Observed baseline
An observed result comes from a retained timer, timestamp, system record, invoice, review log, or other evidence from work that occurred.
Workflow baseline worksheet
Workflow baseline worksheet: measure one process before automation
This workflow baseline template helps you record how one repeated process works today. It keeps hands-on work, wall-clock duration, waiting, review, rework, exceptions, quality, and direct cost inside one defined boundary.
Complete it with evidence from a stated sample. You will have a baseline that a shadow test or pilot can measure against. The baseline can show what happened in the sample. It cannot predict ROI or prove what would have happened under a different process.
If the workflow is still broad, check workflow readiness first.Use internal references and broad process labels. Do not enter customer records, protected information, passwords, credentials, or other sensitive data into this public page. Worksheet values stay in this browser and are not submitted to BaristaLabs.
Workflow baseline
One unit of work
Sample · observed units
Active work · person-time
Elapsed time · wall clock
ObservedEstimatedUnknown
Evidence boundary
A baseline is a record of the sampled process
An observed baseline uses records from work that occurred inside a defined sample period. It names the unit of work, the start and end points, the acceptance rule, and the measures collected for each unit. These details let another person repeat the study or explain why a later comparison is different.
Estimate or modeled scenario
An estimate uses recall, a model, an assumption, or a range and states its method. A modeled version of work that did not occur is a counterfactual estimate, not an observed baseline.
Mark unsupported results unknown rather than entering zero. Unknown means no support is available yet. Blank, zero, estimated, and unknown are different states in the worksheet.
Define one comparable unit and boundary
One unit of work is the smallest repeated item with a clear trigger and an accepted output. Write where measurement starts and ends, then name the source inputs, systems, output, and acceptance rule. Include waiting, review, and exception work between those points. List work outside the boundary as an exclusion.
Choose a representative sample period. Record the number of observed units and any policy, staffing, system, demand, or tooling change that can make the sample less representative. Keep uncertainty visible.
BaristaLabs workflow baseline worksheetbaristalabs.io/learn/workflow-baseline-worksheetUse broad process labels. Keep sensitive records in their approved source system.
Blank worksheet
Complete the workflow baseline worksheet
Use one worksheet for one workflow. Keep raw evidence in the system where your team already stores it and enter only a safe reference here.
Draft baseline10 comparison fields still need an answer.
Privacy: use internal references and broad process labels. Do not enter customer records, protected information, passwords, credentials, or other sensitive data into this public page.
Workflow and measurement boundary
Define one comparable unit and the events inside the study.
Use the name the team uses for this repeated process.
Name the person accountable for the process result.
State the event that starts one unit of work.
Define the repeated item that each observation represents.
State the first event included in time and cost.
State the event that closes the measured unit.
List the records, requests, files, or approved sources needed to start.
Name the finished record, decision, message, file, or update.
State how a reviewer knows the output is complete and acceptable.
Refer to retained source records without copying sensitive data.
Sample
State what was observed and what could make the sample atypical.
Start date for the observed sample.
End date for the observed sample.
Include when time of day matters.
Count the units included in the sample; zero is different from blank.
State who recorded the work and which timers, timestamps, or records were used.
Record policy, staffing, system, demand, or tooling changes during the period.
List work, roles, costs, or cases outside the boundary.
Record missing evidence, estimated fields, and reasons the sample may be atypical.
Active work by meaningful phase
Keep hands-on person-time visible by recognizable phase.
Choose three to five phases the people doing the work can identify. Active work is hands-on person-time. Keep elapsed wall-clock time separate, and do not add review or rework twice.
Phase 01
Minutes, person-hours, cases, rounds, cost, or a qualitative note.
Leave unsupported numeric fields blank and name the missing evidence.
Phase 02
Minutes, person-hours, cases, rounds, cost, or a qualitative note.
Leave unsupported numeric fields blank and name the missing evidence.
Phase 03
Minutes, person-hours, cases, rounds, cost, or a qualitative note.
Leave unsupported numeric fields blank and name the missing evidence.
Observed unit log
Record counts, rates, durations, costs, and qualitative notes for each sampled unit.
Add one entry per sampled unit. Use safe internal references. Zero, blank, and unknown stay distinct; missing timestamps never generate elapsed time.
Unit 01 · Untitled
Use a safe internal reference that points back to retained evidence.
