Reporting control / SYS-02

Field service KPIs your software should expose: definitions before dashboards

A metric-dictionary and reporting test for service operators who need to trace demand, sales, capacity, completion, margin, cash, and retention back to operational records, not merely admire a dashboard.

Worldwide measurement framework / North American product evidenceVerified 10 Aug 2026Independent editorial review

Intent boundary: This is a formula, lineage, and reporting-control guide. It does not compare vendor reporting packages, prescribe universal performance targets, or replace accounting advice.

First deliverableMetric dictionary

Required behaviorDrill to source

Largest riskCohort mismatch

The evidence chain behind a useful KPI

01 / Decision

Start with the management action

A KPI earns space only when a named owner will investigate, coach, schedule, price, collect, or redesign something in response.

02 / Definition

Lock numerator, denominator, and cohort

Status, timestamp, inclusion, exclusion, attribution, currency, tax, refund, and cancellation rules must be explicit before two reports are compared.

03 / Event

Capture the operational source consistently

Calls, leads, estimates, visits, time, line items, completion codes, payments, and agreement events need required fields and reliable timestamps.

04 / Report

Aggregate without hiding the record

Filters, role permissions, time zones, backdating, and report refresh timing should be visible. Managers need to move from total to exception to source.

05 / Review

Explain movement before setting a target

Establish the business baseline, segment it by workflow and market, investigate data quality, then choose a target and review cadence.

A minimum KPI stack for a field-service operator

Demand + sales

From eligible opportunity to approved work

Keep volume, value, and speed separate so one large estimate or a mixed date window does not distort the operating story.

Booking rate = booked service opportunities ÷ eligible inbound service opportunities
Estimate win rate = approved estimates ÷ decided estimates in the same creation cohort
Median approval time uses the estimate-created to approved timestamp

Capacity + execution

From available field time to first-pass completion

A full schedule is not productive capacity, and a completed status does not prove the customer issue was resolved once.

Productive field utilization = productive field hours ÷ available field hours
First-time completion requires an explicit linked-return rule
Callback or rework rate must tie the return visit to the original job

Unit economics

From completed job to explainable contribution

Revenue alone can reward low-margin work. Cost completeness and direct-hour capture determine whether the result is decision-grade.

Gross margin = recognized revenue − direct labor − direct materials − other direct cost
Gross-margin percentage uses recognized revenue as the denominator
Gross margin per field hour exposes route, pricing, and scope effects together

Cash + retention

From invoice to collected and repeated value

Separate billed, collected, overdue, recurring, renewed, and canceled value; each state requires a different intervention.

Days to payment uses invoice and settled-payment timestamps
A/R aging must reconcile to customer-level open balances
Agreement renewal and cancellation need plan version and reason context

Data quality

Measure whether the other metrics can be trusted

Missing source, completion, cost, time, or attribution fields are operating exceptions, not invisible reporting trivia.

Publish the percentage of records missing each required field
Show excluded and unclassified records beside the KPI
Assign correction ownership and preserve the audit trail

The metric dictionary every report should carry

Dictionary field	Required answer	Reproduction test	Failure signal
Purpose + owner	Which decision changes, and who acts?	Ask the owner to name the next action for a high or low value	The metric is decorative or has multiple owners
Numerator + denominator	Exact record states and mathematical operation	Recalculate one period from exported source rows	The dashboard cannot be reproduced
Cohort + timestamps	Which event places a record in the period?	Create in one month and approve, cancel, or refund in another	Results change meaning across date filters
Inclusions + exclusions	Canceled, duplicate, warranty, no-charge, tax, tips, refunds, and unclassified treatment	Insert one example of every exception	Excluded records are invisible or undocumented
Attribution	Call source, salesperson, technician, team, branch, property, and service category rules	Reassign a shared job and inspect credit	One record is credited twice or to the wrong role
Drill-down + correction	Source rows, permission path, edit system, and refresh latency	Correct one source field and time the report update	The total changes without an auditable reason

Direct-test protocol: reproduce the report before trusting it

Create a ten-record control set

Include won, lost, open, canceled, duplicate, refunded, warranty, recurring, multi-tech, and cross-period work with known values.

Pass when: Expected totals and formulas are written before data entry.

Test cohort movement

Create an estimate in one period, approve it in the next, and compare creation-date, decision-date, and completion-date views.

Pass when: The interface names the cohort or the buyer can document it unambiguously.

Reproduce the formula

Export the contributing records and independently calculate count, value, rate, average, median, and margin where supported.

Pass when: The result matches within a documented rounding and refresh tolerance.

Stress permissions and corrections

View the report as admin, manager, dispatcher, and field user; then backdate, reassign, refund, and correct one source record.

Pass when: Visibility and recalculation follow a predictable policy with an audit trail.

Run the management meeting

Give the dashboard to the future owner and require a drill from signal to segment to record to assigned action.

Pass when: The meeting produces an owned exception list without a shadow spreadsheet.

Red flags

Stop treating ambiguity as capability.

A rate is shown without its denominator or eligible population
The same KPI name produces different values across reports without a reconciliation note
Averages hide skewed job values when no median or distribution is available
Managers cannot see which records were excluded
Permissions alter a total without making the visibility boundary obvious
The vendor demo uses polished sample data but cannot export the contributing rows

Measure in operation

Baseline the system, then improve it.

Required source fields completed
KPIs reproduced from export without adjustment
Unclassified or excluded records by metric
Time from dashboard signal to source exception
Corrections that change a closed reporting period
Management actions closed by the next review