Unit of work
What counts as one completion?
The plan defines a repeatable unit, such as one first draft or one completed summary, and states what review time is included.
Measurement methodology
A number is useful only when you can inspect how it was made.
PrivateStride measures workflow time against an agreed baseline. The engagement-specific plan defines the work, data, assumptions, and target before results are judged.
Step 1 · Define
Draw the workflow boundary before timing it.
A broad label such as “draft a memo” is not enough. The baseline document identifies the starting event, finishing event, included tasks, responsible roles, and the kinds of work considered comparable.
Comparable work
What belongs in the comparison?
Work may be grouped by complexity, service line, role, or other factors that materially affect time. Unlike work is not pooled merely to improve the result.
Source
Where does the time come from?
The plan can use direct observation, system timestamps, timekeeping records, structured samples, or validated team estimates. The source and its weaknesses are disclosed.
Step 2 · Baseline
Measure the old workflow before redesigning it.
The observation period, sample, timing method, exclusions, and any segmentation are agreed for the engagement. A median or another documented statistic may be used when it better represents typical work than a simple average.
Core calculation
Recovered time starts with a direct difference.
For comparable completed work:
recovered hours = (baseline time − current time) ÷ 60 × completed units
Times are stated in minutes per unit. A negative result means the workflow added time. It is not silently converted to zero in the underlying review.
Firm-level view
Aggregate only after the work is comparable.
Workflow results can be summed after agreed adjustments for volume and work mix. To express a weekly professional measure:
hours per professional per week = total recovered hours ÷ participating professionals ÷ measured weeks
The dashboard identifies the population and dates behind the denominator.
Step 3 · Interpret
Capacity is not automatically revenue.
Time recovered is an operational measure. What the firm does with that capacity determines the financial result.
Annualized hours
Weekly recovered hours may be multiplied by an agreed number of working weeks. This is a projection, not an observed year.
Estimated economic value
Recovered hours may be multiplied by an agreed rate. The rate, utilization assumption, and calculation are shown. The result is not the same as collected revenue or profit.
Usage and participation
Usage can explain whether the new workflow was adopted. It is a leading indicator, not proof that time was recovered or work quality improved.
Quality and review
Faster work still has to be fit for use.
AI-assisted output is a draft. The firm’s professionals review facts, calculations, citations, tax positions, advice, and client communications using the firm’s normal quality controls. Rework that falls inside the defined workflow boundary belongs in the timing.
PrivateStride does not treat a faster first draft as recovered time if required review or correction simply moved elsewhere in the measured process.
Human accountability
The model does not become the signer.
Professionals remain accountable for judgment and delivered work. The measurement process evaluates workflow performance; it does not certify the accuracy of every output or transfer a professional obligation to software.
Assumptions and limits
A useful result names what else could have changed.
Common sources of uncertainty
- Seasonality or a short observation window.
- Changes in work complexity, volume, or client mix.
- Staffing, experience, management, or policy changes.
- Concurrent software or process improvements.
- Self-reported timing, small samples, or missing records.
- Learning effects during the early adoption period.
How the report handles them
Material assumptions, exclusions, data gaps, and concurrent changes are documented. Where the work cannot be compared reliably, it is separated, qualified, or omitted with the reason shown. An observed change can support a practical decision without proving that AI alone caused every minute of the difference.
The engagement plan, not a generic website formula, controls the measurement used for a guarantee.
Claims discipline
Projection, measured result, and client example are different things.
Estimates are labeled as estimates and carry their assumptions. Measured results identify the period, population, workflow, and method. We do not invent client outcomes, present modeled capacity as collected revenue, or publish an identifiable client result without permission.
Related information
See how the method supports the promise.
Read the public guarantee terms , the site’s privacy notice , and the main implementation overview .
Capacity & AI risk assessment
Start with assumptions you can inspect.
The first audit estimates where capacity may exist. An engagement begins only after the workflow boundaries, baseline plan, and target are agreed.
Book your audit 30 min · No preparation · Confidential