Robot or Human? · Methodology
How the model works
Plain formulas, public data, and a list of every number we had to assume. Version 1.0.0.
1. What each option costs
People, employed. Hours needed per month = workload ÷ output per person-hour. Annual cost = hours × months × hourly wage × employer on-cost multiplier (+ equipment per hour, if you add one).
People or robots, hired as a service. Annual cost = units per year × price per unit.
Robots you buy, lease or subscribe to. The robot handles the share of work it can do (1 − edge share); the rest is still bought at the baseline's unit cost. Robot-hours needed = robot work ÷ robot output per hour. Robots needed = robot-hours ÷ (operating hours × days × uptime), rounded up. Then:
- Capital: (robots × price + integration) repaid as an annuity at the financing rate over the useful life. Maintenance = robots × price × maintenance %.
- Lease and subscription: fees × robots × 12 (+ per-hour fees × robot-hours), with integration annuitised.
- Supervision: robot-hours × staff minutes per robot-hour ÷ 60 × loaded wage.
- Failures: the share of robot work lost to faults is redone at the baseline's unit cost.
- Energy: robot-hours × kWh per hour × price per kWh.
2. Payback, break-even utilisation and NPV
Payback (months) = upfront cost ÷ (baseline monthly cost − option monthly operating cost). "Never" when there is no saving.
Break-even utilisation u* = F ÷ (M × (v − c)), where F is the fixed annual cost of one robot, M its maximum productive hours per year, v the value of the human work one robot-hour displaces (robot output per hour × (1 − failure share) × baseline unit cost) and c the variable cost per robot-hour (supervision, energy, per-hour fees). Above 100% means even a robot that never stops would not pay.
NPV over 5 years = −upfront + Σm=1..60 (baseline costm − option operating costm) ÷ (1 + r/12)m + residual value. Labour-linked costs (wages, supervision, leftover human work, the labour share of service prices) grow with wage growth each year; contract fees, maintenance and energy are held flat. Owned robots are replaced at the end of their life if that falls inside the horizon, and the remaining straight-line value is credited at the end.
The payback and NPV functions are checked against hand-computed examples in scripts/rvh.test.mjs (for example: $72,000 a year of human work against two $50,000 robots with $10,000 integration and 10% maintenance gives a 21.3-month payback and a $200,000 five-year NPV at 0% discount).
3. Projections
Hardware prices follow Wright's law: pricet = price0 × (cumulative unitst ÷ cumulative units0)−b, with b = −log₂(1 − learning rate). We approximate cumulative-unit growth with annual shipment growth (exact only when shipments grow at a steady rate). Lease and subscription fees are split into a hardware share (follows the learning curve) and a service share (follows wages); robot service prices are split into equipment and labour the same way; integration follows wages, because it is mostly people. Robot productivity improves at the scenario rate. Everything is in constant 2026 money. The crossover year is the first year the cheapest robot option costs no more than the baseline; "now" means already true in 2026; "after 2035" means not inside the window.
4. Where the numbers come from
- Wages: US: BLS OEWS May 2025 medians by occupation, national and 8 states. Japan: MHLW 2024 Basic Survey on Wage Structure by occupation, including bonuses. China: NBS 2024 average annual wages by position ÷ (48.6 h × 52). Germany, UK, India, Brazil, Mexico: ILOSTAT mean hourly earnings by ISCO-08 major group from national surveys (Germany uprated from 2022 to 2025 with Eurostat). Major-group averages are not the exact job; the calculator says so next to the wage.
- Employer on-costs: US from BLS ECEC (wages are 70.0% of compensation); Germany from Eurostat labour cost levels; UK derived from statutory employer NIC (15%) and pension (3%). Japan, China, India, Brazil and Mexico are assumptions with the statutory components listed: Japan ×1.16, China ×1.3, India ×1.1, Brazil ×1.6, Mexico ×1.3.
- Robot and service prices: Whirrwork's own price observations (each with a public source URL) plus vendor price pages, municipal contracts and extension-service surveys. Vendor numbers are labelled as vendor numbers. US prices are converted for other countries at ECB reference rates (2026-10-23) and flagged; service prices are also adjusted for local labour cost, which is a modelled estimate.
- Trend points and learning rates: see the trends page and its receipts.
5. Every assumption without a source
These are ours. They are editable in each calculator and highlighted there. If you have a better number with a source, the calculator will take it, and we would like to hear about it.