Leave unsupported numeric fields blank and name the missing evidence.
Review and rework may already be part of active work. Keep them visible, but do not add them to active work twice.
Baseline summary
Keep the sample count, unit, denominator, and uncertainty beside every result.
Enter summaries manually. An average without its sample count and spread can hide a few expensive cases. This worksheet does not calculate savings or ROI.
Active work by phase
Keep the sample count beside totals, rates, medians, ranges, and notes.
Leave unsupported numeric fields blank and name the missing evidence.
Total active work
Keep the sample count beside totals, rates, medians, ranges, and notes.
Leave unsupported numeric fields blank and name the missing evidence.
Elapsed time
Keep the sample count beside totals, rates, medians, ranges, and notes.
Leave unsupported numeric fields blank and name the missing evidence.
Material waiting or dependencies
Keep the sample count beside totals, rates, medians, ranges, and notes.
Leave unsupported numeric fields blank and name the missing evidence.
Review and approval effort
Keep the sample count beside totals, rates, medians, ranges, and notes.
Leave unsupported numeric fields blank and name the missing evidence.
Rework or reopened work
Keep the sample count beside totals, rates, medians, ranges, and notes.
Leave unsupported numeric fields blank and name the missing evidence.
Exceptions and handling
Use exceptions / observed units.
Leave unsupported numeric fields blank and name the missing evidence.
Acceptance failures
Use failed checks / units checked.
Leave unsupported numeric fields blank and name the missing evidence.
Direct tool or provider cost
Keep the sample count beside totals, rates, medians, ranges, and notes.
Leave unsupported numeric fields blank and name the missing evidence.
Pilot comparison and next decision
Plan the comparison without treating the scenario as observed baseline data.
Planned comparison — not observed baseline data
List active work, elapsed time, waiting, review, rework, exceptions, quality, and direct costs using the same definitions.
Keep design, integration, data preparation, testing, and training separate from recurring cost per unit.
Measurement notes
Keep the measures separate before you compare them
Active work
Person-time used to perform the process. Add each person’s work when people work at the same time.
Elapsed time
Wall-clock duration from the stated start to the stated end. It includes waiting, queues, dependencies, and sometimes parallel work.
Review and rework
Keep reviewer effort, rounds, corrections, repeated steps, and reopened work visible. They may already be part of active work, so do not count them twice.
Exceptions and quality
Use counts and rates when a denominator exists. Five failed checks across ten reviewed units is different from five across five hundred.
Direct cost
Include a provider or tool charge only when it belongs to the boundary. Name its charging unit, billing period, and allocation method.
Qualitative notes
Keep a dependency, process change, uncertainty, or exception type as a note when forcing it into hours or money would hide the useful fact.
Controlled comparison
Use the same boundary in a shadow test or pilot
A fair comparison uses the same class of input, unit, start and end points, acceptance rule, and evidence definitions. During a shadow test, people continue making the real decisions while the proposed path runs without sending messages or changing live records.
Collect misses, exceptions, reviewer changes, and direct cost beside time. Keep pilot setup cost separate. Record changes in staff, demand, systems, policy, source data, and scope.
Use the shadow-week guide to plan the comparison. Then connect the baseline to the pilot proof guide and keep item-level branches in the workflow case record.
Stop when the baseline cannot support a fair comparison
- The team cannot define one repeated unit or consistent start and end points.
- The acceptance rule, sample period, or observed-unit count is missing.
- Estimates look observed, or sample changes and exclusions are hidden.
- Review, rework, exceptions, quality, or direct costs sit outside the comparison.
A small sample can reveal missing evidence and expensive exception types. It should not support a broad savings claim unless it represents the work and keeps its limits visible.
Choose the next action
Choose the next action for this workflow
The workflow is still broad
Select one process with a clear owner and reviewable result before collecting a baseline.
Check workflow readinessThe workflow has a boundary
Review process automation when you need help mapping handoffs, exceptions, and integrations. Use the pricing and engagement guide to separate discovery, implementation, and recurring provider costs.
BaristaLabs review
Review one baseline with BaristaLabs
Bring the completed worksheet for one repeated process. BaristaLabs can review the measurement boundary, evidence gaps, exception pattern, and pilot comparison before implementation begins.
Review one baseline