ROI of Automatic Timekeeping: A Practical Law-Firm Model

Automatic timekeeping has an intuitive business case: record work that would otherwise be forgotten, reduce the effort required to reconstruct a day, and catch billing problems before an entry reaches prebill.

The difficult part is translating those improvements into a number a managing partner, CFO, or billing director can trust. A computer can observe work that never becomes billable. A draft can be rejected by the lawyer. An approved entry can be written down before invoicing, deducted by a client, or never collected. Time saved can create welcome capacity without reducing a single dollar of cost.

A useful model follows the work through those stages instead of multiplying every observed minute by a standard rate. It also lets the firm replace vendor assumptions with its own billing history and pilot results.

Start with the revenue path, not the activity log

Four different quantities are often collapsed into “recovered revenue”:

1. Observed activity is evidence that work occurred.
2. Approved time is work the timekeeper reviewed and released as a time entry.
3. Billed value is the amount that survives internal review and appears on an invoice.
4. Collected value is what the client ultimately pays.

The distinctions matter. Automatic capture can give a lawyer a more complete record of the day, but it should not declare every email, meeting, or document edit billable. Matter attribution, narrative quality, billing rules, and human judgment still sit between activity and an approved entry.

This is also why a historical benchmark should be treated carefully. An ABA practice-management article estimates that waiting until the end of the day can lose 10%–15% of potentially billable time. That is a practitioner’s estimate, not a controlled industry benchmark or a forecast for any particular firm. The firm’s own delayed-entry patterns are more useful.

Build the case from two separate benefit pools

The public Hourglass model keeps two potential benefits visible rather than hiding them inside one percentage.

1. Additional billed time

The calculator’s first estimate is:

timekeepers × 40 additional billed hours × billing rate

Attorney and paralegal counts and rates are entered separately. The 40-hour annual assumption is directional; it is not a promise that every licensed user will produce that result. It describes additional billed hours, not every minute the software observes and not collected cash.

For a purchase decision, improve the estimate with three adjustments:

• Use the number of eligible timekeepers who perform work where additional hourly entries can affect revenue.
• Apply sustained adoption from a representative pilot, rather than counting every purchased seat.
• Convert incremental billed value to expected collections using the firm’s own effective collection experience.

That creates a more conservative formula:

eligible timekeepers × adoption × incremental billed hours × effective collected rate

Where the effective collected rate is unavailable, the firm can approximate it as:

worked rate × billing realization × collection realization

Billing realization is billed value divided by worked value. Collection realization is collected value divided by billed value. Overall realization is collected value divided by worked value. Keeping those denominators explicit prevents a standard billing rate from being mistaken for cash.

2. Potential deduction recovery

For firms with e-billed work, the calculator adds an entry-level compliance estimate. Its starting input is deductions that remain after appeals. The default is 7%, which is equivalent to a 93% collection rate for the modeled revenue.

The model uses a 3% deduction floor, or a 97% collection ceiling. It then estimates that automatic entry-level checks may close 25%–75% of the gap between the firm’s deduction rate and that floor:

annual modeled fees × (deduction rate − 3% floor) × 25%–75%

The gap cannot fall below zero. At the default 7% deduction rate, the recoverable gap is four percentage points. Closing 25%–75% of that gap means a directional improvement of one to three percentage points, not eliminating all deductions and not guaranteeing 97% collection.

Only include issues the system can plausibly influence before export, such as supported narrative requirements, block billing, task and activity codes, prohibited terms, and configured outside counsel guideline rules. Exclude deductions caused by rates, budgets, staffing decisions, invoice formatting, unauthorized scope, or substantive disputes outside the time-entry workflow.

The annual attorney and paralegal hour requirements in the calculator are editable because they help size this modeled e-billed fee base. The defaults of 1,800 attorney hours and 1,400 paralegal hours are model inputs, not universal industry requirements.

Do not count the same benefit twice

Overlap is the easiest way to make an ROI model look better than reality.

If additional billed hours are already valued at an effective collected rate, do not apply the same compliance recovery percentage to them again. If an hour saved from reconstruction becomes additional billed work, value the collected work or the returned capacity, but not both. And if fewer prebill corrections are already reflected in improved billing realization, do not add a second line for the same avoided write-down.

