Arrow Measure a recurring unit of work
Choose a consistent request type and period. Start with observed manual handling time, then enter a plausible covered share, routine review time and full exception-handling time.
Free tool04 / Operational capacity ↗Model the workload
Model the workload
that still needs a person.
Estimate what happens to workload when a workflow is partly automated. Include the routine review, the requests automation cannot cover and the exceptions that come back to a person.
Your inputs
Model automation with review and exceptions.
Use one recurring workflow and one currency throughout. Covered share is the portion attempted by automation. Exception rate applies only within that covered share. Exception minutes replace routine review for those cases.
Fictional property-diagnostics data for demonstration. These are not France DPE results or business facts.
Method / 04A useful result.
See how discovery and intake connect in the France DPE case ↗
A useful result.
A method you can inspect.
See how discovery and intake connect in the France DPE case ↗Compare before and after workload
Uncovered requests retain the original manual time. Covered, successful requests use routine review time. Covered exceptions use their full handling time. A workflow can increase workload if those assumptions are unfavorable.
Test the fragile assumption
Inspect the exception-rate sensitivity table and capacity per 100 staff-hours. Use a pilot to replace assumptions with observed times before treating the modeled capacity as an operational commitment.
Calculation rules & scope
Modeled minutes per request = (1 − covered share) × manual minutes + covered share × [(1 − exception rate) × review minutes + exception rate × exception minutes]. Monthly hours multiply that result by request volume and divide by 60. Capacity is the whole number of requests that fit in 100 staff-hours. Optional net labor value subtracts monthly operating cost from the modeled hours reclaimed times your hourly value; it is not cash savings or revenue.
Across industries, from question to action.
For a DPE-style intake, routine review could check a complete property request while an exception needs a full manual clarification. In a hotel, the unit might be a booking inquiry; in software, a support triage. Keep the same unit across manual and automated scenarios. This calculator does not assume staff reductions or increased conversion.
For the property-diagnostics example, explore the France DPE case study: public source quality and an actionable request belong in the same journey. The sample values in this tool are fictional.
Good questions.
Clear answers.
Is this a measured productivity result?
No. It is a scenario based on the values you enter. Actual workload depends on implementation, request mix, quality requirements and the measured exception rate.
What counts as an exception?
An exception is a request inside the covered share that cannot complete the routine path. Its handling time is the full human time for that case, replacing routine review in the model.
Can the result show negative time savings?
Yes. More review or expensive exceptions can make the modeled workflow take longer than the manual baseline. The tool preserves that result.
Is optional net labor value a financial forecast?
No. It values modeled time at your chosen hourly amount and subtracts your monthly operating cost assumption. It does not establish cash savings, revenue or a return on investment.
Continue the loop
All free tools ↗The next useful move.
02 — AI Systems / implementation
Lead Routing Blueprint Builder
Build and test lead routing rules.
Routing JSON + implementation briefOpen tool ↗
GEO / source quality
Brand Facts Checker
Check conflicting business facts.
Conflict report + source-of-truth JSONOpen tool ↗
Your next step / Arrow review
Make the result actionable.
Tell us what you want to improve with Automation Capacity Calculator. Your request goes to the Arrow team. You choose whether to include a short result summary.