| Job | Input | Default | Why |
|---|---|---|---|
| Commercial floor cleaning | volume | 40000 | Square feet scrubbed per day: a big-box store or small warehouse. Use yours. |
| Commercial floor cleaning | hours | 8 | Hours per day the floor is available for cleaning. |
| Commercial floor cleaning | days | 22 | Cleaning days per month. |
| Commercial floor cleaning | months | 12 | Active months per year. |
| Commercial floor cleaning | ratio | 1.6 | Robot sq ft per hour relative to a person with a walk-behind scrubber. Vendor theoretical maxima are about 3.6x (Avidbots Neo 2: 33,583 ft²/h); real floors with turns, obstacles and tank changes run far below that. |
| Commercial floor cleaning | integration | 2000 | Site mapping, training and docking setup per site. |
| Commercial floor cleaning | maint_pct | 0.08 | Yearly maintenance and consumables (pads, squeegees, batteries) as a share of purchase price. |
| Commercial floor cleaning | life | 5 | Useful life in years. |
| Commercial floor cleaning | uptime | 0.85 | Share of the cleaning window the robot is actually cleaning (charging, tank changes). |
| Commercial floor cleaning | failure | 0.03 | Share of robot work lost to faults that a person has to cover. |
| Commercial floor cleaning | edge | 0.15 | Share of the area robots skip (corners, under racking, spills, stairs) that a person still cleans. |
| Commercial floor cleaning | sup_min | 6 | Minutes of staff time per robot-hour (start, empty, refill, rescue). |
| Commercial floor cleaning | kwh_h | 1.2 | kWh per robot operating hour. |
| Warehouse order picking (AMR-assisted) | volume | 4000 | Order lines picked per day. |
| Warehouse order picking (AMR-assisted) | hours | 16 | Operating hours per day (two shifts). |
| Warehouse order picking (AMR-assisted) | days | 22 | Operating days per month. |
| Warehouse order picking (AMR-assisted) | months | 12 | Active months per year. |
| Warehouse order picking (AMR-assisted) | human_rate | 80 | Picks per hour for a person walking a cart. Varies hugely with order profile; use your own WMS figure. |
| Warehouse order picking (AMR-assisted) | robot_rate | 60 | Picks per robot-hour. Collaborative AMRs do the walking; people still do the picking, so this is robot throughput, not autonomy. |
| Warehouse order picking (AMR-assisted) | integration | 30000 | WMS integration, mapping and change management for a first site. Integration often costs more than the robots. |
| Warehouse order picking (AMR-assisted) | life | 5 | Contract or useful life in years. |
| Warehouse order picking (AMR-assisted) | uptime | 0.9 | Share of the shift each robot is available. |
| Warehouse order picking (AMR-assisted) | failure | 0.02 | Share of robot work lost to faults. |
| Warehouse order picking (AMR-assisted) | edge | 0.1 | Share of picks that stay manual (oversize, hazmat, rush orders). |
| Warehouse order picking (AMR-assisted) | sup_min | 30 | Picker minutes per robot-hour. In assisted picking a person meets the robot at each pick face, so this is large by design. |
| Warehouse order picking (AMR-assisted) | kwh_h | 0.3 | kWh per robot-hour. |
| Pallet moving | volume | 200 | Pallet moves per day on fixed routes (dock to racking, line-side). |
| Pallet moving | hours | 16 | Operating hours per day. |
| Pallet moving | days | 22 | Operating days per month. |
| Pallet moving | months | 12 | Active months per year. |
| Pallet moving | human_rate | 12 | Moves per operator-hour on these routes. |
| Pallet moving | ratio | 0.7 | AMR moves per hour relative to a person: slower travel, no shortcuts. |
| Pallet moving | integration | 5000 | Site prep, floor marking and change management on top of the vendor package. |
| Pallet moving | maint_pct | 0.06 | Yearly maintenance and software after the included period, share of price. |
| Pallet moving | life | 6 | Useful life in years. |
| Pallet moving | uptime | 0.85 | Availability after charging. |
| Pallet moving | failure | 0.03 | Work lost to faults. |
| Pallet moving | edge | 0.25 | Moves that need a person (damaged pallets, ad-hoc moves, trailers). |
| Pallet moving | sup_min | 5 | Staff minutes per robot-hour (calling moves, clearing blockages). |
| Pallet moving | kwh_h | 1 | kWh per robot-hour. |
| Pallet moving | human_equip_h | 0 | Truck or forklift cost per operator-hour. Default 0 means the comparison ignores the forklift you might retire; add yours. |
| Security patrol | volume | 24 | Patrol hours needed per day (24 = round the clock). |
| Security patrol | hours | 24 | Operating hours per day. |
| Security patrol | days | 30.4 | Days per month (every day). |
| Security patrol | months | 12 | Active months per year. |
| Security patrol | human_rate | 1 | One guard-hour covers one patrol-hour. |
| Security patrol | ratio | 1 | One robot-hour covers one patrol-hour of a defined route (outdoor lots, perimeters). |
| Security patrol | integration | 3000 | Site survey, connectivity and alarm-centre integration. |