A clean worksheet labels every benefit by its source and asks, “Would this dollar still exist if the other benefit lines were zero?” If the answer is no, the lines probably overlap.

Keep capacity separate from cash

Less end-of-day reconstruction and fewer narrative corrections can matter even when they do not create immediate revenue. The ABA’s 2024 Artificial Intelligence TechReport reports that saving time or increasing efficiency was lawyers’ leading perceived benefit of AI. Perceived efficiency, however, is not automatically a cash return.

Measure reconstruction and review time before and during a pilot. Report the result first as capacity returned. Convert it to a hard-dollar benefit only when the firm has an operational plan that changes overtime, contractor spend, headcount, or another actual cost. If lawyers use that capacity for additional billed work, use the collected value of that work instead of also claiming the labor cost as savings.

Add the full cost before calling the result ROI

The Hourglass ROI calculator is deliberately a directional impact calculator. It lets a firm edit attorney and paralegal counts, rates, annual hour requirements, whether clients require e-billing, and deductions after appeals. It shows estimated additional billed value and potential deduction recovery. It does not display an “annual ROI” multiple.

A firm-specific net-ROI analysis should add:

• subscription payments;
• the initial integration fee;
• implementation and customer-specific services;
• security and procurement review;
• training and change management;
• internal project labor; and
• recurring administration.

Separate one-time costs from recurring costs, and use the proposed commercial terms rather than a universal seat price.

Then calculate each decision measure by its proper name:

• Net benefit: hard-dollar benefits minus total cost.
• Net ROI: net benefit divided by total cost.
• Benefit-cost ratio: hard-dollar benefits divided by total cost.
• Payback period: the first month cumulative hard-dollar benefits exceed cumulative costs.

A large impact estimate can coexist with an unattractive first-year net ROI if implementation costs are high or adoption is weak. Conversely, a conservative benefit estimate may still pass if the break-even threshold is small.

Use three scenarios, with the conservative case able to stand alone

Create conservative, base, and upside columns, but do not fill them with generic vendor percentages.

For the conservative case, use the lower credible pilot result, sustained adoption rather than initial enrollment, the firm’s historical downside collection rate, zero unmeasured compliance benefit, and every known cost plus a contingency. The base case can use the pilot cohort’s median and expected cost. The upside case can use a supported upper bound, but never the best timekeeper, best week, or highest-rate matter as a firm-wide forecast.

Stress-test the inputs that drive the answer most:

• What if only half the planned users sustain the workflow?
• What if 40 additional billed hours becomes 20?
• What if none of the deduction gap is recoverable?
• What if the effective collected rate is materially below the entered billing rate?
• What if implementation takes longer or requires more internal labor?

The point is not to make the upside case disappear. It is to show the buyer which assumptions must be proven.

Design the pilot to produce finance-grade evidence

Agree on the measurement method before launch:

1. Choose a cooperative but representative cohort that includes strong, average, and inconsistent timekeepers.
2. Document the evaluation window, exclusions, fee arrangements, and baseline billing cycles.
3. Define adoption as sustained review and approval behavior, not logins.
4. Measure incremental approved and billed entries against each participant’s historical baseline and, where practical, a comparable non-user cohort.
5. Segment results by role, practice, client, matter type, and baseline timekeeping quality.
6. Follow incremental entries through billing and collection using the firm’s downstream data; Hourglass’s production trail ends when an approved entry is sent to the billing system.
7. Record review time, edit rates, and compliance warnings separately from downstream financial outcomes.

This makes the eventual business case auditable. It also reveals whether the value is broad or concentrated among a few users.

Change the model when the fee arrangement changes

The public calculator models hourly work, with an optional deduction-recovery component for e-billing. It should not be forced onto every matter.

For capped or fixed-fee work, additional recorded hours may improve margin visibility without increasing revenue. For contingency work, time records may support management or fee applications, but hourly value is not collected in the ordinary way. Court-fee work may require a different recovery assumption.

The practical rule is simple: model the economic mechanism the fee arrangement actually permits. If an extra approved hour cannot produce additional fees, do not call it recovered revenue.

Automatic timekeeping can create meaningful value, but the credible case is not “all observed time times rate.” Start with additional billed time, isolate addressable deduction recovery, keep capacity separate, subtract the full cost, and let a representative pilot replace assumptions with the firm’s own evidence.