong to the allowed catalog, and the task, activity, narrative, and matter make sense together. This is also the point to identify work that may need to be split rather than forced under one code.

6. Require human review

The timekeeper or authorized reviewer should inspect the evidence, change or clear a suggestion, correct the narrative, and explicitly approve the entry. Automation should assist judgment, not conceal uncertainty or rewrite a truthful record to evade client review.

7. Keep later controls

Entry-time coding does not replace prebill, invoice, or e-billing checks. Rates, expenses, budgets, prior authorization, cross-entry patterns, file syntax, and submission deadlines may depend on information outside one entry. Use the control stage that can actually see the required facts.

What useful AI coding assistance should and should not do

Useful assistance narrows the decision using client and matter context. It can compare the work evidence with the narrative, suggest allowed task and activity values, flag an inconsistent pairing, present alternatives, and abstain when the evidence is insufficient.

It should not claim universal code correctness. It cannot infer a negotiated exception that has not been recorded, decide whether legal work was necessary, or guarantee that a client will pay the entry. A confident classification is not the same thing as a client-approved code or a payable invoice.

Hourglass follows this assistive model. The firm defines its task and activity codes, and those catalogs can be synchronized from the billing system. Hourglass uses the work record and applicable context to suggest values; when the evidence is insufficient, it can leave a field for the user. Approved entry-level rules can require or validate codes at firm, client, or matter scope, explain a problem, and suggest a correction.

The reviewer retains control. No draft reaches the billing system without explicit human approval. After approval, Hourglass sends the entry fields to the firm’s billing system, but it does not claim control over how every downstream platform transforms or displays them. Its current coding workflow focuses on task and activity codes, not expense processing or invoice-wide controls.

See the broader billing-compliance workflow and the current billing-system integration catalog for more context.

Measure coding as a process

One “accuracy” percentage cannot explain whether a coding workflow is healthy. Billing leaders should track separate measures:

Client reductions and appeal outcomes can be analyzed separately when the firm has reliable downstream data. They should not be treated as native proof of suggestion accuracy: an accurately coded entry can still be reduced for staffing, rate, budget, duration, necessity, or another reason.

The practical goal is not memorizing a public code table. It is making the right client-constrained decision while the work is still clear, preserving the system mapping, and keeping a person responsible for the final record.