| Security patrol | life | 3 | Contract length in years. |
| Security patrol | uptime | 0.9 | Share of hours the robot patrols (charging is scheduled). |
| Security patrol | failure | 0.03 | Hours lost to faults, covered by a guard. |
| Security patrol | edge | 0.35 | Share of guard hours spent on things a patrol robot cannot do: responding, access control, talking to people. A person stays on call for these. |
| Security patrol | sup_min | 6 | Remote-monitoring minutes per robot-hour (watching alerts, often shared across sites). |
| Security patrol | kwh_h | 0 | Included in the contract. |
| Home lawn mowing | volume | 4.3 | Mows per month in season (weekly). |
| Home lawn mowing | months | 7 | Mowing months per year; depends on climate. |
| Home lawn mowing | hours | 10 | Hours per day the robot may run. |
| Home lawn mowing | days | 30 | Days per month. |
| Home lawn mowing | robot_rate | 0.5 | Mows completed per robot-hour for a quarter-acre yard (about 2 hours per full mow). |
| Home lawn mowing | life | 5 | Useful life in years. |
| Home lawn mowing | uptime | 0.8 | Share of allowed hours the mower can work (rain, charging). |
| Home lawn mowing | failure | 0.03 | Mows lost to faults or stuck robots. |
| Home lawn mowing | edge | 0.15 | Edging and trimming the robot can't do, as a share of a service visit you still pay for. |
| Home lawn mowing | sup_min | 0 | Your own time is not costed here (you would spend it anyway, or enjoy it). |
| Home lawn mowing | kwh_h | 0.05 | kWh per robot-hour. |
| Commercial mowing | volume | 12 | Acres mowed per working day across your sites. |
| Commercial mowing | hours | 10 | Hours per day the mower may run. |
| Commercial mowing | days | 22 | Working days per month. |
| Commercial mowing | months | 8 | Mowing months per year. |
| Commercial mowing | human_rate | 2 | Acres per hour for an experienced operator on a commercial zero-turn mower. |
| Commercial mowing | ratio | 0.7 | Autonomous mower acres per hour relative to a person. |
| Commercial mowing | integration | 1000 | Site mapping and crew training. |
| Commercial mowing | life | 3 | Lease term in years. |
| Commercial mowing | uptime | 0.85 | Availability. |
| Commercial mowing | failure | 0.03 | Work lost to faults. |
| Commercial mowing | edge | 0.25 | Trimming, slopes and obstacles a person still handles. |
| Commercial mowing | sup_min | 20 | Crew minutes per robot-hour: transport, supervision while it mows nearby. |
| Commercial mowing | kwh_h | 3 | kWh per robot-hour. |
| Commercial mowing | human_equip_h | 0 | Your mower's fuel and ownership cost per hour. Default 0 ignores it. |
| Pool cleaning | volume | 1 | One month of pool care. |
| Pool cleaning | months | 12 | Months the pool is open. |
| Pool cleaning | hours | 3 | Robot cycle hours per day it runs. |
| Pool cleaning | days | 8 | Cleaning cycles per month. |
| Pool cleaning | robot_rate | 0.042 | A month of cleaning needs about 24 robot-hours (8 cycles of 3 hours). |
| Pool cleaning | integration | 0 | Plug it in. |
| Pool cleaning | maint_pct | 0.08 | Yearly parts (filters, brushes, eventually a motor) as a share of price. |
| Pool cleaning | life | 5 | Useful life in years. |
| Pool cleaning | uptime | 0.95 | Availability. |
| Pool cleaning | failure | 0.03 | Months lost to faults. |
| Pool cleaning | edge | 0.5 | Share of the service you still need: water chemistry, filter and equipment checks. Many owners do these themselves (then set 0). |
| Pool cleaning | sup_min | 0 | Your time is not costed. |
| Pool cleaning | kwh_h | 0.2 | kWh per robot-hour. |
| Window and facade cleaning | volume | 120000 | Facade square feet cleaned per year (e.g. 60,000 sq ft twice). |
| Window and facade cleaning | months | 12 | Months per year (volume is annual). |
| Window and facade cleaning | hours | 8 | Working hours per day. |
| Window and facade cleaning | days | 20 | Working days per month. |
| Window and facade cleaning | robot_rate | 1500 | Facade sq ft per drone-hour including setup and moves. |
| Window and facade cleaning | integration | 5000 | Pilot certification, insurance and gear to run your own drone. |
| Window and facade cleaning | life | 3 | Subscription commitment in years. |
| Window and facade cleaning | uptime | 0.6 | Share of working hours flyable (wind, rain, battery swaps). |
| Window and facade cleaning | failure | 0.02 | Work lost to faults. |
| Window and facade cleaning | edge | 0.2 | Share still done by rope or lift (recessed windows, detail work, inspection). |
| Window and facade cleaning | sup_min | 120 | Crew minutes per drone-hour: pilot plus a ground person for the hose and the public. |
| Window and facade cleaning | kwh_h | 2 | kWh per drone-hour. |
| Roof inspection | volume | 10 | Roof inspections per month. |
| Roof inspection | months | 12 | Months per year. |
| Roof inspection | hours | 8 | Working hours per day. |
| Roof inspection | days | 20 | Working days per month. |
| Roof inspection | edge | 0.1 | Inspections where someone still climbs (soft spots, interior attic checks). |
| Cell tower inspection | volume | 20 | Towers inspected per month. |
| Cell tower inspection | months | 12 | Months per year. |
| Cell tower inspection | hours | 8 | Hours per day. |
| Cell tower inspection | days | 20 | Days per month. |
| Cell tower inspection | edge | 0.15 | Towers that still need a climb (hands-on testing, torque checks, repairs). |
| Wind turbine blade inspection | volume | 50 | Turbines inspected per year. |
| Wind turbine blade inspection | months | 12 | Months per year. |
| Wind turbine blade inspection | hours | 8 | Hours per day. |
| Wind turbine blade inspection | days | 20 | Days per month. |
| Wind turbine blade inspection | edge | 0.1 | Share of turbines needing rope access anyway (repairs, lightning-protection tests). |
| Solar farm inspection | volume | 50 | MW of plant inspected per year. |
| Solar farm inspection | months | 12 | Months per year. |
| Solar farm inspection | hours | 8 | Hours per day. |
| Solar farm inspection | days | 20 | Days per month. |
| Solar farm inspection | edge | 0.05 | Share needing a person on site anyway (string-level electrical tests). |
| Crop spraying | volume | 2000 | Acres sprayed per year (all passes). |
| Crop spraying | months | 12 | Volume is annual. |
| Crop spraying | hours | 8 | Flying hours per spray day. |
| Crop spraying | days | 10 | Spray days per month, averaged. |
| Crop spraying | robot_rate | 25 | Acres per drone-hour in practice, including refills and moves. Spec sheets quote more. |
| Crop spraying | integration | 12000 | Extra batteries, generator, mixing trailer, certification (FAA Part 137 in the US) and insurance. |
| Crop spraying | maint_pct | 0.1 | Yearly repairs and battery replacement, share of price. |
| Crop spraying | life | 4 | Useful life in years. |
| Crop spraying | uptime | 0.7 | Share of spray hours flyable (wind, rain). |
| Crop spraying | failure | 0.03 | Work lost to faults. |
| Crop spraying | edge | 0.05 | Acres a drone cannot do well (very large flat fields at high volume rates). |
| Crop spraying | sup_min | 120 | Crew minutes per drone-hour: pilot plus a mixer/visual observer. |
| Crop spraying | kwh_h | 10 | Generator or grid energy per drone-hour (kWh). |
| Drone light show vs fireworks | volume | 1 | Shows per year. |
| Drone light show vs fireworks | months | 12 | Volume is annual. |
| Drone light show vs fireworks | edge | 0 | Nothing left over. |
| Restaurant food running | volume | 10 | Runner hours needed per day (lunch and dinner). |
| Restaurant food running | hours | 12 | Opening hours per day. |
| Restaurant food running | days | 30 | Open days per month. |
| Restaurant food running | months | 12 | Months per year. |
| Restaurant food running | human_rate | 1 | One runner-hour. |
| Restaurant food running | ratio | 0.5 | Share of a runner-hour of work one robot-hour replaces (it carries; servers still serve). |
| Restaurant food running | integration | 500 | Mapping the dining room and staff training. |
| Restaurant food running | maint_pct | 0.05 | Yearly maintenance share of price. |
| Restaurant food running | life | 4 | Useful life in years. |
| Restaurant food running | uptime | 0.9 | Availability. |
| Restaurant food running | failure | 0.03 | Work lost to faults. |
| Restaurant food running | edge | 0.3 | Running the robot cannot do: stairs, terraces, crowded floors, big parties. |
| Restaurant food running | sup_min | 5 | Staff minutes per robot-hour beyond normal serving (loading, clearing paths). |
| Restaurant food running | kwh_h | 0.1 | kWh per robot-hour. |
| Hospital deliveries | volume | 80 | Deliveries per day (meds, linens, specimens, meals). |
| Hospital deliveries | hours | 20 | Operating hours per day. |
| Hospital deliveries | days | 30.4 | Every day. |
| Hospital deliveries | months | 12 | Months per year. |
| Hospital deliveries | human_rate | 3 | Deliveries per porter-hour including waiting for lifts. |
| Hospital deliveries | ratio | 0.8 | Robot deliveries per hour relative to a porter (slower, but no breaks). |
| Hospital deliveries | integration | 25000 | Lift and door interfaces, Wi-Fi survey, charging bays. Often the hardest part in older buildings. |
| Hospital deliveries | maint_pct | 0.1 | Yearly service contract share of price (owned). |
| Hospital deliveries | life | 7 | Useful life in years. |
| Hospital deliveries | uptime | 0.85 | Availability after charging. |
| Hospital deliveries | failure | 0.03 | Runs lost to faults. |
| Hospital deliveries | edge | 0.3 | Deliveries that stay human: STAT runs, patient-facing handovers, odd items. |
| Hospital deliveries | sup_min | 6 | Staff minutes per robot-hour loading and unloading (beyond what a porter run also needs). |
| Hospital deliveries | kwh_h | 0.5 | kWh per robot-hour. |
| CNC / injection machine tending | volume | 16 | Hours per day the machine needs tending. |
| CNC / injection machine tending | hours | 16 | Shift hours per day. |
| CNC / injection machine tending | days | 22 | Working days per month. |
| CNC / injection machine tending | months | 12 | Months per year. |
| CNC / injection machine tending | human_rate | 1 | One operator-hour. |
| CNC / injection machine tending | ratio | 1 | The machine cycle sets the pace, so one robot-hour covers one operator-hour. |
| CNC / injection machine tending | integration | 40000 | Gripper, fixtures, safety, guarding, programming. Integrators say the arm is often half or less of the total. |
| CNC / injection machine tending | life | 8 | Useful life in years. |
| CNC / injection machine tending | uptime | 0.92 | Availability. |
| CNC / injection machine tending | failure | 0.03 | Hours lost to faults and jams. |
| CNC / injection machine tending | edge | 0.1 | Changeovers, first-article checks and odd parts that need a person. |
| CNC / injection machine tending | sup_min | 8 | Operator minutes per robot-hour (loading trays, inspections, restarts). |
| CNC / injection machine tending | kwh_h | 0.3 | kWh per robot-hour. |
| Welding | volume | 8 | Manual welding hours of work per day. |
| Welding | hours | 8 | Shift hours per day. |
| Welding | days | 22 | Working days per month. |
| Welding | months | 12 | Months per year. |
| Welding | human_rate | 1 | One manual welding hour. |
| Welding | ratio | 2 | Output of one cobot-hour relative to a manual welder-hour. The vendor says most shops see 2-4x; we take the low end. |
| Welding | integration | 15000 | Fixtures, guarding, training beyond the package. Integrators quote $15,000-$40,000 for un-packaged systems. |
| Welding | life | 8 | Useful life in years. |
| Welding | uptime | 0.85 | Arc-ready share of the shift (loading, tip changes). |
| Welding | failure | 0.03 | Hours lost to faults. |
| Welding | edge | 0.3 | Welds that stay manual: one-offs, awkward joints, repairs. |
| Welding | sup_min | 30 | Welder minutes per robot-hour (fixturing, loading, inspecting). Cobot welding is a person plus a robot. |
| Welding | kwh_h | 6 | kWh per hour of welding. |
| End-of-line palletizing | volume | 3000 | Cases stacked per day. |
| End-of-line palletizing | hours | 16 | Line hours per day. |
| End-of-line palletizing | days | 22 | Working days per month. |
| End-of-line palletizing | months | 12 | Months per year. |
| End-of-line palletizing | human_rate | 250 | Cases per person-hour (light cases; heavy ones are slower and harder on backs). |
| End-of-line palletizing | ratio | 1.2 | Robot cases per hour relative to a person. |
| End-of-line palletizing | integration | 10000 | Conveyor tie-in, safety scanners, pattern programming. |
| End-of-line palletizing | maint_pct | 0.04 | Yearly maintenance share of price. |
| End-of-line palletizing | life | 10 | Useful life in years. |
| End-of-line palletizing | uptime | 0.93 | Availability (pallet changes). |
| End-of-line palletizing | failure | 0.02 | Output lost to faults. |
| End-of-line palletizing | edge | 0.05 | Mixed or damaged cases stacked by hand. |
| End-of-line palletizing | sup_min | 6 | Operator minutes per robot-hour (pallet swaps, slip sheets). |
| End-of-line palletizing | kwh_h | 0.5 | kWh per robot-hour. |
| Warehouse inventory counting | volume | 30000 | Pallet locations counted per month (e.g. 15,000 locations twice). |
| Warehouse inventory counting | hours | 16 | Hours per day the drone may fly. |
| Warehouse inventory counting | days | 22 | Days per month. |
| Warehouse inventory counting | months | 12 | Months per year. |
| Warehouse inventory counting | human_rate | 60 | Locations per person-hour with an order picker or reach truck for high bays. |
| Warehouse inventory counting | robot_rate | 150 | Locations per drone-hour. |
| Warehouse inventory counting | integration | 5000 | WMS integration and site setup. |
| Warehouse inventory counting | life | 3 | Contract years. |
| Warehouse inventory counting | uptime | 0.8 | Availability (charging). |
| Warehouse inventory counting | failure | 0.03 | Work lost to faults. |
| Warehouse inventory counting | edge | 0.1 | Locations that still need a person (unreadable labels, exceptions to resolve). |
| Warehouse inventory counting | sup_min | 3 | Staff minutes per drone-hour. |
| Warehouse inventory counting | kwh_h | 0.2 | kWh per drone-hour. |
| Warehouse inventory counting | human_equip_h | 0 | Lift truck cost per counter-hour. Default 0 ignores it. |
| Residential snow clearing | volume | 4 | Clearings per winter month. |
| Residential snow clearing | months | 4 | Snow months per year. |
| Residential snow clearing | hours | 6 | Hours per day it may run. |
| Residential snow clearing | days | 30 | Days per month. |
| Residential snow clearing | robot_rate | 1 | Clearings per robot-hour for a two-car driveway. |
| Residential snow clearing | integration | 0 | DIY setup. |
| Residential snow clearing | maint_pct | 0.05 | Yearly parts share of price. |
| Residential snow clearing | life | 6 | Useful life in years. |
| Residential snow clearing | uptime | 0.9 | Availability. |
| Residential snow clearing | failure | 0.05 | Clearings lost to faults or stuck robots. |
| Residential snow clearing | edge | 0.2 | Visits you still need: heavy wet snow, the plough berm at the end of the drive. |
| Residential snow clearing | sup_min | 0 | Your time is not costed. |
| Residential snow clearing | kwh_h | 1.5 | kWh per robot-hour. |
| Any other job (build your own) | volume | 100 | Units of work per day. |
| Any other job (build your own) | hours | 8 | Operating hours per day. |
| Any other job (build your own) | days | 22 | Working days per month. |
| Any other job (build your own) | months | 12 | Months per year. |
| Any other job (build your own) | human_rate | 10 | Units per person-hour. |
| Any other job (build your own) | ratio | 1 | Robot units per hour relative to a person. |
| Any other job (build your own) | purchase | 50000 | Robot price. Replace with a quote. |
| Any other job (build your own) | lease | 1500 | Lease per robot-month. |
| Any other job (build your own) | raas | 1500 | RaaS per robot-month. |
| Any other job (build your own) | integration | 20000 | Integration. In factories this is often 1-3x the robot price. |
| Any other job (build your own) | maint_pct | 0.06 | Yearly maintenance share of price. |
| Any other job (build your own) | life | 6 | Useful life in years. |
| Any other job (build your own) | uptime | 0.85 | Availability. |
| Any other job (build your own) | failure | 0.03 | Work lost to faults. |
| Any other job (build your own) | edge | 0.15 | Work that stays human. |
| Any other job (build your own) | sup_min | 6 | Staff minutes per robot-hour. |
| Any other job (build your own) | kwh_h | 0.5 | kWh per robot-hour. |
| All jobs | rate | 0.08 | Annual financing / discount rate. Nobody publishes yours; 8% is a round mid-2026 commercial borrowing guess. |
| All jobs | kwh_price | 0.15 | Electricity price per kWh (USD). Check your bill; energy is rarely a big line. |
| All jobs | horizon | 5 | Years in the NPV comparison. |
| All jobs | svc_labour_share | 0.6 | Share of a human service price that is labour (rises with wages). Used to escalate and to adjust US service prices to other countries. |
| All jobs | robot_svc_labour_share | 0.4 | Share of a robot service price that is labour (pilot, technician). The rest is equipment, which learning curves make cheaper. |
| All jobs | fee_hw_share | 0.6 | Share of a lease or RaaS fee that pays for hardware (learns down). The rest is service and support (moves with wages). |
6. What a cost model cannot tell you
- Integration and change management dominate real projects and vary by an order of magnitude. Our defaults are mid-range guesses; get quotes.
- Robots rarely replace a job one-for-one. They replace tasks; the "edge share" and "supervision" inputs are where most of the honesty lives.
- Vendors fail. iRobot went through Chapter 11 in December 2025 (iRobot); Walmart dropped shelf-scanning robots in 2020 (Reuters); Japan's Henn-na Hotel retired more than half its 243 robots in 2019 (The Verge). A five-year NPV assumes your vendor is still there in year five.
- Exchange rates, interest rates and wages move. Crossovers in emerging economies are especially sensitive to the exchange rate.
- None of this is financial advice. It is arithmetic with receipts.
7. Versions and reuse
Model version 1.0.0. Changes to formulas or defaults bump the version; every published projection is logged append-only with its version on the track record page. Data and model are CC BY 4.0: download them, or call the API and MCP tool.
Cite this
APA
Whirrwork. (2026). Robot or Human? cost model and methodology (Version 1.0.0) [Data set and interactive model]. https://whirrwork.com/tools/robot-vs-human/methodology/
Chicago
Whirrwork. “Robot or Human? cost model and methodology.” Version 1.0.0. 2026. Accessed October 25, 2026. https://whirrwork.com/tools/robot-vs-human/methodology/.
MLA
Whirrwork. “Robot or Human? cost model and methodology.” Whirrwork, 2026, whirrwork.com/tools/robot-vs-human/methodology/. Accessed 25 Oct 2026.
BibTeX
@misc{whirrwork2026robot,
author = {{Whirrwork}},
title = {Robot or Human? cost model and methodology},
year = {2026},
version = {1.0.0},
howpublished = {\url{https://whirrwork.com/tools/robot-vs-human/methodology/}},
note = {Accessed 2026-10-25. Modelled estimates, not facts.}
}All sources
- Occupational Employment and Wage Statistics, May 2025 (median hourly wage by occupation) U.S. Bureau of Labor Statistics, 2025-05
- OEWS May 2025 state estimates (median hourly wage by occupation and state) U.S. Bureau of Labor Statistics, 2025-05
- Employer Costs for Employee Compensation, June 2026: wages and salaries 70.0% of private-industry compensation (series CMU2020000000000P) U.S. Bureau of Labor Statistics, 2026-06
- Mean nominal hourly earnings of employees by sex and occupation (ISCO-08), DF_EAR_EHRA_SEX_OCU_CUR_NB ILOSTAT (International Labour Organization), from national labour force and earnings surveys, 2025
- Average monthly earnings of employees by sex and economic activity, DF_EAR_EMTA_SEX_ECO_CUR_NB ILOSTAT (International Labour Organization), 2022
- 令和6年賃金構造基本統計調査 (Basic Survey on Wage Structure 2024), table 1 by occupation Ministry of Health, Labour and Welfare, Japan (via e-Stat), 2024-06
- Average annual wage of employed persons in urban units in 2024 (by position and by region) National Bureau of Statistics of China, 2025-05-20
- National economy 2025: average weekly working hours of enterprise employees 48.6 National Bureau of Statistics of China, 2026-01-19
- Labour cost levels by NACE Rev. 2 activity (lc_lci_lev): Germany 2022-2025, total labour cost vs wages and salaries per hour Eurostat, 2025
- Rates and thresholds for employers 2025 to 2026 (employer Class 1 NIC 15% above £5,000) HM Revenue & Customs, 2025-04-06
- Workplace pensions: what you, your employer and the government pay (employer minimum 3%) GOV.UK, 2025
- Consumer price index (2010 = 100), FP.CPI.TOTL World Bank, World Development Indicators, 2026-10-08
- Euro foreign exchange reference rates (cross rates to USD), via Frankfurter European Central Bank, 2026-10-23
- ISSA 612 cleaning times: 20-inch walk-behind automatic scrubber about 9,274 sq ft per hour; wet mop about 1,333-1,429 ISSA, as reproduced by Parish Supply, 2026-10
- Neo 2 disc cleaning head technical specifications: maximum theoretical productivity 33,583 ft²/h (26-inch) Avidbots, 2026-10 · vendor
- Gausium autonomous scrubbers: purchase $29,600-$95,880; RaaS $829-$2,669/month on a 36-month term RobotLAB (reseller), 2026-10 · vendor
- Avidbots Neo lease $2,250-$2,900/month, as quoted by an Avidbots representative MLive, 2024-01
- Locus Robotics offers warehouse robots for $2,000 per month as a service CIVL, 2026-10
- Big Joe Pallet Mover AMR brochure: $45,000 each incl. deployment, training, annual visit and 3-year software subscription Jtec Products (dealer), 2024-08-14 · vendor
- Knightscope robots: effective $6-$12 per hour on an annual 24/7 Machine-as-a-Service contract Knightscope, 2026-10 · vendor
- Mammotion LUBA 3 AWD robot mower: $2,399-$3,299 by coverage tier Mammotion US store, 2026-10 · vendor
- Robotic mower installation: up to 1 acre $499, over 1 acre $799 Zippy Lawnz (installer), 2026-10 · vendor
- Annual robotic mower care plans $149-$249 Smart Lawn NY (installer), 2026-10 · vendor
- Lawn mowing cost: $30-$85 per visit for a 1/8 to 1/4 acre yard, about $50 on average HomeGuide, 2026
- Scythe M.52 autonomous commercial mower lease from $1,800/month (Standard, incl. 40 autonomous acres) Scythe Robotics, 2026-10 · vendor
- Weekly professional pool service $100-$150 per month PoolRobots.com, 2026-10 · vendor
- Maytronics Dolphin residential robotic pool cleaners $849-$1,449 (mid-range models) Maytronics US store, 2026-10-25 · vendor
- Conventional lift/rope exterior cleaning bids $0.50-$3.50 per sq ft (for comparison with drone cleaning) Sky PSI (drone cleaning company), 2026-10 · vendor
- Drone facade/window cleaning, commercial buildings: $0.12-$0.25 per sq ft Vista Drone Cleaning, 2026-10 · vendor
- Drone window/facade cleaning, Southern California: $0.25-$0.50 per sq ft Peninsula Pressure Pro, 2026-10 · vendor
- Lucid Bots Sherpa cleaning drone subscription for operators: $2,950-$4,600 per month Lucid Bots, 2026-10 · vendor
- Roof inspection cost: traditional physical inspection $75-$200 HomeGuide, 2026
- Drone roof inspection cost: standard visual $150-$400; thermal $400-$600 HomeGuide, 2026-01
- Drone inspection cost vs traditional methods: tower climb $800-$1,500 vs drone $350-$600; roof ladder $300-$800 vs drone $150-$400; solar manual $0.008-$0.012/W vs drone $0.003-$0.006/W AeroDeploy (drone inspection company), 2026 · vendor
- Cell tower drone inspection, average US rate $300-$600 per tower Scanixx, 2026-10 · vendor
- Wind turbine inspection costs: rope access $1,500-$3,000 per turbine vs drone $300-$600 SkyVisor (drone inspection software company), 2026-10 · vendor
- Wind turbine drone inspection in France: €350-€700 per turbine within a farm campaign C-Drone, 2026-10 · vendor
- Drone thermography of a ground-mounted solar farm (France): €150-€400 per MWp C-Drone, 2026-10 · vendor
- 2025 Iowa Farm Custom Rate Survey: self-propelled ground broadcast spraying $8.70/acre average; aerial application $11.85/acre average (materials excluded) Iowa State University Extension and Outreach, 2025
- 2025 ASDC Impact Survey: US average custom drone spray rate $13/acre (down from $21 in 2024) American Spray Drone Coalition, 2026-01
- DJI Agras T50 from $19,999; spray-ready package $31,500 Southeast Drone Solutions (dealer), 2026-10 · vendor
- Drone fumigation in Mexico: MXN 300-800 per hectare, application only EAVision México, 2026-10 · vendor
- Outsourced drone spraying in Brazil: BRL 60-150 per hectare Planon Ambiental, 2026-10 · vendor
- Average drone show cost by region (SPH Engineering market data): North America $52,500 (~218 drones); EU $42,034; UK $90,600; China $114,560; India $91,700; LatAm $25,550 Cyberdrone, citing SPH Engineering, 2026-10 · vendor
- City council agenda: Rails to Trails fireworks display, $9,468 (Flashing Thunder) City of Watertown, Minnesota, 2025-07-08
- Board of Trustees: Zambelli Fireworks contract, $42,500 base for a 22-minute July 3, 2026 show Town of Berthoud, Colorado, 2026-02-10
- Pudu BellaBot: buy $14,500; RaaS lease $409/month RobotLAB (reseller), 2026-10 · vendor
- UK serving robot rental £490-£590/month on a 24-month term incl. maintenance Hutech Robotics, 2026-10 · vendor
- Tug robots leased for $1,500-$2,000 a month; University of Maryland Medical Center per-trip cost $2.40 by robot vs $5.50 by hand (WSJ, Feb 2014, reproduced by Aethon) The Wall Street Journal via Aethon, 2014-02-23
- Hospitals have purchased TUGs for $75,000-$140,000; leases $1,500-$2,000/month (Becker's Hospital Review, reproduced by Aethon) Becker's Hospital Review via Aethon, 2015-03-31
- Universal Robots UR5e cobot arm $39,083-$43,544 (US distributor listing) Automation Distribution, 2026-10 · vendor
- RaaS machine tending: six-axis robot at about $10 per hour (Formic) Plastics Machinery & Manufacturing, 2026-10
- Cobot Welder starting price $105,000 (complete system incl. UR8 Long, power source, torch, cart, software); "most shops see 2-4x the output" Hirebotics, 2026-10 · vendor
- Cobot welding price breakdown: integration/commissioning $15,000-$40,000; base system 50-60% of total investment; maintenance 3-5% per year THG Automation (integrator), 2026 · vendor
- Turnkey FANUC CRX cobot palletizing cell, from $94,750 RobotPalletizer.com, 2026-10 · vendor
- Complete cobot palletizer cell $119,850 PicPac USA, 2026-10 · vendor
- Langham Logistics pays about $4,000 per month per Gather AI inventory drone Indianapolis Business Journal via Progressive Logistics, 2026-10
- Corvus One inventory drone, small distribution centre (<150,000 sq ft): about $2,500-$4,500 per month RaaS Actel Robotics, 2026 · vendor
- Snow removal: driveway $30-$70 per visit (standard driveway up to 90 ft) HomeGuide, 2026
- Yarbo Y40P Core with Snow Blower Module Gen 2: $7,299 Yarbo US store, 2026-10 · vendor
- World Robotics 2026: 603,307 industrial robots installed in 2025 (+11%); forecast 655,000 in 2026 and 806,000 in 2029 International Federation of Robotics, 2026-09
- Global sales of professional service robots surge 24% (2025): ~250,000 units; transport & logistics 117,500 (+21%); professional cleaning +40% International Federation of Robotics, 2026-09-30
- Interact Analysis: mobile robot shipments grew 70% in 2021; each time shipments double, prices expected to fall about 10-15% Robotics 24/7, 2022-12
- Wright, T. P. (1936). Factors affecting the cost of airplanes. Journal of the Aeronautical Sciences, 3(4), 122-128 Journal of the Aeronautical Sciences, 1936
- Nagy, B., Farmer, J. D., Bui, Q. M., & Trancik, J. E. (2013). Statistical basis for predicting technological progress. PLoS ONE, 8(2), e52669 (62 technologies; Wright's law among the best hindcasters; errors grow with horizon) PLoS ONE, 2013
- Ziegler, M. S., & Trancik, J. E. (2021). Re-examining rates of lithium-ion battery technology improvement and cost decline. Energy & Environmental Science, 14, 1635-1651 (about 20% cost decline per doubling) Energy & Environmental Science, 2021
- Battery pack prices cited below $100/kWh for the first time in 2020, while market average sits at $137/kWh (89% real decline from >$1,100/kWh in 2010) BloombergNEF, 2020-12-16
- Lithium-ion battery pack prices see largest drop since 2017, falling to $115 per kilowatt-hour BloombergNEF, 2024-12-10
- Lithium-ion battery pack prices fall to $108 per kilowatt-hour (2025, -8%) BloombergNEF, 2025-12-09
- Velodyne introduces 3D real-time LiDAR Puck (VLP-16) at $7,999 Velodyne via PRWeb, 2014-09-03 · vendor
- Velodyne slashes the price of its most popular LiDAR sensor in half (VLP-16 to $3,999) RoboticsTomorrow, 2018-01-02
- China's Hesai to halve lidar prices next year (ATX under $200 in 2025) Reuters, 2024-11-27
- Unitree Robotics unveils G1 humanoid for $16K (H1 was $90,000) The Robot Report, 2024-05-13
- Unitree G1 humanoid official store price $13,500-$15,999 Unitree, 2026-10 · vendor
- DJI announces $15,000 agricultural drone designed to spray crops (Agras MG-1, 10 kg tank) The Verge, 2015-11-26
- iRobot introduces Roomba Intelligent FloorVac, $199.95 iRobot, 2002-09-18 · vendor
- Roomba robot vacuums with self-emptying or auto-wash dock, $449.99-$699.99 iRobot US store, 2026-10-25 · vendor
- Universal Robots wants to conquer the universe (of robotic arms): UR5 costs $35,000 IEEE Spectrum, 2015
- iRobot announces strategic transaction (prepackaged Chapter 11; Picea to acquire) iRobot, 2025-12-14 · vendor
- Walmart drops plans to use robots for tracking inventory Reuters, 2020-11-02
- Japan's robot hotel lays off half the robots after they created more work for humans The Verge, 2019-01-15